University of Southern California Dissertations and Theses

Substance use and technology: testing an innovative method for recruitment of young men who have sex with men

(USC Thesis Other)

Substance use and technology: testing an innovative method for recruitment of young men who have sex with men

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Substance Use and Technology: Testing an Innovative Method for Recruitment of Young Men
Who Have Sex with Men

by
Jeremy J. Gibbs, MSW

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(SOCIAL WORK)

August 2018

SUBSTANCE USE AND TECHNOLOGY ii
Table of Contents
List of Figures and Tables .......................................................................................................... iv
Dedication................................................................................................................................... v
Acknowledgments......................................................................................................................vi
Abstract ...................................................................................................................................... 1
Introduction ................................................................................................................................ 3
References............................................................................................................................. 7
Chapter 1: Background.............................................................................................................. 15
Significance of Problem ...................................................................................................... 15
Sampling Methodology Used with This Population ............................................................. 16
Expected Contributions to the Field ..................................................................................... 18
References........................................................................................................................... 21
Chapter 2: Relevant Theory....................................................................................................... 30
Minority Stress Theory ........................................................................................................ 30
Social Information Processing Perspective .......................................................................... 31
References........................................................................................................................... 35
Chapter 3: Methods ................................................................................................................... 40
Procedures........................................................................................................................... 41
Stage 1: Building a Sampling Frame .............................................................................. 41
Stage 2: Enumeration of the Sampling Frame ................................................................ 42
Stage 3: Recruitment and Data Collection ...................................................................... 47
Measurement ....................................................................................................................... 51
Analytic Procedure .............................................................................................................. 53
Aim 1 ............................................................................................................................ 53
Aim 2 ............................................................................................................................ 54
Power Analysis ................................................................................................................... 54
References........................................................................................................................... 56
Chapter 4 (Paper 1): Using a Geosocial Networking Application for Random Sampling of
Young Men Who Have Sex with Men: Development of an Innovative Method ......................... 68
Introduction ......................................................................................................................... 68
Venue-Based Convenience Sample Recruitment ............................................................ 69
Internet-Based Recruitment ........................................................................................... 70
Respondent-Driven Sampling ........................................................................................ 71
Venue-Based Probability Sample Recruitment ............................................................... 73
Introduction of Geosocial Networking Smartphone Apps .............................................. 74
Methods .............................................................................................................................. 76
Stage 1: Building a Sampling Frame .............................................................................. 77
Stage 2: Enumeration of the Sampling Frame ................................................................ 78
Stage 3: Recruitment and Data Collection ...................................................................... 83
Barriers and Strengths of GSNA-Based Probability Sampling ............................................. 86
Navigation of the GSNA Partnership ............................................................................. 86
Determination of the Geographic Sampling Frame ........................................................ 88
Conclusions ......................................................................................................................... 90
References........................................................................................................................... 94
SUBSTANCE USE AND TECHNOLOGY iii
Chapter 5 (Paper 2): Young Men Who Have Sex with Men and Substance Use: A Comparison
of Venue-Based Sampling and Geosocial Networking Application Sampling .......................... 118
Introduction ....................................................................................................................... 118
Geosocial Networking Applications............................................................................. 120
Methods ............................................................................................................................ 122
Building the Sampling Frame and Enumeration ........................................................... 122
Recruitment and Data Collection ................................................................................. 125
Measurement ............................................................................................................... 128
Analytic Procedure ...................................................................................................... 130
Results .............................................................................................................................. 130
Discussion ......................................................................................................................... 134
References......................................................................................................................... 140
Chapter 6 (Paper 3): Venue-Based versus Geosocial Networking Application-Based
Recruitment of Young Men Who Have Sex with Men: An Examination of Feasibility ............ 160
Introduction ....................................................................................................................... 160
Methods ............................................................................................................................ 163
Recruitment Preparation .............................................................................................. 163
Recruitment Implementation........................................................................................ 166
Measurement ............................................................................................................... 169
Analytic Procedure ...................................................................................................... 171
Results .............................................................................................................................. 171
Discussion ......................................................................................................................... 175
References......................................................................................................................... 179
Chapter 7: Conclusions ........................................................................................................... 191
References......................................................................................................................... 199

SUBSTANCE USE AND TECHNOLOGY iv
List of Figures and Tables
Figure 3.1. Geographic Sampling Frame A ............................................................................... 61
Figure 3.2. Geographic Sampling Frame B (GSF A Present) ..................................................... 62
Figure 3.3. Geographic Sampling Frame C (GSF B Present) ..................................................... 63
Figure 3.5. Interpolated User Density on Mondays at 10 a.m. .................................................... 65
Figure 3.6. Day-Time-Sampling-Locations for Mondays at 10 a.m. in GSF B ........................... 66
Figure 3.7. Day-Time-Sampling-Locations for Mondays at 10 a.m. in GSF C ........................... 67
Figure 4.1. Geographic Sampling Frame A ............................................................................. 108
Figure 4.2. Geographic Sampling Frame B (GSF A Present) ................................................... 109
Figure 4.3. Geographic Sampling Frame C (GSF B Present) ................................................... 110
Figure 4.4. Geographic Sampling Frame B with Grid of 79 Sampling Points ........................... 111
Figure 4.5. Interpolated User Density on Mondays at 10 a.m. .................................................. 112
Figure 4.6. Summation of Neighborhoods ............................................................................... 113
Figure 4.7. Math Algebra Process of Removing Summation of Neighborhood Overlap ........... 114
Figure 4.8. Day-Time-Sampling-Locations for Mondays at 10 a.m. in GSF B ......................... 115
Figure 4.9. Steps for Analyzing User Density Data in ArcMap 10.5 ........................................ 116
Figure 4.10. Day-Time-Sampling-Locations for Mondays at 10 a.m. in GSF C ....................... 117
Table 5.1. Demographic Differences between Samples ........................................................... 152
Table 5.2. Substance Use Covariate Differences between Samples .......................................... 153
Table 5.3. Lifetime Substance Use Differences between Samples............................................ 154
Table 5.4. Differences in Substance Use during the Prior Month between Samples ................. 155
Table 5.5. Correlations between Study Variables..................................................................... 156
Table 5.6. Relationship between Covariates and Substance Use during the Prior Month
Moderated by Recruitment Type ............................................................................................. 157
Figure 5.1. Moderation of GCC and Number of Days Binge Drinking by Recruitment Type ... 158
Figure 5.2. Moderation of GCC and Number of Days Drinking Alcohol by Recruitment Type 159
Table 6.1. Recruitment Period Statistics .................................................................................. 186
Table 6.2. Summary of Recruitment Method Work Hours ....................................................... 187
Table 6.3. Summary of Recruitment Method Costs ................................................................. 188
Figure 6.1. Recruitment Flow Chart for Venue-Based and App-Based Recruitment ................ 189
Figure 6.2. Cost per Participant by Recruitment Method based on Study Sample Size ............. 190
Table 7.1. Population Estimates of App and Gay Venue Use among YMSM ........................... 201

SUBSTANCE USE AND TECHNOLOGY v
Dedication
This dissertation is dedicated to two groups of people: the sexual minority people who
came before me and improved the lives of those to come after them, and to my family, who
nurtured the desire in me to pursue a career committed to social justice and science. This
dissertation is only possible because of the sexual minority individuals, who when it was difficult
to acknowledge who they were, spoke up and blazed a trail for the next generation. To the gay
men, who befriended my grandparents while my father was still a child, this is for you, you made
a difference.
This work is also for my family, of whom I truly believe I am the least extraordinary. To
my grandparents, Dr. James J. Gibbs and Betty J. Gibbs: your lives inspired me to pursue a life
committed to social justice, mental health, and science. And this study is also dedicated to my
parents, James D. Gibbs and Carol M. Gibbs. You taught me that a sacrifice made in love is
more life giving than any accomplishment.

SUBSTANCE USE AND TECHNOLOGY vi
Acknowledgments
These last seven years have been a race that I would not have finished without those who
supported me and my work. This is an attempt to acknowledge the impact they had:
First and foremost, to Dr. Jeremy Goldbach for mentoring me and making me dig deep
within myself to deliver a work I did not know I could accomplish. Your counsel, support,
kindness, and challenge have shaped me into the researcher I am today.
To Dr. Eric Rice, for unofficially mentoring me and being so generous with your data and
ideas. My work is built on your shoulders.
To Dr. Robin Petering, for being my confidant and my inspiration. Sharing this journey
with you has been my pleasure.
To Drs. Dorian Traube, Alice Echols, Karen Kemp, Sheree Schrager, Michàlle Mor
Barak and other faculty at USC, for critiquing and nurturing this work.
To my dissertation study team: John “Jack” Senese, Justin Zhang, Garrett Weskamp,
Bryan Hancock, and Spencer Vincente, for doing to heavy lifting for this work.
To my family and friends in Los Angeles, Baltimore, São Paulo, Amsterdam,
Philadelphia, Half Moon Bay, and everywhere else in the world, for all your support and care.
To my loving husband, Leandro T. Duarte, for sharing with me your deep empathy for
those mistreated by the world. I am a more compassionate and effective social work researcher
because of you.
To Hornet, for taking a chance on me, and for their dedication to supporting sexual
minority men’s health.
To the National Institutes on Drug Abuse, for funding this dissertation and other LGBTQ
research.
SUBSTANCE USE AND TECHNOLOGY vii
And to all the sexual minority participants in this study, for their contribution to science
and improving the lives of those to come after them.
SUBSTANCE USE AND TECHNOLOGY 1
Abstract
Young men who have sex with men (YMSM; aged 18 to 24) report significantly higher
rates of substance use compared to their heterosexual peers. Studies examining the psychosocial
determinants of substance use among YMSM have historically relied on samples of men
recruited from venues that arguably condone substance use (e.g., gay bars). This may artificially
inflate substance use prevalence estimates in this community and limit the generalizability of
these studies to the full population of YMSM. Further, recent research has indicated that YMSM
are increasingly using smartphone technology (i.e., geosocial networking applications, or
GSNAs) to find sexual partners and forgoing the use of venues. To date, no study has
investigated whether a sample of YMSM drawn from a GSNA differs from a sample recruited
through venues. Therefore, the overarching goal of this dissertation was to recruit two samples of
YMSM through venue-based probability sampling and GSNA-based probability sampling to (a)
compare substance use and the psychosocial covariates of substance use of the two samples, and
(b) investigate the feasibility of implementing each recruitment method (i.e., cost and
recruitment efficiency).
A cross-sectional online survey design was utilized with a sample of 111 YMSM
recruited in Los Angeles, CA. Sixty-eight participants were recruited through venue-based
probability sampling procedures and 43 were recruited using GSNA-based probability sampling.
The GSNA method systematically identified locations for recruitment using spatial analysis of
app-user densities in the geographic sampling frame. Sample substance use (e.g., alcohol, binge
drinking, cocaine, poppers), substance use covariates (e.g., discrimination, depression, gay
community connection), sexual risk (e.g., recent unprotected sex, HIV testing) and demographics
were compared using chi-square tests and two-sample t-tests. The relationships between
SUBSTANCE USE AND TECHNOLOGY 2
covariates and substance use were also tested for moderation by recruitment type using linear
and logistic regressions. All costs and hours of work were recorded throughout the recruitment
preparation and implementation process.
The samples significantly differed in demographics characteristics (i.e., race and
ethnicity, revealing sexual orientation to all parents, employment, education). No significant
differences in psychosocial covariates of substance use or sexual risk emerged. Regressions
indicated significant differences (higher in the venue sample) in lifetime and recent substance
use between the two samples of men. The relationship between gay community connection and
recent alcohol consumption was moderated by recruitment type. Venue-based recruitment was
more than 2 times more expensive and required more than 3 times more work hours to complete.
Results indicate that recruitment method affects substance use prevalence levels for
YMSM. Although differences in prevalence emerged, findings that psychosocial covariates of
substance use did not differ between samples offer support for population generalizability. This
dissertation also found evidence regarding the feasibility of using technology to recruit
probability samples of YMSM. Venue-based procedures cost more than 2 times as much as
GSNA procedures, a notable difference. GSNA-based methods may be especially applicable to
areas in which venues are not readily accessible (e.g., rural areas and international contexts in
which being a sexual minority is stigmatized). Future research should consider the application of
these methods outside of an urban area to investigate feasibility.

SUBSTANCE USE AND TECHNOLOGY 3
Introduction
Young men who have sex with men (YMSM; 18 to 24 years old) report higher rates of
alcohol and substance use (e.g., marijuana, depressants, stimulants, hallucinogens, and inhalants)

compared to their heterosexual peers (Boyd, McCabe, & d’Arcy, 2003; Gattis, Sacco, &
Cunningham-Williams, 2012; Goldberg, Strutz, Herring, & Halpern, 2013; Hagger-Johnson et
al., 2013; Hatzenbuehler, Corbin, & Fromme, 2008; McCabe, Boyd, Hughes, & d’Arcy, 2003;
McCabe, Hughes, Bostwick, West, & Boyd, 2009). Additionally, YMSM substance use is
associated with increased frequency of both unprotected insertive and receptive anal sex,
increasing their risk of HIV and other sexually transmitted infections (Operario et al., 2006;
Parsons, Lelutiu-Weinberger, Botsko, & Golub, 2013). Further, using substances (e.g.,
methamphetamine, Viagra, and poppers) during sexual activity increases the risk of unprotected
anal intercourse (Carey et al., 2009). Given the various negative health implications of substance
use, including addiction, cardiovascular disease, stroke, cancer, and liver disease (National
Institute on Drug Abuse, 2012), fully understanding the factors that promote substance use by
YMSM is paramount to reducing this public health concern.
Much of the current literature examining the psychosocial determinants of substance use
has relied on samples of YMSM recruited from venues (e.g., gay bars, bath houses, local gay
events), which by their nature are more likely to condone use of alcohol or drugs (Kelly, Davis,
& Schlesinger, 2015; Kipke, Weiss, Ramirez, et al., 2007; Kipke, Weiss, & Wong, 2007;
Mutchler et al., 2011; Operario et al., 2006; Thiede et al., 2003; Traube, Schrager, Holloway,
Weiss, & Kipke, 2013). These samples may lead to both inflated substance use estimates and
limited generalizability of significant psychosocial determinants (e.g., discrimination,
internalized stigma, sensation seeking, depression), primarily due to the venue-based nature of
SUBSTANCE USE AND TECHNOLOGY 4
recruitment. Venue-based studies systematically exclude YMSM who are not connected to the
gay venue community and individuals who may not identify as a sexual minority (Meyer &
Wilson, 2009). Other sampling methods such as respondent-driven sampling have been reported
as a viable option for YMSM recruitment, but recent research has identified issues with
homogeneity (i.e., low-income, unemployed, racial and ethnic minority samples; Shoptaw et al.,
2009) and limited feasibility due to the time needed to recruit a full sample (Kuhns et al., 2015).
Emerging literature has suggested that YMSM often use technology to meet other
YMSM and rely less on venues to provide social interaction (Grov, 2012; Grov & Crow, 2012;
Zablotska, Holt, & Prestage, 2012). Geosocial networking applications (GSNAs) like Hornet,
which has reported more than 25 million users worldwide (Hornet, 2018), are estimated to be
used by two thirds of men who have sex with men (Phillips et al., 2014). Although GSNAs
require the use of a smart device, an estimated 94% of 18- to 29-year-olds in the United States
have access to this technology (Pew Research Center, 2018).
GSNAs provide a profile-based home screen environment in which YMSM can browse
the profiles of other YMSM in the same geographic region. Because profiles are ordered based
on their relative proximity to the user, YMSM users may click on a profile and see the distance
to that user from their current location. Whereas YMSM previously had limited access to
socialization, except through gay venues, they can now open a GSNA and easily find other
YMSM. Portions of the YMSM community that previously attended gay venues specifically to
meet other men (and not to use alcohol or other substances) may no longer use those venues to
facilitate connections. Therefore, GSNAs are the next logical step in advancing YMSM
recruitment methods. Recruitment of YMSM using GSNA is relatively simple. These procedures
have been developed and used in previous studies (Buckingham et al., 2017; Duncan et al., 2018;
SUBSTANCE USE AND TECHNOLOGY 5
Gibbs & Rice, 2016; Goedel & Duncan, 2015, 2016; Goedel, Hagan, et al., 2017; Goedel,
Halkitis, & Duncan, 2016; Goedel, Safren, Mayer, & Duncan, 2017; Holloway, Pulsipher, Gibbs,
Barman-Adhikari, & Rice, 2015; Holloway et al., 2014; Holloway et al., 2017; Lachowsky et al.,
2016; Lorimer, Flowers, Davis, & Frankis, 2016; Macapagal, Coventry, Puckett, Phillips, &
Mustanski, 2016; Rendina, Jimenez, Grov, Ventuneac, & Parsons, 2014; Rice et al., 2012;
Siegler et al., 2015).
Despite the potential to revolutionize data collection with YMSM, the use of this
technology in research has been criticized for being biased toward YMSM who are single,
younger, and more sexually risky (Beymer et al., 2014; Burrell et al., 2012; Choi, Wong, &
Fong, 2017; Eaton et al., 2016; Lehmiller & Ioerger, 2014; Phillips et al., 2014). These
criticisms, however, are based on a limited number of studies that either did not use random
sampling procedures (Burrell et al., 2012) or compared samples of venue-recruited YMSM based
on their GSNA use rather than sampling using a GSNA (Eaton et al., 2016; Lehmiller & Ioerger,
2014; Phillips et al., 2014). To date, no study has compared the substance use characteristics and
associated psychosocial determinants of a probability sample recruited through a GSNA to a
probability sample of venue-recruited YMSM.
To address this gap in the literature, the overarching goal of this dissertation was to
compare the substance use characteristics of two samples of YMSM individuals: (a) a probability
sample recruited using a GSNA; and (b) a probability sample recruited using venue-based
procedures. To achieve this goal, the specific aims were as follows:
Aim 1 (Chapter 4): Develop and implement a GSNA-based probability sampling
methodology.
SUBSTANCE USE AND TECHNOLOGY 6
Aim 2 (Chapter 5): Investigate whether differences exist in substance use, psychosocial
determinants of substance use, and demographics between venue-recruited and GSNA-recruited
probability samples of YMSM.
Aim 3 (Chapter 6): Compare the feasibility (i.e., cost and recruitment efficiency) of
GSNA recruitment methods with venue-based recruitment methods for YMSM.

SUBSTANCE USE AND TECHNOLOGY 7
References
Beymer, M. R., Weiss, R. E., Bolan, R. K., Rudy, E. T., Bourque, L. B., Rodriguez, J. P., &
Morisky, D. E. (2014). Sex on demand: Geosocial networking phone apps and risk of
sexually transmitted infections among a cross-sectional sample of men who have sex with
men in Los Angeles County. Sexually Transmitted Infections, 90, 567–572.
doi:10.1136/sextrans-2013-051494
Boyd, C. J., McCabe, S. E., & d’Arcy, H. (2003). Ecstasy use among college undergraduates:
Gender, race and sexual identity. Journal of Substance Abuse Treatment, 24, 209–215.
doi:10.1016/S0740-5472(03)00025-4
Buckingham, L., Becher, J., Voytek, C. D., Fiore, D., Dunbar, D., Davis-Vogel, A., … Frank, I.
(2017). Going social: Success in online recruitment of men who have sex with men for
prevention HIV vaccine research. Vaccine, 35, 3498–3505.
doi:10.1016/j.vaccine.2017.05.002
Burrell, E. R., Pines, H. A., Robbie, E., Coleman, L., Murphy, R. D., Hess, K. L., … Gorbach, P.
M. (2012). Use of the location-based social networking application Grindr as a
recruitment tool in rectal microbicide development research. AIDS and Behavior, 16,
1816–1820. doi:10.1007/s10461-012-0277-z
Carey, J. W., Mejia, R., Bingham, T., Ciesielski, C., Gelaude, D., Herbst, J. H., … Stall, R.
(2009). Drug use, high-risk sex behaviors, and increased risk for recent HIV infection
among men who have sex with men in Chicago and Los Angeles. AIDS and Behavior,
13, 1084–1096. doi:10.1007/s10461-008-9403-3
Choi, E. P. H., Wong, J. Y. H., & Fong, D. Y. T. (2017). The use of social networking
applications of smartphone and associated sexual risks in lesbian, gay, bisexual, and
SUBSTANCE USE AND TECHNOLOGY 8
transgender populations: A systematic review. AIDS care, 29, 145–155.
doi.org/10.1080/09540121.2016.1211606
Duncan, D. T., Goedel, W. C., Stults, C. B., Brady, W. J., Brooks, F. A., Blakely, J. S., &
Hagen, D. (2018). A study of intimate partner violence, substance abuse, and sexual risk
behaviors among gay, bisexual, and other men who have sex with men in a sample of
geosocial-networking smartphone application users. American Journal of Men’s Health,
12, 292–301. doi:10.1177/1557988316631964
Eaton, L. A., Maksut, J. L., Gamarel, K. E., Siembida, E. J., Driffin, D. D., & Baldwin, R.
(2016). Online sex partner meeting venues as a risk factor for testing HIV positive among
a community-based sample of black men who have sex with men. Sexually transmitted
diseases, 43, 360–364. doi: 10.1097/OLQ.0000000000000454
Gattis, M. N., Sacco, P., & Cunningham-Williams, R. M. (2012). Substance use and mental
health disorders among heterosexual identified men and women who have same-sex
partners or same-sex attraction: Results from the National Epidemiological Survey on
Alcohol and Related Conditions. Archives of Sexual Behavior, 41, 1185–1197.
doi:10.1007/s10508-012-9910-1
Gibbs, J. J., & Rice, E. (2016). The social context of depression symptomology in sexual
minority male youth: Determinants of depression in a sample of Grindr users. Journal of
Homosexuality, 63, 278–299. doi:10.1080/00918369.2015.1083773
Goedel, W. C., & Duncan, D. T. (2015). Geosocial-networking app usage patterns of gay,
bisexual, and other men who have sex with men: Survey among users of Grindr, a mobile
dating app. JMIR Public Health and Surveillance, 1, e4. doi:10.2196/publichealth.4353
SUBSTANCE USE AND TECHNOLOGY 9
Goedel, W. C., & Duncan, D. T. (2016). Contextual factors in geosocial-networking smartphone
application use and engagement in condomless anal intercourse among gay, bisexual, and
other men who have sex with men who use Grindr. Sexual Health, 13, 549–554.
doi:10.1071/SH16008
Goedel, W. C., Hagen, D., Halkitis, P. N., Greene, R. E., Griffin-Tomas, M., Brooks, F. A., …
Duncan, D. T. (2017). Post-exposure prophylaxis awareness and use among men who
have sex with men in London who use geosocial-networking smartphone applications.
AIDS Care, 29, 579–586. doi:10.1080/09540121.2016.1259455
Goedel, W. C., Halkitis, P. N., & Duncan, D. T. (2016). Behavior- and partner-based HIV risk
perception and sexual risk behaviors in men who have sex with men (MSM) who use
geosocial-networking smartphone applications in New York City. Journal of Urban
Health, 93, 400–406. doi:10.1007/s11524-016-0043-z
Goedel, W. C., Safren, S. A., Mayer, K. H., & Duncan, D. T. (2017). Community-level norms
and condomless anal intercourse among gay, bisexual, and other men who have sex with
men who use geosocial-networking smartphone applications in the Deep South. Journal
of HIVAIDS & Social Services, 16, 382–389. doi:10.1080/15381501.2017.1344599
Goldberg, S., Strutz, K. L., Herring, A. A., & Halpern, C. T. (2013). Risk of substance abuse and
dependence among young adult sexual minority groups using a multidimensional
measure of sexual orientation. Public Health Reports, 128, 144–152.
doi:10.1177/003335491312800304
Grov, C. (2012). HIV risk and substance use in men who have sex with men surveyed in
bathhouses, bars/clubs, and on Craigslist.org: Venue of recruitment matters. AIDS and
Behavior, 16, 807–817. doi:10.1007/s10461-011-9999-6
SUBSTANCE USE AND TECHNOLOGY 10
Grov, C., & Crow, T. (2012). Attitudes about and HIV risk related to the “most common place”
MSM meet their sex partners: Comparing men from bathhouses, bars/clubs, and
Craigslist.org. AIDS Education and Prevention, 24, 102–116.
doi:10.1521/aeap.2012.24.2.102
Hagger-Johnson, G., Taibjee, R., Semlyen, J., Fitchie, I., Fish, J., Meads, C., & Varney, J.
(2013). Sexual orientation identity in relation to smoking history and alcohol use at age
18/19: Cross-sectional associations from the Longitudinal Study of Young People in
England (LSYPE). BMJ Open, 3, e002810. doi:10.1136/bmjopen-2013-002810
Hatzenbuehler, M. L., Corbin, W. R., & Fromme, K. (2008). Trajectories and determinants of
alcohol use among LGB young adults and their heterosexual peers: Results from a
prospective study. Developmental Psychology, 44, 81–90. doi:10.1037/0012-
1649.44.1.81
Holloway, I. W., Pulsipher, C. A., Gibbs, J., Barman-Adhikari, A., & Rice, E. (2015). Network
influences on HIV risk behaviors of gay, bisexual and other men who have sex with men
using geosocial networking applications. AIDS and Behavior, 19(Suppl. 2), 112–122.
doi:10.1007/s10461-014-0989-3
Holloway, I. W., Rice, E., Gibbs, J., Winetrobe, H., Dunlap, S., & Rhoades, H. (2014).
Acceptability of smartphone application-based HIV prevention for among young men
who have sex with men. AIDS and Behavior, 18, 285–296. doi:10.1007/s10461-013-
0671-1
Holloway, I. W., Tan, D., Gildner, J. L., Beougher, S. C., Pulsipher, C., Montoya, J. A., …
Leibowitz, A. (2017). Facilitators and barriers to pre-exposure prophylaxis willingness
SUBSTANCE USE AND TECHNOLOGY 11
among young men who have sex with men who use geosocial networking applications in
California. AIDS Patient Care and STDs, 31, 517–527. doi:10.1089/apc.2017.0082
Hornet. (2018). About us. Retrieved from https://love.hornet.com/about/
Kelly, J., Davis, C., & Schlesinger, C. (2015). Substance use by same sex attracted young
people: Prevalence, perceptions and homophobia. Drug and Alcohol Review, 34, 358–
365. doi:10.1111/dar.12158
Kipke, M. D., Weiss, G., Ramirez, M., Dorey, F., Ritt-Olson, A., Iverson, E., & Ford, W. (2007).
Club drug use in Los Angeles among young men who have sex with men. Substance Use
& Misuse, 42, 1723–1743. doi:10.1080/10826080701212261
Kipke, M. D., Weiss, G., & Wong, C. F. (2007). Residential status as a risk factor for drug use
and HIV risk among young men who have sex with men. AIDS and Behavior, 11(Suppl.
2), 56–69. doi:10.1007/s10461-006-9204-5
Kuhns, L. M., Kwon, S., Ryan, D. T., Garofalo, R., Phillips, G., II, & Mustanski, B. S. (2015).
Evaluation of respondent-driven sampling in a study of urban young men who have sex
with men. Journal of Urban Health, 92, 151–167. doi:10.1007/s11524-014-9897-0
Lachowsky, N. J., Lal, A., Forrest, J. I., Card, K. G., Cui, Z., Sereda, P., … Hogg, R. S. (2016).
Including online-recruited seeds: A respondent-driven sample of men who have sex with
men. Journal of Medical Internet Research, 18, e51. doi:10.2196/jmir.5258
Lehmiller, J. J., & Ioerger, M. (2014). Social networking smartphone applications and sexual
health outcomes among men who have sex with men. PLOS One, 9, e86603.
doi:10.1371/journal.pone.0086603
Lorimer, K., Flowers, P., Davis, M., & Frankis, J. (2016). Young men who have sex with men’s
use of social and sexual media and sex-risk associations: Cross-sectional, online survey
SUBSTANCE USE AND TECHNOLOGY 12
across four countries. Sexually Transmitted Infections, 92, 371–376.
doi:10.1136/sextrans-2015-052209
Macapagal, K., Coventry, R., Puckett, J. A., Phillips, G., II, & Mustanski, B. (2016). Geosocial
networking app use among men who have sex with men in serious romantic relationships.
Archives of Sexual Behavior, 45, 1513–1524. doi:10.1007/s10508-016-0698-2
McCabe, S. E., Boyd, C., Hughes, T. L., & d’Arcy, H. (2003). Sexual identity and substance use
among undergraduate students. Substance Abuse, 24, 77–91.
doi:10.1023/A:1023768215020
McCabe, S. E., Hughes, T. L., Bostwick, W. B., West, B. T., & Boyd, C. J. (2009). Sexual
orientation, substance use behaviors and substance dependence in the United States.
Addiction, 104, 1333–1345. doi:10.1111/j.1360-0443.2009.02596.x
Meyer, I. H., & Wilson, P. A. (2009). Sampling lesbian, gay, and bisexual populations. Journal
of Counseling Psychology, 56, 23–31. doi:10.1037/a0014587
Mutchler, M. G., McKay, T., Candelario, N., Liu, H., Stackhouse, B., Bingham, T., & Ayala, G.
(2011). Sex drugs, peer connections, and HIV: Use and risk among African-American,
Latino, and multiracial young men who have sex with men in Los Angeles and New
York. Journal of Gay & Lesbian Social Services, 23, 271–295.
doi:10.1080/10538720.2011.560100
National Institute on Drug Abuse. (2012). Health consequences of drug abuse. Retrieved from
http://www.drugabuse.gov/related-topics/medical-consequences-drug-abuse
Operario, D., Choi, K.-H., Chu, P. L., McFarland, W., Secura, G. M., Behel, S., … Valleroy, L.
(2006). Prevalence and correlates of substance use among young Asian Pacific Islander
men who have sex with men. Prevention Science, 7, 19–29. doi:10.1007/s11121-005-
SUBSTANCE USE AND TECHNOLOGY 13
0018-x
Parsons, J. T., Lelutiu-Weinberger, C., Botsko, M., & Golub, S. A. (2013). Predictors of day-
level sexual risk for young gay and bisexual men. AIDS and Behavior, 17, 1465–1477.
doi:10.1007/s10461-012-0206-1
Pew Research Center. (2018). Mobile fact sheet: Who owns cellphones and smartphones.
Retrieved from http://www.pewinternet.org/fact-sheet/mobile/
Phillips, G., II, Magnus, M., Kuo, I., Rawls, A., Peterson, J., Jia, Y., … Greenberg, A. E. (2014).
Use of geosocial networking (GSN) mobile phone applications to find men for sex by
men who have sex with men (MSM) in Washington, DC. AIDS and Behavior, 18, 1630–
1637. doi:10.1007/s10461-014-0760-9
Rendina, H. J., Jimenez, R. H., Grov, C., Ventuneac, A., & Parsons, J. T. (2014). Patterns of
lifetime and recent HIV testing among men who have sex with men in New York City
who use Grindr. AIDS and Behavior, 18, 41–49. doi:10.1007/s10461-013-0573-2
Rice, E., Holloway, I., Winetrobe, H., Rhoades, H., Barman-Adhikari, A., Gibbs, J., … Dunlap,
S. (2012). Sex risk among young men who have sex with men who use Grindr, a
smartphone geosocial networking application. Journal of AIDS & Clinical Research, S4,
005. doi:10.4172/2155-6113.S4-005
Shoptaw, S., Weiss, R. E., Munjas, B., Hucks-Ortiz, C., Young, S. D., Larkins, S., … Gorbach,
P. M. (2009). Homonegativity, substance use, sexual risk behaviors, and HIV status in
poor and ethnic men who have sex with men in Los Angeles. Journal of Urban Health,
86(Suppl. 1), 77–92. doi:10.1007/s11524-009-9372-5
Siegler, A. J., Sanchez, T., Sineath, R. C., Grey, J., Kahle, E., & Sullivan, P. S. (2015).
Knowledge and awareness of acute human immunodeficiency virus infection among
SUBSTANCE USE AND TECHNOLOGY 14
mobile app-using men who have sex with men: A missed public health opportunity. Open
Forum Infectious Diseases, 2, ofv016. doi:10.1093/ofid/ofv016
Thiede, H., Valleroy, L. A., MacKellar, D. A., Celentano, D. D., Ford, W. L., Hagan, H., …
Torian, L. V. (2003). Regional patterns and correlates of substance use among young
men who have sex with men in 7 US urban areas. American Journal of Public Health, 93,
1915–1921. doi:10.2105/AJPH.93.11.1915
Traube, D. E., Schrager, S. M., Holloway, I. W., Weiss, G., & Kipke, M. D. (2013).
Environmental risk, social cognition, and drug use among young men who have sex with
men: Longitudinal effects of minority status on health processes and outcomes. Drug and
Alcohol Dependence, 127, 1–7. doi:10.1016/j.drugalcdep.2012.06.007
Zablotska, I. B., Holt, M., & Prestage, G. (2012). Changes in gay men’s participation in gay
community life: Implications for HIV surveillance and research. AIDS and Behavior, 16,
669–675. doi:10.1007/s10461-011-9919-9

SUBSTANCE USE AND TECHNOLOGY 15
Chapter 1: Background
Significance of Problem
In the last 15 years, research on behavioral health disparities between young men who
have sex with men (YMSM) and their heterosexual peers has increased. Overwhelmingly, this
research indicates that YMSM are at higher risk of both using substances and substance use
dependence. YMSM have higher odds of using marijuana (McCabe, Boyd, Hughes, & d’Arcy,
2003; McCabe, Hughes, Bostwick, West, & Boyd, 2009), ecstasy (Boyd, McCabe, & d’Arcy,
2003; McCabe et al., 2003), and other hard drugs (McCabe et al., 2009). In addition to higher
levels of alcohol consumption compared to their heterosexual peers (Hagger-Johnson et al.,
2013), YMSM drink to intoxication more often (Hatzenbuehler, Corbin, & Fromme, 2008) and
exhibit higher odds of developing alcohol dependence (Goldberg, Strutz, Herring, & Halpern,
2013)

than their peers. This is most notable during the transition from adolescence to young
adulthood; college-enrolled YMSM experience significantly higher increases in binge drinking
compared to their heterosexual peers (Hatzenbuehler et al., 2008). Dependence is a major risk for
this population. A 2012 national epidemiological survey found that YMSM have a significantly
higher prevalence of substance use disorder symptoms for alcohol, depressants, stimulants,
marijuana, hallucinogens, and inhalants (Gattis, Sacco, & Cunningham-Williams, 2012).
Substance use and dependence has been linked to numerous negative health-related
outcomes (e.g., cardiovascular disease, stroke, cancer, and liver disease; National Institute on
Drug Abuse, 2012). For YMSM, the risk associated with substance use is intensified due to a
well-established relationship between substance use and unprotected anal sex. Binge drinking
(i.e., more than five drinks in one sitting), marijuana, poppers, methamphetamine, gamma-
hydroxybutyric acid, hallucinogens, cocaine, amphetamine, crack cocaine, and heroin use are
SUBSTANCE USE AND TECHNOLOGY 16
each associated with higher odds of unprotected anal sex in this population (Colfax et al., 2004;
Parsons, Lelutiu-Weinberger, Botsko, & Golub, 2013). Further, polysubstance use is also
associated with higher odds of unprotected sex (Operario et al., 2006). A 2005 study found that
as the number of drugs being used increases (i.e., higher levels of polysubstance use), so do the
odds of engaging in unprotected anal sex (Colfax et al., 2005). In addition to general use, several
studies have examined sexual risk-taking associated with alcohol and substance use both before
and during sexual activity. Findings from these studies indicated a significant association of
binge drinking, poppers, methamphetamine, and Viagra use during sexual activity with
unprotected anal sex (Carey et al., 2009; Colfax et al., 2004). Estimates indicate that binge
drinking and popper use during sexual activity may increase the likelihood of unprotected sex by
3 and 2 times, respectively (Colfax et al., 2004).
Sampling Methodology Used with This Population
Studies examining the psychosocial determinants of substance use (e.g., discrimination,
victimization, disclosure stress, internalized stigma) for YMSM have relied on samples recruited
from gay venues that arguably condone substance use (e.g., gay bars, local gay events; Kelly,
Davis, & Schlesinger, 2015; Kipke, Weiss, Ramirez, et al., 2007; Kipke, Weiss, & Wong, 2007;
Mutchler et al., 2011; Operario et al., 2006; Thiede et al., 2003; Traube, Schrager, Holloway,
Weiss, & Kipke, 2013). Venue-based sampling is the most widely accepted recruitment
methodology used to sample YMSM. This type of recruitment involves the research team
identifying locations where YMSM congregate and then approaching perceived potential
participants, inquiring about eligibility, and inviting them to participate in the study (Meyer &
Wilson, 2009).
SUBSTANCE USE AND TECHNOLOGY 17
Although many studies using venue-based sampling with YMSM have relied on
convenience samples (Kelly et al., 2015; Mutchler et al., 2011), more rigorous procedures have
been developed that rely on preliminary enumeration (i.e., counting of potential participants) to
identify both times and locations of YMSM congregation (Kipke, Weiss, Ramirez, et al., 2007;
Kipke, Weiss, & Wong, 2007; MacKellar et al., 2007; Operario et al., 2006; Thiede et al., 2003;
Traube et al., 2013). Venue-based stratified probability sampling helps ensure that participants
are recruited based on probability rather than convenience (Ford et al., 2009). Despite this
enhancement, these recruitment strategies have been criticized for excluding YMSM who either
do not identify as gay or bisexual but engage in same-sex behaviors or have a limited connection
to the gay community (Meyer & Wilson, 2009). This sampling bias is a critical concern in
substance abuse research because it may artificially inflate the reporting of substance use
prevalence in this community. Further, it may cloud our understanding of risk factors for
substance use that would be uncovered in a more representative sample of YMSM. Because
substance use and sexual risk reduction interventions are developed using findings from these
potentially nongeneralizable studies, there is reason for concern (Kurtz, Stall, Buttram, Surratt, &
Chen, 2013).
Two other major alternatives for recruiting YMSM have been used in previous literature:
online website recruitment (i.e., chat room and banner ads) and respondent-driven sampling
(RDS). Internet website recruitment methods have limited viability because they typically rely
on nonprobability samples (Horvath, Bowen, & Williams, 2006; Horvath, Rosser, & Remafedi,
2008; Hospers, Kok, Harterink, & de Zwart, 2005; Fernández et al., 2004; Fernández et al.,
2007; Pequegnat et al., 2007; Sullivan et al., 2011). Additionally, although online website
recruitment tends to require less resources and time than venue-based recruitment, there are
SUBSTANCE USE AND TECHNOLOGY 18
significant disadvantages to recruiting a convenience sample, such as the lack of control related
to physical geolocation (i.e., a website user could be located anywhere in the world).
RDS has also been promoted as an option for YMSM recruitment, but research using this
method has reported that this procedure may not be as promising. Shoptaw et al. (2009), in their
study on men who have sex with men in Los Angeles, found that using RDS was not an
appropriate recruitment strategy because it led to homogenous recruitment chains of mostly men
of color and low-income or unemployed men. Homogeny of socioeconomic status has been
found in other YMSM studies using RDS (Kuhns et al., 2015; Reisner et al., 2010), in addition to
limited feasibility because sample recruitment required a time period far beyond expectations
(e.g., more than 3 years; Bryant, 2014; Kuhns et al., 2015; Lachowsky et al., 2016).
Given these limitations, the present study compared the most widely accepted form of
recruiting a probability sample of YMSM (i.e., venue-based stratified probability sampling) to a
novel method that used location-based geosocial networking applications (GSNAs). In 2009, the
first GSNA hit the smartphone application market (Grindr, 2015). These technologies use
geolocation data to allow YMSM users to contact and connect in real time with other users. The
format of GSNAs have also made it possible to filter by age and recruit a probability sample of
YMSM. Since this time, several studies have used GSNAs to recruit a probability sample of
YMSM (Gibbs & Rice, 2016; Holloway, Pulsipher, Gibbs, Barman-Adhikari, & Rice, 2015;
Holloway et al., 2014; Rice et al., 2012).
Expected Contributions to the Field
One of the primary strengths of this dissertation is its innovation: (a) it tested a novel
strategy for recruiting a probability sample of YMSM; (b) it strengthened the methods of this
novel recruitment strategy by using an enumeration stage for estimating GSNA user density; and
SUBSTANCE USE AND TECHNOLOGY 19
(c) the procedures for estimating GSNA user density and random sampling did not require the
study team to be present in the physical location, reducing cost and increasing efficiency.
This dissertation tested GSNA recruitment methods and compared them to venue-based
recruitment methods by examining characteristics of the sample, cost, and recruitment efficiency.
The methods built and tested in this dissertation are applicable to other areas of research that rely
on samples of men who have sex with men, including cross-sectional research and recruitment of
random samples of men for longitudinal and experimental research. Further, the geographic
information systems procedures used for building the geographic sampling frame are applicable
to fields that use social technology with global positioning system (GPS) data. These methods
allow researchers to understand where technology is being used most heavily, recruit samples
with more geographic variance, and identify geographic bias in samples.
In addition to testing a novel method of sample recruitment, this dissertation built on the
rigor of this novel methodology through incorporation of the enumeration procedures used in
venue-based sampling. Although the GSNA-recruited sample was selected using similar random
recruitment methods to those described by Rice et al. (2012), an enumeration phase was used
first to systematically identify locations for recruitment parallel to the formative procedures of
the Healthy Young Men’s Study (Ford et al., 2009). In previous studies, a single location for
recruitment was determined either based on convenience (Buckingham et al., 2017; Duncan et
al., 2018; Goedel & Duncan, 2015, 2016; Goedel, Hagan, et al., 2017; Goedel, Halkitis, &
Duncan, 2016; Goedel, Safren, Mayer, & Duncan, 2017; Holloway et al., 2017; Lachowsky et
al., 2016; Lorimer, Flowers, Davis, & Frankis, 2016; Macapagal, Coventry, Puckett, Phillips, &
Mustanski, 2016; Rendina, Jimenez, Grov, Ventuneac, & Parsons, 2014; Siegler et al., 2015) or
an expected high density of users (Gibbs & Rice, 2016; Holloway et al., 2015; Holloway et al.,
SUBSTANCE USE AND TECHNOLOGY 20
2014; Rice et al., 2012). Using geographic information systems technology, the user density of
the geographic sampling frame (i.e., the study geographic area) was calculated following the
procedures used by Delaney, Kramer, Waller, Flanders, and Sullivan (2014) in Atlanta, Georgia,
which can inform systematic identification of GPS locations for random recruitment. Use of
these GPS points ensured that all YMSM using the GSNA located in the study area had an equal
probability of being selected. Therefore, these sampling location identification procedures
created a known geographic sampling frame with maximized coverage of the study sampling
site. This added level of rigor lays the methodological framework for assessing the user density
of an entire city.
Because GSNAs rely on GPS information supplied by a smart device (i.e., phone or
tablet) to determine current location, GSNAs can be misled by changing the GPS data supplied.
Therefore, observing density data or random sampling of an area only requires changing the GPS
information provided (using a third-party application) to the GSNA, rather than being located at
that specific GPS point. Because studies using this method can do so from any location, the
limits of study team location are eliminated. This has vast implications for sampling YMSM in
rural areas in the United States and abroad.

SUBSTANCE USE AND TECHNOLOGY 21
References
Boyd, C. J., McCabe, S. E., & d’Arcy, H. (2003). Ecstasy use among college undergraduates:
Gender, race and sexual identity. Journal of Substance Abuse Treatment, 24, 209–215.
doi:10.1016/S0740-5472(03)00025-4
Bryant, J. (2014). Using respondent-driven sampling with ‘hard to reach’ marginalised young
people: Problems with slow recruitment and small network size. International Journal of
Social Research Methodology, 17, 599–611. doi:10.1080/13645579.2013.811921
Buckingham, L., Becher, J., Voytek, C. D., Fiore, D., Dunbar, D., Davis-Vogel, A., … Frank, I.
(2017). Going social: Success in online recruitment of men who have sex with men for
prevention HIV vaccine research. Vaccine, 35, 3498–3505.
doi:10.1016/j.vaccine.2017.05.002
Carey, J. W., Mejia, R., Bingham, T., Ciesielski, C., Gelaude, D., Herbst, J. H., … Stall, R.
(2009). Drug use, high-risk sex behaviors, and increased risk for recent HIV infection
among men who have sex with men in Chicago and Los Angeles. AIDS and Behavior,
13, 1084–1096. doi:10.1007/s10461-008-9403-3
Colfax, G., Coates, T. J., Husnik, M. J., Huang, Y., Buchbinder, S., Koblin, B., … EXPLORE
Study Team. (2005). Longitudinal patterns of methamphetamine, popper (amyl nitrite),
and cocaine use and high-risk sexual behavior among a cohort of San Francisco men who
have sex with men. Journal of Urban Health, 82, i62–i70. doi:10.1093/jurban/jti025
Colfax, G., Vittinghoff, E., Husnik, M. J., McKirnan, D., Buchbinder, S., Koblin, B., …
EXPLORE Study Team. (2004). Substance use and sexual risk: A participant- and
episode-level analysis among a cohort of men who have sex with men. American Journal
of Epidemiology, 159, 1002–1012. doi:10.1093/aje/kwh135
SUBSTANCE USE AND TECHNOLOGY 22
Delaney, K. P., Kramer, M. R., Waller, L. A., Flanders, W. D., & Sullivan, P. S. (2014). Using a
geolocation social networking application to calculate the population density of sex-
seeking gay men for research and prevention services. Journal of Medical Internet
Research, 16, e249. doi:10.2196/jmir.3523
Duncan, D. T., Goedel, W. C., Stults, C. B., Brady, W. J., Brooks, F. A., Blakely, J. S., & Hagen,
D. (2018). A study of intimate partner violence, substance abuse, and sexual risk
behaviors among gay, bisexual, and other men who have sex with men in a sample of
geosocial-networking smartphone application users. American Journal of Men’s Health,
12, 292–301. doi:10.1177/1557988316631964
Fernández, M. I., Varga, L. M., Perrino, T., Collazo, J. B., Subiaul, F., Rehbein, A., … Bowen,
G. S. (2004). The Internet as recruitment tool for HIV studies: Viable strategy for
reaching at-risk Hispanic MSM in Miami? AIDS Care, 16, 953–963.
doi:10.1080/09540120412331292480
Fernández, M. I., Warren, J. C., Varga, L. M., Prado, G., Hernandez, N., & Bowen, G. S. (2007).
Cruising in cyber space: Comparing Internet chat room versus community venues for
recruiting Hispanic men who have sex with men to participate in prevention studies.
Journal of Ethnicity in Substance Abuse, 6, 143–162. doi:10.1300/J233v06n02_09
Ford, W. L., Weiss, G., Kipke, M. D., Ritt-Olson, A., Iverson, E., & Lopez, D. (2009). The
Healthy Young Men’s Study: Sampling methods to recruit a random cohort of young
men who have sex with men. Journal of Gay & Lesbian Social Services, 21, 357–373.
doi:10.1080/10538720802498280
Gattis, M. N., Sacco, P., & Cunningham-Williams, R. M. (2012). Substance use and mental
health disorders among heterosexual identified men and women who have same-sex
SUBSTANCE USE AND TECHNOLOGY 23
partners or same-sex attraction: Results from the National Epidemiological Survey on
Alcohol and Related Conditions. Archives of Sexual Behavior, 41, 1185–1197.
doi:10.1007/s10508-012-9910-1
Gibbs, J. J., & Rice, E. (2016). The social context of depression symptomology in sexual
minority male youth: Determinants of depression in a sample of Grindr users. Journal of
Homosexuality, 63, 278–299. doi:10.1080/00918369.2015.1083773
Goedel, W. C., & Duncan, D. T. (2015). Geosocial-networking app usage patterns of gay,
bisexual, and other men who have sex with men: Survey among users of Grindr, a mobile
dating app. JMIR Public Health and Surveillance, 1, e4. doi:10.2196/publichealth.4353
Goedel, W. C., & Duncan, D. T. (2016). Contextual factors in geosocial-networking smartphone
application use and engagement in condomless anal intercourse among gay, bisexual, and
other men who have sex with men who use Grindr. Sexual Health, 13, 549–554.
doi:10.1071/SH16008
Goedel, W. C., Hagen, D., Halkitis, P. N., Greene, R. E., Griffin-Tomas, M., Brooks, F. A., …
Duncan, D. T. (2017). Post-exposure prophylaxis awareness and use among men who
have sex with men in London who use geosocial-networking smartphone applications.
AIDS Care, 29, 579–586. doi:10.1080/09540121.2016.1259455
Goedel, W. C., Halkitis, P. N., & Duncan, D. T. (2016). Behavior- and partner-based HIV risk
perception and sexual risk behaviors in men who have sex with men (MSM) who use
geosocial-networking smartphone applications in New York City. Journal of Urban
Health, 93, 400–406. doi:10.1007/s11524-016-0043-z
Goedel, W. C., Safren, S. A., Mayer, K. H., & Duncan, D. T. (2017). Community-level norms
and condomless anal intercourse among gay, bisexual, and other men who have sex with
SUBSTANCE USE AND TECHNOLOGY 24
men who use geosocial-networking smartphone applications in the Deep South. Journal
of HIV/AIDS & Social Services, 16, 382–389. doi:10.1080/15381501.2017.1344599
Goldberg, S., Strutz, K. L., Herring, A. A., & Halpern, C. T. (2013). Risk of substance abuse and
dependence among young adult sexual minority groups using a multidimensional
measure of sexual orientation. Public Health Reports, 128, 144–152.
doi:10.1177/003335491312800304
Grindr. (2015). Grindr: About. Retrieved from https://www.grindr.com/about/
Hagger-Johnson, G., Taibjee, R., Semlyen, J., Fitchie, I., Fish, J., Meads, C., & Varney, J.
(2013). Sexual orientation identity in relation to smoking history and alcohol use at age
18/19: Cross-sectional associations from the Longitudinal Study of Young People in
England (LSYPE). BMJ Open, 3, e002810. doi:10.1136/bmjopen-2013-002810
Hatzenbuehler, M. L., Corbin, W. R., & Fromme, K. (2008). Trajectories and determinants of
alcohol use among LGB young adults and their heterosexual peers: Results from a
prospective study. Developmental Psychology, 44, 81–90. doi:10.1037/0012-
1649.44.1.81
Holloway, I. W., Rice, E., Gibbs, J., Winetrobe, H., Dunlap, S., & Rhoades, H. (2014).
Acceptability of smartphone application-based HIV prevention for among young men
who have sex with men. AIDS and Behavior, 18, 285–296. doi:10.1007/s10461-013-
0671-1
Holloway, I. W., Pulsipher, C. A., Gibbs, J., Barman-Adhikari, A., & Rice, E. (2015). Network
influences on HIV risk behaviors of gay, bisexual and other men who have sex with men
using geosocial networking applications. AIDS and Behavior, 19(Suppl. 2), 112–122.
doi:10.1007/s10461-014-0989-3
SUBSTANCE USE AND TECHNOLOGY 25
Holloway, I. W., Tan, D., Gildner, J. L., Beougher, S. C., Pulsipher, C., Montoya, J. A., …
Leibowitz, A. (2017). Facilitators and barriers to pre-exposure prophylaxis willingness
among young men who have sex with men who use geosocial networking applications in
California. AIDS Patient Care and STDs, 31, 517–527. doi:10.1089/apc.2017.0082
Horvath, K. J., Bowen, A. M., & Williams, M. L. (2006). Virtual and physical venues as contexts
for HIV risk among rural men who have sex with men. Health Psychology, 25, 237–242.
doi:10.1037/0278-6133.25.2.237
Horvath, K. J., Rosser, B. S., & Remafedi, G. (2008). Sexual risk taking among young Internet-
using men who have sex with men. American Journal of Public Health, 98, 1059–1067.
doi:10.2105/AJPH.2007.111070
Hospers, H. J., Kok, G., Harterink, P., & de Zwart, O. (2005). A new meeting place: Chatting on
the Internet, e-dating and sexual risk behaviour among Dutch men who have sex with
men. AIDS, 19, 1097–1101. doi:10.1097/01.aids.0000174457.08992.62
Kelly, J., Davis, C., & Schlesinger, C. (2015). Substance use by same sex attracted young
people: Prevalence, perceptions and homophobia. Drug and Alcohol Review, 34, 358–
365. doi:10.1111/dar.12158
Kipke, M. D., Weiss, G., Ramirez, M., Dorey, F., Ritt-Olson, A., Iverson, E., & Ford, W. (2007).
Club drug use in Los Angeles among young men who have sex with men. Substance Use
& Misuse, 42, 1723–1743. doi:10.1080/10826080701212261
Kipke, M. D., Weiss, G., & Wong, C. F. (2007). Residential status as a risk factor for drug use
and HIV risk among young men who have sex with men. AIDS and Behavior, 11(Suppl.
2), 56–69. doi:10.1007/s10461-006-9204-5
Kuhns, L. M., Kwon, S., Ryan, D. T., Garofalo, R., Phillips, G., II, & Mustanski, B. S. (2015).
SUBSTANCE USE AND TECHNOLOGY 26
Evaluation of respondent-driven sampling in a study of urban young men who have sex
with men. Journal of Urban Health, 92, 151–167. doi:10.1007/s11524-014-9897-0
Kurtz, S. P., Stall, R. D., Buttram, M. E., Surratt, H. L., & Chen, M. (2013). A randomized trial
of a behavioral intervention for high risk substance-using MSM. AIDS and Behavior, 17,
2914–2926. doi:10.1007/s10461-013-0531-z
Lachowsky, N. J., Lal, A., Forrest, J. I., Card, K. G., Cui, Z., Sereda, P., … Hogg, R. S. (2016).
Including online-recruited seeds: A respondent-driven sample of men who have sex with
men. Journal of Medical Internet Research, 18, e51. doi:10.2196/jmir.5258
Lorimer, K., Flowers, P., Davis, M., & Frankis, J. (2016). Young men who have sex with men’s
use of social and sexual media and sex-risk associations: Cross-sectional, online survey
across four countries. Sexually Transmitted Infections, 92, 371–376.
doi:10.1136/sextrans-2015-052209
Macapagal, K., Coventry, R., Puckett, J. A., Phillips, G., II, & Mustanski, B. (2016). Geosocial
networking app use among men who have sex with men in serious romantic relationships.
Archives of Sexual Behavior, 45, 1513–1524. doi:10.1007/s10508-016-0698-2
MacKellar, D. A., Gallagher, K. M., Finlayson, T., Sanchez, T., Lansky, A., & Sullivan, P. S.
(2007). Surveillance of HIV risk and prevention behaviors of men who have sex with
men—a national application of venue-based, time-space sampling. Public Health
Reports, 122(1, Suppl.), 39–47. doi:10.1177/00333549071220S107
McCabe, S. E., Boyd, C., Hughes, T. L., & d’Arcy, H. (2003). Sexual identity and substance use
among undergraduate students. Substance Abuse, 24, 77–91.
doi:10.1023/A:1023768215020
McCabe, S. E., Hughes, T. L., Bostwick, W. B., West, B. T., & Boyd, C. J. (2009). Sexual
SUBSTANCE USE AND TECHNOLOGY 27
orientation, substance use behaviors and substance dependence in the United States.
Addiction, 104, 1333–1345. doi:10.1111/j.1360-0443.2009.02596.x
Meyer, I. H., & Wilson, P. A. (2009). Sampling lesbian, gay, and bisexual populations. Journal
of Counseling Psychology, 56, 23–31. doi:10.1037/a0014587
Mutchler, M. G., McKay, T., Candelario, N., Liu, H., Stackhouse, B., Bingham, T., & Ayala, G.
(2011). Sex drugs, peer connections, and HIV: Use and risk among African-American,
Latino, and multiracial young men who have sex with men in Los Angeles and New
York. Journal of Gay & Lesbian Social Services, 23, 271–295.
doi:10.1080/10538720.2011.560100
National Institute on Drug Abuse. (2012). Health consequences of drug abuse. Retrieved from
http://www.drugabuse.gov/related-topics/medical-consequences-drug-abuse
Operario, D., Choi, K.-H., Chu, P. L., McFarland, W., Secura, G. M., Behel, S., … Valleroy, L.
(2006). Prevalence and correlates of substance use among young Asian Pacific Islander
men who have sex with men. Prevention Science, 7, 19–29. doi:10.1007/s11121-005-
0018-x
Parsons, J. T., Lelutiu-Weinberger, C., Botsko, M., & Golub, S. A. (2013). Predictors of day-
level sexual risk for young gay and bisexual men. AIDS and Behavior, 17, 1465–1477.
doi:10.1007/s10461-012-0206-1
Pequegnat, W., Rosser, B. R. S., Bowen, A. M., Bull, S. S., DiClemente, R. J., Bockting, W. O.,
… Konstan, J. (2007). Conducting Internet-based HIV/STD prevention survey research:
Considerations in design and evaluation. AIDS and Behavior, 11, 505–521.
doi:10.1007/s10461-006-9172-9
SUBSTANCE USE AND TECHNOLOGY 28
Reisner, S. L., Mimiaga, M. J., Johnson, C. V., Bland, S., Case, P., Safren, S. A., & Mayer, K. H.
(2010). What makes a respondent-driven sampling “seed” productive? Example of
finding at-risk Massachusetts men who have sex with men. Journal of Urban Health, 87,
467–479. doi:10.1007/s11524-010-9439-3
Rendina, H. J., Jimenez, R. H., Grov, C., Ventuneac, A., & Parsons, J. T. (2014). Patterns of
lifetime and recent HIV testing among men who have sex with men in New York City
who use Grindr. AIDS and Behavior, 18, 41–49. doi:10.1007/s10461-013-0573-2
Rice, E., Holloway, I., Winetrobe, H., Rhoades, H., Barman-Adhikari, A., Gibbs, J., … Dunlap,
S. (2012). Sex risk among young men who have sex with men who use Grindr, a
smartphone geosocial networking application. Journal of AIDS & Clinical Research, S4,
005. doi:10.4172/2155-6113.S4-005
Shoptaw, S., Weiss, R. E., Munjas, B., Hucks-Ortiz, C., Young, S. D., Larkins, S., … Gorbach,
P. M. (2009). Homonegativity, substance use, sexual risk behaviors, and HIV status in
poor and ethnic men who have sex with men in Los Angeles. Journal of Urban Health,
86(Suppl. 1), 77–92. doi:10.1007/s11524-009-9372-5
Siegler, A. J., Sanchez, T., Sineath, R. C., Grey, J., Kahle, E., & Sullivan, P. S. (2015).
Knowledge and awareness of acute human immunodeficiency virus infection among
mobile app-using men who have sex with men: A missed public health opportunity. Open
Forum Infectious Diseases, 2, ofv016. doi:10.1093/ofid/ofv016
Sullivan, P. S., Khosropour, C. M., Luisi, N., Amsden, M., Coggia, T., Wingood, G. M., &
DiClemente, R. J. (2011). Bias in online recruitment and retention of racial and ethnic
minority men who have sex with men. Journal of Medical Internet Research, 13, e38.
doi:10.2196/jmir.1797
SUBSTANCE USE AND TECHNOLOGY 29
Thiede, H., Valleroy, L. A., MacKellar, D. A., Celentano, D. D., Ford, W. L., Hagan, H., …
Torian, L. V. (2003). Regional patterns and correlates of substance use among young
men who have sex with men in 7 US urban areas. American Journal of Public Health, 93,
1915–1921. doi:10.2105/AJPH.93.11.1915
Traube, D. E., Schrager, S. M., Holloway, I. W., Weiss, G., & Kipke, M. D. (2013).
Environmental risk, social cognition, and drug use among young men who have sex with
men: Longitudinal effects of minority status on health processes and outcomes. Drug and
Alcohol Dependence, 127, 1–7. doi:10.1016/j.drugalcdep.2012.06.007

SUBSTANCE USE AND TECHNOLOGY 30
Chapter 2: Relevant Theory
This dissertation used two models for understanding the behaviors (i.e., substance use
and utilization of technology for communication) of YMSM. Minority stress theory (MST)
posits that sexual minority individuals experience unique stressors related to their sexual identity
(Meyer, 2003). These stressors are in turn associated with worse substance use outcomes
(Hatzenbuehler, Hilt, & Nolen-Hoeksema, 2010; Mays, Cochran, & Barnes, 2007), because
using substances is a form of coping with increased stress (Feinstein & Newcomb, 2016). In
addition to MST, this dissertation applied an explanatory model to address the impact of
technology on subcultures. The social information processing (SIP) perspective explains that
humans have the capacity and ability to adapt to the way media are used. More specifically,
individuals learn how to connect with others through new technology (Walther, 1996), which
then has an impact on the culture of communities (Mowlabocus, 2010).
Minority Stress Theory
Given disparities in substance use and the associated health and sexual risks among
YMSM, social researchers have developed explanatory models using an MST framework. MST
highlights the additional stresses experienced by sexual minorities due to their sexual minority
identity status. This model of understanding stress is based on Lazarus and Folkman’s (1984)
stress and coping perspective, which explains that negative social experiences are distal stressors
from an individual’s environment that are then internalized through subjective appraisal and
manifested as proximal stressors (Lazarus & Folkman, 1984; Meyer, 2003). Distal stress
includes discrimination and victimization. Proximal stress includes concealment of sexual
identity, expectation of rejection (after a disclosure of being a sexual minority), and internalized
stigma (i.e., internalized homophobia; Meyer, 2003).
SUBSTANCE USE AND TECHNOLOGY 31
Numerous studies have associated minority stressors with both illicit drug use and
substance abuse disorder symptomology (Duncan, Hatzenbuehler, & Johnson, 2014; McCabe,
Bostwick, Hughes, West, & Boyd, 2010; Rosario, Schrimshaw, & Hunter, 2009; Ryan, Huebner,
Diaz, & Sanchez, 2009; Shoptaw et al., 2009; Traube, Schrager, Holloway, Weiss, & Kipke,
2013; Wong, Weiss, Ayala, & Kipke, 2010). Experiences of homophobia and other forms of
discrimination are associated with higher odds of both illicit drug use (Traube et al., 2013; Wong
et al., 2010) and having a substance abuse disorder (McCabe et al., 2010). Findings from a study
on sexual minority hate crimes indicate that higher prevalence of sexual minority victimization
in a geographical area is associated with increased illicit substance use (Duncan et al., 2014).
Further, rejection after disclosure of sexual minority status is associated with both illicit drug use
and symptoms of substance abuse disorders (Rosario, Schrimshaw, & Hunter, 2009; Ryan et al.,
2009). One study found that higher levels of family rejection following disclosure is associated
with more than 3 times the odds of illicit substance use (Ryan et al., 2009). Internalized stigma
(i.e., homonegativity or homophobia) is also correlated with illicit substance use (Shoptaw et al.,
2009). Further, minority stressors are highly associated with increased mental health distress
(e.g., depression; Almeida, Johnson, Corliss, Molnar, & Azrael, 2009; Russell, Ryan, Toomey,
Diaz, & Sanchez, 2011; Ryan et al., 2009). Distress in turn is associated with increased odds of
both binge drinking and other substance abuse symptoms in this population (Rosario,
Schrimshaw, & Hunter, 2006; Wong, Kipke, & Weiss, 2008).
Social Information Processing Perspective
SIP provides an explanatory framework for understanding how computer-mediated
communication (CMC) can be as impactful, if not more impactful, in building relationships
compared to face-to-face communication (Walther, 1996). A SIP model recognizes that
SUBSTANCE USE AND TECHNOLOGY 32
geosocial networking applications (GSNAs) have provided a new cultural and digital framework,
and YMSM have adopted this technology because it enhances interpersonal exchange
(Mowlabocus, 2010). Walther (1996) explained that for CMC technology to be as effective as
face-to-face interaction, it must allow three types of relationship-building interactions:
impersonal, interpersonal, and hyperpersonal.
The GSNA home screen digital environment is by nature an impersonal interaction,
which is defined as any exchange that conveys formal information (Walther, 1996). By entering
the home screen of most GSNAs, YMSM can gather physical information about other users by
observing profile photos (Holloway et al., 2014; Landovitz et al., 2013). If the user chooses to
display a face or body photo, this type of digital observation may be analogous to viewing an
individual in a venue. GSNA profiles also can provide users with an individual’s likes, dislikes,
physical attributes, sexual proclivities, HIV status, and last date of testing for sexually
transmitted infections (Holloway et al., 2014). This type of impersonal interaction is not readily
available in a face-to-face venue context, and if investigated in face-to-face communication,
more time may be required to naturally share this information. Although this type of impersonal
interaction may be available on a gay dating website (e.g., Manhunt) or social media (e.g.,
Facebook) profile context, GSNAs offer location data, allowing proximity information to be
exchanged (Mowlabocus, 2010).
Interpersonal interaction is any exchange of information for developing social
relationships. SIP states that over time, CMC can build the same level of interpersonal
interaction as face-to-face communication. However, for CMC to be as effective, communicators
using the CMC technology must be driven to develop social relationships. The literature has
indicated that YMSM use GSNAs for the primary role of developing social relationships. A
SUBSTANCE USE AND TECHNOLOGY 33
study by Holloway et al. (2014) of Grindr users in Los Angeles asked participants to report on
their rationale for using GSNAs. Four of the top-reported rationales for GSNA use were related
to social relationships: 80% reported a desire to “make new friends,” 67% reported to “meet
people to hook up with,” 65% reported to “meet people to date,” and 65% reported to “connect
with the gay community” (Holloway et al., 2014). SIP explains that although interpersonal
interaction can be achieved, the process is slower in CMC compared to face-to-face interactions
because less social information is exchanged in text messages. Therefore, the rate of exchange
directly influences the type and depth of interpersonal interactions. Additionally, whereas
Walther (1996) originally focused on text-based CMC, recent literature has identified that the
ability to send photographs, videos, gifs, emojis, and emoticons can improve interpersonal digital
exchanges. These types of visual and pictorial messages have enhanced the ability to convey
emotions and develop interpersonal interactions through CMC (Elder, 2018; Liebman & Gergle,
2016).
The third type of interaction, termed hyperpersonal, refers to social interactions that are
more socially desirable than what is commonly expected in parallel face-to-face communication
(Walther, 1996). Different than face-to-face communications, CMC on GSNAs allows message
senders to have an optimized self-presentation. GSNA users have full control over the
information they share on their profile and the direct messages they send (Jaspal, 2017). Users
can choose to provide ideal photos of themselves and focus attention on language rather than
physical presentation, which is the focus of face-to-face interactions (Walther, 1996). This
cognitive reallocation allows users to shift their focus away from in-the-moment presentation to
language and photograph selection. Additionally, a key feature of hyperpersonal interactions is
having an idealized perception of the receiver. Essentially, by removing the face-to-face
SUBSTANCE USE AND TECHNOLOGY 34
interaction, users of CMC fill gaps in the receiver’s personal information with idealized qualities
(Walther, 1996). Walther (1996) explained that the level of idealized perception is based on two
factors: group social categorization and proximity. Group social categorization is a similar
process as selving and othering, which is described in social identity theory. When individuals
can categorize others as part of a personal reference group, personal ideals and values are
assumed to also be shared (Tajfel, 1981; Turner, 1982). The GSNAs that YMSM use are
advertised as digital environments for gay and bisexual men (Rice et al., 2012). This means that
by opening the home screen of a GSNA, YMSM can assume that the individuals they are
viewing are part of the same reference group. This hyperpersonal interaction is only enhanced by
the ability of users to view the proximity of the receivers of their messages. Walther (1996)
explained that when senders of CMC know that the receivers of their message are closer in
proximity, it intensifies the interaction.
Although the original intent of GSNA technologies for sexual minority populations was
to facilitate finding sexual partners (Leslie, 2009), the nature of GSNA use has evolved because
GSNAs allow for all three types of interaction described in SIP. Recent literature has identified
that YMSM use GSNAs multiple times throughout the day (Goedel & Duncan, 2016; Holloway
et al., 2014) and for multiple purposes, including to make new friends, connect with the gay
community, and kill time (Holloway et al., 2014; Phillips et al., 2014; Rice et al., 2012).
Qualitative research has identified that YMSM report less of a need to visit gay-specific venues
because they have access to men anywhere through GSNAs (Tudor, 2012). Additionally,
research has indicated that YMSM are more readily using technology and foregoing the use of
venues to meet other YMSM (Grov, 2012; Grov & Crow, 2012; Zablotska, Holt, & Prestage,
2012).
SUBSTANCE USE AND TECHNOLOGY 35
References
Almeida, J., Johnson, R. M., Corliss, H. L., Molnar, B. E., & Azrael, D. (2009). Emotional
distress among LGBT youth: The influence of perceived discrimination based on sexual
orientation. Journal of Youth and Adolescence, 38, 1001–1014. doi:10.1007/s10964-009-
9397-9
Duncan, D. T., Hatzenbuehler, M. L., & Johnson, R. M. (2014). Neighborhood-level LGBT hate
crimes and current illicit drug use among sexual minority youth. Drug and Alcohol
Dependence, 135, 65–70. doi:10.1016/j.drugalcdep.2013.11.001
Elder, A. M. (2018). What words can’t say: Emoji and other non-verbal elements of
technologically-mediated communication. Journal of Information, Communication and
Ethics in Society, 16, 2–15. doi:10.1108/JICES-08-2017-0050
Feinstein, B. A., & Newcomb, M. E. (2016). The role of substance use motives in the
associations between minority stressors and substance use problems among young men
who have sex with men. Psychology of Sexual Orientation and Gender Diversity, 3, 357–
366. doi:10.1037/sgd0000185
Goedel, W. C., & Duncan, D. T. (2016). Contextual factors in geosocial-networking smartphone
application use and engagement in condomless anal intercourse among gay, bisexual, and
other men who have sex with men who use Grindr. Sexual Health, 13, 549–554.
doi:10.1071/SH16008
Grov, C. (2012). HIV risk and substance use in men who have sex with men surveyed in
bathhouses, bars/clubs, and on Craigslist.org: Venue of recruitment matters. AIDS and
Behavior, 16, 807–817. doi:10.1007/s10461-011-9999-6
SUBSTANCE USE AND TECHNOLOGY 36
Grov, C., & Crow, T. (2012). Attitudes about and HIV risk related to the “most common place”
MSM meet their sex partners: Comparing men from bathhouses, bars/clubs, and
Craigslist.org. AIDS Education and Prevention, 24, 102–116.
doi:10.1521/aeap.2012.24.2.102
Hatzenbuehler, M. L., Hilt, L. M., & Nolen-Hoeksema, S. (2010). Gender, sexual orientation,
and vulnerability to depression. In J. C. Chrisler & D. R. McCreary (Eds.), Handbook of
gender research in psychology (Vol. 2, pp. 133–151). New York, NY: Springer.
Holloway, I. W., Rice, E., Gibbs, J., Winetrobe, H., Dunlap, S., & Rhoades, H. (2014).
Acceptability of smartphone application-based HIV prevention for among young men
who have sex with men. AIDS and Behavior, 18, 285–296. doi:10.1007/s10461-013-
0671-1
Jaspal, R. (2017). Gay men’s construction and management of identity on Grindr. Sexuality &
Culture, 21, 187–204. doi:10.1007/s12119-016-9389-3
Landovitz, R. J., Tseng, C.-H., Weissman, M., Haymer, M., Mendenhall, B., Rogers, K., …
Shoptaw, S. (2013). Epidemiology, sexual risk behavior, and HIV prevention practices of
men who have sex with men using GRINDR in Los Angeles, California. Journal of
Urban Health, 90, 729–739. doi:10.1007/s11524-012-9766-7
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York, NY: Springer.
Leslie, J. (2009, October 1). Grindr app makes it easy to be easy. Out & About Nashville.
Retrieved from https://www.outandaboutnashville.com/story/grindr-app-makes-it-easy-
be-easy
Liebman, N., & Gergle, D. (2016, February). It’s (not) simply a matter of time: The relationship
between CMC cues and interpersonal affinity. Paper presented at the 19th ACM
SUBSTANCE USE AND TECHNOLOGY 37
Conference on Computer-Supported Cooperative Work & Social Computing, San
Francisco, CA.
Mays, V. M., Cochran, S. D. & Barnes, N. W. (2007). Race, race-based discrimination, and
health outcomes among African Americans. Annual Review of Psychology, 58, 201–225.
doi:10.1146/annurev.psych.57.102904.190212
McCabe, S. E., Bostwick, W. B., Hughes, T. L., West, B. T., & Boyd, C. J. (2010). The
relationship between discrimination and substance use disorders among lesbian, gay, and
bisexual adults in the United States. American Journal of Public Health, 100, 1946–1952.
doi:10.2105/AJPH.2009.163147
Meyer, I. H. (2003). Prejudice, social stress, and mental health in lesbian, gay, and bisexual
populations: Conceptual issues and research evidence. Psychosocial Bulletin, 129, 674–
697. doi:10.1037/0033-2909.129.5.674
Mowlabocus, S. (2010). Gaydar culture: Gay men, technology and embodiment in the digital
age. Burlington, VT: Ashgate.
Phillips, G., II, Magnus, M., Kuo, I., Rawls, A., Peterson, J., Jia, Y., … Greenberg, A. E. (2014).
Use of geosocial networking (GSN) mobile phone applications to find men for sex by
men who have sex with men (MSM) in Washington, DC. AIDS and Behavior, 18, 1630–
1637. doi:10.1007/s10461-014-0760-9
Rice, E., Holloway, I., Winetrobe, H., Rhoades, H., Barman-Adhikari, A., Gibbs, J., … Dunlap,
S. (2012). Sex risk among young men who have sex with men who use Grindr, a
smartphone geosocial networking application. Journal of AIDS & Clinical Research, S4,
005. doi:10.4172/2155-6113.S4-005
SUBSTANCE USE AND TECHNOLOGY 38
Rosario, M., Schrimshaw, E. W., & Hunter, J. (2006). A model of sexual risk behaviors among
young gay and bisexual men: Longitudinal associations of mental health, substance
abuse, sexual abuse, and the coming-out process. AIDS Education and Prevention, 18,
444–460. doi:10.1521/aeap.2006.18.5.444
Rosario, M., Schrimshaw, E. W., & Hunter, J. (2009). Disclosure of sexual orientation and
subsequent substance use and abuse among lesbian, gay, and bisexual youths: Critical
role of disclosure reactions. Psychology of Addictive Behaviors, 23, 175–184.
doi:10.1037/a0014284
Russell, S. T., Ryan, C., Toomey, R. B., Diaz, R. M., & Sanchez, J. (2011). Lesbian, gay,
bisexual, and transgender adolescent school victimization: Implications for young adult
health and adjustment. Journal of School Health, 81, 223–230. doi:10.1111/j.1746-
1561.2011.00583.x
Ryan, C., Huebner, D., Diaz, R. M., & Sanchez, J. (2009). Family rejection as a predictor of
negative health outcomes in White and Latino lesbian, gay, and bisexual young adults.
Pediatrics, 123, 346–352. doi:10.1542/peds.2007-3524
Shoptaw, S., Weiss, R. E., Munjas, B., Hucks-Ortiz, C., Young, S. D., Larkins, S., … Gorbach,
P. M. (2009). Homonegativity, substance use, sexual risk behaviors, and HIV status in
poor and ethnic men who have sex with men in Los Angeles. Journal of Urban Health,
86(Suppl. 1), 77–92. doi:10.1007/s11524-009-9372-5
Tajfel, H. (1981). Human groups and social categories: Studies in social psychology. New York,
NY: Cambridge University Press.
Traube, D. E., Schrager, S. M., Holloway, I. W., Weiss, G., & Kipke, M. D. (2013).
Environmental risk, social cognition, and drug use among young men who have sex with
SUBSTANCE USE AND TECHNOLOGY 39
men: Longitudinal effects of minority status on health processes and outcomes. Drug and
Alcohol Dependence, 127, 1–7. doi:10.1016/j.drugalcdep.2012.06.007
Tudor, M. (2012). Cyberqueer techno-practices: Digital space-making and networking among
Swedish gay men (Doctoral dissertation). Stockholm University, Stockholm, Sweden.
Turner, J. C. (1982). Towards a cognitive redefinition of the social group. In H. Tajfel (Ed.).
Social identity and intergroup relations (pp. 15–40). New York, NY: Cambridge
University Press.
Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal, and
hyperpersonal interaction. Communication Research, 23, 3–43.
doi:10.1177/009365096023001001
Wong, C. F., Kipke, M. D., & Weiss, G. (2008). Risk factors for alcohol use, frequent use, and
binge drinking among young men who have sex with men. Addictive Behaviors, 33,
1012–1020. doi:10.1016/j.addbeh.2008.03.008
Wong, C. F., Weiss, G., Ayala, G., & Kipke, M. D. (2010). Harassment, discrimination, violence
and illicit drug use among young men who have sex with men. AIDS Education and
Prevention, 22, 286–298. doi:10.1521/aeap.2010.22.4.286
Zablotska, I. B., Holt, M., & Prestage, G. (2012). Changes in gay men’s participation in gay
community life: Implications for HIV surveillance and research. AIDS and Behavior, 16,
669–675. doi:10.1007/s10461-011-9919-9

SUBSTANCE USE AND TECHNOLOGY 40
Chapter 3: Methods
This dissertation aimed to inform the use of innovative sampling methods by rigorously
comparing a geosocial networking application (GSNA)-recruited sample with a venue-recruited
sample. Specifically, this dissertation: (a) investigated differences between sample characteristics
of a GSNA-recruited sample and a venue-recruited sample of young men who have sex with men
(YMSM), (b) established the feasibility of using user density data to identify locations for
random recruitment using a GSNA, and (c) compared the costs and recruitment efficiency of
GSNA- and venue-based recruitment.
A cross-sectional online survey design was conducted in Los Angeles in April 2017
through April 2018 using two forms of recruitment. A sample of 111 YMSM was recruited, 68
using venue-based stratified probability sampling procedures and 43 using GSNA-based
probability sampling procedures. The venue-based stratified probability sampling methods were
informed by the Healthy Young Men’s Study conducted in Los Angeles (Ford et al., 2009). The
GSNA methods were informed by a previous study also conducted in Los Angeles (Rice et al.,
2012). Geographic information systems methods were also used to build the GSNA sampling
frame and were informed by a 2014 study conducted in Atlanta, Georgia (Delaney, Kramer,
Waller, Flanders, & Sullivan, 2014).
The geographic sampling frame (GSF) in which recruitment was conducted went through
two evolutions. Initially GSF A was defined by the boundaries of zip code 90036, a portion of
West Hollywood, California (see Figure 3.1). Upon initial pilot tests of the GSNA-based
probability sampling methods and discussions with representatives from the study GSNA (i.e.,
Hornet), it was determined that the area covered by GSF A lacked the GSNA user diversity
needed to meet the projected sample size. Therefore, the area of the study GSF was increased by
SUBSTANCE USE AND TECHNOLOGY 41
almost 4 times to form GSF B, which was defined by census bureau tracts and encompassed a
portion of West Hollywood, Hollywood, and Koreatown (see Figure 3.2). Recruitment for the
venue-based sample was completed in GSF B while the methods for recruitment with the GSNA
were being pilot tested. Once the methods for GSNA were established, it was determined
through consultations with the health strategist of the study GSNA that the GSF should be
expanded to maximize the number of unique users. As such, the GSF for the GSNA-based
sample was expanded to the southern portion of Los Angeles County, with a northern boundary
of Burbank, southern boundary of Long Beach, western boundary of Santa Monica, and eastern
boundary of Pomona (see Figure 3.3).
Both recruitment methods followed three stages: (1) building of the sampling frame, (2)
enumeration of the study sampling frame, and (3) recruitment of participants using a monthly
sampling calendar.
Procedures
Stage 1: Building a Sampling Frame
Venue-based stratified probability sample. Following the procedures of the Healthy
Young Men’s Study, for the first 2 months of the study, an exhaustive list of venues frequented
by YMSM in GSF A was developed using local lesbian, gay, bisexual, and transgender
publications and Internet sites (Ford et al., 2009). Next, the study team conducted brief
interviews with men in public areas in GSF A. No identifying information was collected during
these interviews. One study team member approached potential interviewees who appeared to fit
the study population (i.e., 18 to 24 years old, male), explained that the research team was
attempting to compile a list of venues frequented by YMSM, and asked interviewees to report
known locations. Brief interviews continued until the study team no longer received unique
SUBSTANCE USE AND TECHNOLOGY 42
venue reports (a final list of 41 venues). The final investigation involved a meeting with the
community advisory board (CAB; i.e., eight YMSM key informants). During the CAB meeting,
the study team presented the compiled list of venues and asked the CAB to nominate missing
venues and eliminate venues believed to be inappropriate. The CAB was also asked to provide
insight regarding the best times to find YMSM during four standard 4-hour sampling periods (8
a.m.–12 p.m., 12 p.m.–4 p.m., 4 p.m.–8 p.m., and 8 p.m.–12 a.m.). Participants received a $30
gift card incentive, and the work resulted in an exhaustive list of 35 venues currently frequented
by YMSM in GSF A. Investigation of each venue using Google’s “Popular Times” function,
information from each venue’s website, and information gathered from the focus group
identified 173 periods with theoretically high YMSM attendance (i.e., venue, day, and time
period; VDT). When the GSF was expanded to GSF B, the same procedures were followed for
identifying additional appropriate venues. From brief street interviews, discussion with the CAB,
and online research, eight additional venues and 26 additional periods were identified (199 VDTs
in total).
GSNA probability sample. The CAB was asked to report on the most popular GSNA
used by YMSM in the GSF. Each proposed GSNA needed to meet three criteria to be included:
(a) ability to filter users by age, (b) ordering of user profiles by relative proximity to the study
smart device, and (c) reporting of the relative distance of users from the study smart device.
Representatives from the top three GSNAs were contacted by the study’s principal investigator
to inquire about interest in partnership in the current study. From these conversations, the study
GSNA was selected (i.e., Hornet).
Stage 2: Enumeration of the Sampling Frame
Venue-based stratified probability sample. Venue enumeration was conducted during
SUBSTANCE USE AND TECHNOLOGY 43
4 months in GSF A and an additional month when the GSF was expanded to GSF B. Type I
enumeration required one study team member to visit each venue during the identified 4-hour
periods (i.e., 199 VDT sampling periods) identified in Stage 1 (Ford et al., 2009). For 60
minutes, the study team member counted the number of individuals who appeared to fit the study
criteria (i.e., male, aged 18–24). That number was then multiplied by 4 to approximate a 4-hour
period. As in Ford et al. (2009), pilot enumerations were also conducted wherein one study team
member counted individuals who appeared to meet the study criteria and a second study team
member asked those individuals if they met the eligibility criteria (i.e., date of birth, gender, and
self-identification as gay, bisexual, or uncertain about sexual orientation or report of having sex
with men. Based on the pilot enumerations, team members correctly identified eligible
individuals 65% of the time. Therefore, the number of individuals counted in each type I
enumeration was multiplied by 65% to approximate the individuals who were hypothetically
eligible. For example, if 10 people were counted during the 1-hour type I enumeration, then
during a 4-hour period, approximately 40 people would be expected at that venue. To
approximate how many eligible people the study team could correctly identify, 40 was then
multiplied by 65%, resulting in a type I enumeration of 26 individuals during the VDT sampling
period.
An estimated eight individuals in a 4-hour period was identified as the threshold for an
appropriate VDT based on recommendations from previous studies (Ford et al., 2009; MacKellar
et al., 2007). If a type I enumeration for a specific VDT resulted in an estimate of fewer than
eight eligible individuals, then the VDT required a type II enumeration. Of the 173 type I
enumerations completed in GSF A, 122 were determined to be appropriate and 51 required a
SUBSTANCE USE AND TECHNOLOGY 44
type II enumeration. Of the additional 26 type I enumerations completed in GSF B, 22 were
determined to be appropriate and four required a type II enumeration.
Type II enumerations were conducted by two study team members for two 1-hour periods
separated by 1 hour and followed the same procedures as the pilot enumerations. The number of
individuals counted who met the study criteria during the two 1-hour periods was multiplied by 2
to approximate the number of eligible men at the specific VDT during a 4-hour period. If a VDT
was found to have an estimate of fewer than eight, it was excluded from the sampling frame. Of
the 55 type II enumerations completed, 12 were identified as appropriate and 43 as inappropriate
and thus excluded as potential VDTs. At the end of Stage 2, 156 VDTs were identified as
appropriate for recruitment.
GSNA probability sample. The following procedures were used for both GSF A and
GSF B and informed by procedures used to collect user density data in Atlanta, Georgia, by
Delaney and colleagues (2014). Using the study GSNA, anonymous geospatial observation data
were collected from the application interface.
For GSF A, a density map was created by observing the user density data in a 4-by-5 grid
of 1-kilometer-separated geographic points (20 points in total), which covered GSF A. The same
procedure was completed for GSF B for roughly an 11-by-8 grid of 1-kilometer-separated points
(nine points were unnecessary due to the nonuniform nature of the eastern boundary; see Figure
3.4), resulting in 79 points. Observing density data was accomplished by changing the GPS
information supplied to the study smart device (using a third-party application titled Fake GPS
Location), therefore allowing the study team to stay in a fixed location. This procedure allowed
for rapid user density observation of the study sampling area, which minimized any potential
change in the user density during the period required to observe all points in the GSF.
SUBSTANCE USE AND TECHNOLOGY 45
After opening the study GSNA, a study team member filtered the visible profiles by
reported age (i.e., 18 to 24). Each profile listed the associated geographic distance from the study
smart device. Second, the study team member scrolled through the listed user profiles and noted
the number of users (k) within a specific distance (j; i.e., 1 mile) of the study smart device.
Geographic user density data was determined by calculating the area of a circle with a radius j,
and then dividing k by that area (density = k/[πj
2
]). For example, if 20 profiles listed the distance
from the study smart device as less than 1 mile, the average user density at that geographic point
was 6.37 users per square mile (i.e., 20/[π1
2
] = 23.87).
The observation of user density data was completed for each sampling, day, and time
period (SDTP; four periods × 7 days = 28) for all points in GSF A and GSF B. The periods used
were the midpoints of each VDT period: 10 a.m., 2 p.m., 6 p.m., and 10 p.m. Two items of data
were collected for each geographic grid point: a GPS point and an associated user density value.
This procedure was completed at each grid point 3 times for every SDTP during a 1-month
period. The user density values were then averaged across the three collection times for each grid
point of each SDTP. For example, in GSF B, the user density was collected for all 79 points on
Monday at 10 a.m., on three Mondays, and then averaged across each of the grid points. This
means that Point 35 of 79 on Mondays at 10 a.m. had a density of 6.68, 7.00, and 5.41 on three
Mondays. This density was then averaged (6.37 users per square mile) and converted to square
kilometers to approximate the expected density at that specific point on any Monday at 10 a.m.
In total, 6,636 points of data (i.e., 28 SDTP × 79 points × three observations) were collected to
estimate the user density on every day and period in GSF B. Manually collecting 84 sets (28
SDTPs × three collection times) of user density data from 79 points took an average of 22.96
minutes (SD = 2.01 minutes) per observation period.
SUBSTANCE USE AND TECHNOLOGY 46
The user density data from GSF B were then imported into ArcMap 10.5. The data file
contained 79 observations (each grid point) with two types of data: (a) a converted GPS point to
X–Y coordinates and (b) a corresponding user density for each SDTP. The data were visualized
for each SDTP using the X–Y coordinates, converted into an interpolated raster (i.e., a map layer
that averages the density between data points to create a full GSF density layer), and then
divided to approximate the density at any point in GSF B (see Figure 3.5). Circular focal
statistical analyses were then used to calculate the radial distance required to capture eight users
(to allow comparison to venue-based procedures). These different-colored sections corresponded
with the average radial distance required to observe eight users on a specific day and time:
Monday at 10 a.m. (see Figure 7). The number of sampling points needed for each SDTP for
GSF B was determined by three criteria: point-based circular buffers must have (a) maximized
coverage of the entire GSF B, (b) minimized buffer overlap, and (c) minimized coverage of areas
outside the GSF B boundary. Points were placed on the map using trial and error. Once the
necessary number of points (adhering to the three criteria) were placed on the map, the buffer
tool was used to visualize the radial distance achieved by each point. This process continued
until the three criteria were met (see Figure 3.6). The center of each circle in Figure 3.6
represents the GPS coordinates of the five unique day, time, and sampling locations (DTSLs) for
recruitment of YMSM. Each SDTP user density data was processed using the same procedures
in ArcMap 10.5. Once completed, 140 unique DTSLs had been determined. Each DTSL had a
corresponding day and time for recruitment, a GPS point, and a buffer distance for recruiting.
Upon piloting recruitment in GSF A and GSF B, the GSF was expanded to GSF C due to
a lack of user diversity (i.e., limited unique users). GSF C covered an area approximately 50
times larger than GSF B. Because of the considerable area, the process for collecting user density
SUBSTANCE USE AND TECHNOLOGY 47
data through a 1-kilometer-spaced grid could not be achieved in a timely manner (approximately
20 hours). Therefore, five points were chosen as the locations for recruitment (see Figure 3.7).
Similar to the process of placing sampling points in ArcMap 10.5, these five points were chosen
using the three criteria, but with a standard buffer area of 8 miles. Once completed, GSF C
featured 140 DTSLs.
Stage 3: Recruitment and Data Collection
Recruitment for the venue-based sample and pilot testing of GSNA-based recruitment
occurred in GSF B. When it was determined that recruitment of the GSNA sample required GSF
C, venue-based sampling had concluded. The following procedures were used throughout the
data collection process.
A monthly sampling calendar was used to randomly select both VDTs and DTSLs.
Creation of calendars followed the procedures used in Ford et al. (2009), except for two
adaptations: exclusion of large events taking place in the GSF and procedures to limit one
recruitment strategy per sampling period. At the beginning of each recruitment month and before
the monthly sampling calendar was created, the study team researched YMSM events and
holidays in the GSF (referred to as wild cards; Ford et al., 2009) and recorded any that were
likely to be large and therefore would change the density of YMSM in the GSF (e.g., lesbian,
gay, bisexual, and transgender pride events). These events were placed in the sampling calendar
to inhibit recruitment during the associated period.
To create the monthly sampling calendar, all VDTs and DTSLs were loaded into a
spreadsheet. VDTs and DTSLs were categorized as weekday day, weekday night, weekend day,
and weekend night to allow for matching between recruitment methods, because it was
hypothesized that different individuals would be in the GSF at each of these times. One
SUBSTANCE USE AND TECHNOLOGY 48
randomly selected VDT or DTSL was placed on the appropriate day and time on the calendar
and a matched random selection was taken from the other recruitment method. If an event was
randomly selected and the corresponding day and time was not available, an additional random
selection was taken from that same recruitment strategy and so on, until an appropriate selection
event was randomly selected. Selection of VDTs and DTSLs continued until 10 VDTs and three
or four DTSLs (the study GSNA preferred to limit monthly recruitment periods to not
overburden users) were placed on the study calendar. Once the targeted sample size was
achieved for a recruitment strategy, the corresponding remaining sampling events for that month
were disregarded.
Venue-based stratified probability sample. In accordance with the monthly sampling
calendar, two or three study team members recruited a sample of participants from the specified
VDT (Ford et al., 2009). The same procedures in type II enumeration were followed. While one
study team member counted the number of men who appeared to be eligible, the remaining study
team members individually approached potential participants and asked if they were willing to
answer eligibility questions to support research. If the individual declined, he was thanked and
the refusal was noted. If the potential participant agreed to participate, he completed the
eligibility survey on the study iPad. At the end of the eligibility survey, the potential participant
created a unique study ID code using a list of questions about himself (e.g., provide the last two
letters of your first name, the two digits of your birth month, and the last two digits of your
phone number) and input his contact information (i.e., email address, mobile phone number).
The potential participant then received a $5 gift card incentive for completing the eligibility
survey. This $5 incentive was added as a procedure following initial pilot testing of the venue-
SUBSTANCE USE AND TECHNOLOGY 49
based procedures, during which it was observed that those who were approached to participate
were apathetic about being screened.
The eligibility data was received by the contact information manager, who determined if
individuals were eligible to take the study survey (i.e., met the study eligibility criteria and had a
unique study ID code indicating they had not previously taken the main study survey). The
contact information manager then provided a list of new eligible unique study IDs to the data
manager, who never had access to the contact information of the participants. The data manager
used the unique study IDs to create unique one-time web links to the main study survey and
provided the contact information manager with these links. The contact information manager
then messaged the potential participant using his preferred mode of contact with instructions to
take the main study survey, his unique study ID, and the unique link to the main study survey.
The contact information manager never had access to the main study survey data.
Once entering the main study survey website, potential participants had the opportunity
to read informed consent information. After giving informed consent, participants were
instructed to enter their unique study ID code. The main study survey asked participants to report
on demographic information, substance use characteristics, substance use determinants, and
sexual risk behaviors. Once completing the main study survey, participants were asked to
indicate if they had previously completed the study. This item was asked as a safeguard against
duplicate data. Once a participant completed the main study survey, the submitted data were
available to the data manager. The data manager provided a list of unique study ID codes of
individuals who completed the main study survey and a unique link to an associated
supplemental survey to the contact information manager. The contact information manager then
sent each participant a $25 downloadable gift card. In addition, the contact information manager
SUBSTANCE USE AND TECHNOLOGY 50
sent each participant who completed the main study survey a link to the supplemental survey
with instructions on how to complete the supplemental survey. The supplemental survey
followed the same procedures as the main study survey, except participants who completed it
received a $15 downloadable gift card.
GSNA probability sample. Recruitment for the GSNA sample followed similar
procedures outlined by Rice et al. (2012), except for the use of DTSLs, use of the monthly
sampling calendar for determining DTSLs, and direct messages sent from the study GSNA
company rather than from the study team. Instead of visiting the specified geographic point, the
study GSNA was technologically misled and provided with GPS coordinates for the DTSL
specified on the monthly sampling calendar. As noted, each DTSL had an associated GPS point
and radial distance (i.e., 8 miles). The recruitment process began by one study team member
opening the GSNA on the study smart device and making an anonymous user profile. After
filtering the visible user profiles by age eligibility (i.e., aged 18–24), the study team member
recorded all public GSNA user IDs within 8 miles of the study smart device. Concurrently, the
study GSNA (i.e., Hornet) sent out a preset direct message in the GSNA to each GSNA user
within 8 miles of the DTSL. The direct message appeared as a personal message from Hornet
and remained in the users’ message inbox. The message explained the study, noted incentives for
participation, and provided the link to the GSNA eligibility survey (e.g., “Hey, you appear to be
eligible for a USC study! Do you have 2 minutes to answer a few questions? If you’re eligible
and you take the survey you can earn up to $50”). The GSNA eligibility survey was the same as
the venue-based recruitment eligibility survey.
Once the information contact manager determined a potential participant’s eligibility
(met the study criteria and did not have a unique study ID code indicating the user had already
SUBSTANCE USE AND TECHNOLOGY 51
taken the study survey), the contact information manager followed the same procedures as those
for the venue-based recruited sample (i.e., emailed a list of unique study ID codes to the data
manger, who replied with unique main study survey links, etc.). Following recruitment, all
procedures were identical to the venue-based procedures, with the exception of no eligibility
survey incentive and a $35 main study survey incentive. Pilot tests of the GSNA-based methods
indicated that a $5 eligibility survey incentive did not affect the rate of survey completion,
whereas a higher advertised main study survey incentive increased completions of the eligibility
survey.
Measurement
Aim 1 of this dissertation focused on comparing both samples by demographics,
substance use, psychosocial determinants of substance use, and sexual risk. Respondents
reported their age, sexual identity, race and ethnicity, relationship status, highest education level,
employment status, monthly income, and GSNA use (days and periods of use, frequency of use,
length of time spent using app). Additionally, level of connection with the gay community was
assessed using the 8-item Identification with the Gay Community Scale (Vanable, McKirnan, &
Stokes, 1998),

which has been shown to have an acceptable level of internal consistency (α = .67;
Fernández et al., 2007; Holloway et al., 2012).
Substance use was measured with a binary lifetime item and an item measuring the
number of times of use during the previous 30 days (i.e., 0, 1–2, 3–9, 10–19, 20–39, 40 or more)
for each substance. In addition to alcohol consumption, binge drinking (i.e., five or more drinks
in 2 hours), and marijuana use, substance use of three classes of substances were assessed: illicit
(i.e., methamphetamine, cocaine, heroin, ecstasy, poppers, and LSD); prescription (i.e., pain
SUBSTANCE USE AND TECHNOLOGY 52
pills, tranquilizers, stimulants, muscle relaxants, erectile dysfunction drugs, and steroids); and
emerging (i.e., bath salts, salvias, inhalants, and other synthetics).
The psychosocial risk factors measured in this study were based on previous research
(i.e., experiences of homophobia, sensation seeking, depression, and internalized homophobia,
(Kelly, Davis, & Schlesinger, 2015; Kipke, Weiss, Ramirez, et al., 2007; Kipke, Weiss, & Wong,
2007; Mutchler et al., 2011; Operario et al., 2006; Thiede et al., 2003; Traube, Schrager,
Holloway, Weiss, & Kipke, 2013). Experiences of homophobia was assessed using the 6-item
Lifetime Experience of Homophobia questionnaire, which has previously been used with similar
samples and showed acceptable reliability (α = .69; Choi, Hudes, & Steward, 2008). The Brief
Sensation Seeking Inventory was used to assess sensation seeking (Hoyle, Stephenson,
Palmgreen, Lorch, & Donohew, 2002). In previous samples of men who have sex with men, this
instrument showed a moderate level of reliability, with a Cronbach’s alpha ranging from .63 to
.68 (Hoyle et al., 2002; Newcomb, Clerkin, & Mustanski, 2011). The 4-item Center for
Epidemiologic Studies Depression Scale was used to measure level of depression symptomology
(Melchior, Huba, Brown, & Reback, 1993). When used with a YMSM population, this scale
exhibited a high level of internal consistency (α = .83; Gibbs & Rice, 2016). To assess level of
internalized stigma toward a sexual minority identity (i.e., internalized homophobia), the 9-item
Internalized Homophobia Scale was used (Herek, Cogan, Gillis, & Glunt, 1998).
Sexual risk items focused on number of both receptive and insertive sex partners during
the previous 30 days, partner concurrency during the previous 30 days, use of protective
measures (e.g., condoms, pre-exposure prophylaxis) during last sexual encounter, substance use
before and during last sexual encounter, sexually transmitted infections (i.e., lifetime and during
SUBSTANCE USE AND TECHNOLOGY 53
the prior 6 months), and HIV testing (e.g., lifetime and during the prior year, 6 months, and 3
months).
Aim 2 addressed costs and recruitment efficiency of each method, which required
documentation of all costs: recruitment preparation (e.g., community advisory board work hours,
community advisory board incentives, research assistant hours, enumeration hours); recruitment
implementation (mileage, research assistant recruitment hours, incentives); and software and
hardware (Wang et al., 2003). To address recruitment efficiency, the study personnel
documented in each sampling period the number of potential participants who were approached,
agreed to be questioned, were found to be eligible, agreed to participate, and completed the
survey.
Analytic Procedure
Aim 1
Analysis was conducted in SPSS version 24 by first comparing the demographic
characteristics, substance use, substance use determinants, and sexual risk characteristics of the
two samples (i.e., t-tests, chi-square analyses). Second, bivariate correlations were completed to
identify relationships between substance use covariates (internalized stigma, depression
symptomology, lifetime experiences of homophobia, sensation seeking, gay community
connection) and monthly substance use (days consuming alcohol, days binge drinking, marijuana
use, and use of any other substance other than marijuana or alcohol). Significant and trending
toward significant correlations were investigated via regressions to examine the association
between each psychosocial covariate and substance use variable, while including recruitment
type as a moderator.
SUBSTANCE USE AND TECHNOLOGY 54
Aim 2
To address recruitment efficiency, recruitment periods were treated as individual
observations. Recruitment periods were grouped by method and compared using two-sample t-
tests for the number of individuals observed (unique and total number, including repeats), who
completed the eligibility survey, deemed eligible, who completed the main survey, and who
completed the supplemental survey. Additionally, the venue-based methods informing this study
used enumerations to identify locations where an estimated eight eligible individuals would be
observed (Ford et al., 2009). Therefore, to test whether these estimates were accurate,
recruitment periods from each method were compared based on whether eight or more
potentially eligible participants were observed, using two chi-square tests to identify differences.
A direct comparison of costs was conducted using previously described cost subgroups
(Wang et al., 2003). Additionally, the recruitment methods were compared based their cost per
participant (i.e., main survey completed), because a different number of participants was
recruited for each method. The findings from the cost analysis and recruitment efficiency
analysis were then combined to estimate the cost per participant for each recruitment method,
based on the final sample size. All final cost estimates are provided in 2018 U.S. dollars.
Power Analysis
The primary innovation of this dissertation was its ability to compare estimates of
substance use among samples recruited via GSNA versus venue sampling methods. Based on
prior research (Eisenberg & Wechsler, 2003; Kelly et al., 2015; Kubicek et al., 2007; McCabe,
Boyd, Hughes, & d’Arcy, 2003; Newcomb, Ryan, Greene, Garofalo, & Mustanski, 2014;
O’Connell et al., 2004; Thiede et al., 2003) and preliminary data (Gibbs & Rice, 2016; Rice et
al., 2012), average rates of substance use obtained by these methods, and the odds ratios
SUBSTANCE USE AND TECHNOLOGY 55
representing their difference, are presented in Table 3.1. Using G*Power version 3.1 (Faul,
Erdfelder, Lang, & Buchner, 2007) and assuming a logistic regression approach with 80%
power, base rates based on prior research, and the listed odds ratios for the differences, the
necessary sample size to detect statistical significance ranged from 52 to 2,975. To ensure it was
feasible to complete this study in the proposed time frame, the study team sought to recruit 94
YMSM, which theoretically would provide sufficient statistical power to investigate recruitment
methodology differences in use of marijuana, methamphetamines, and poppers. To account for
incomplete or missing data, complete case analysis with current GSNA data estimated 70%
complete cases (Gibbs & Rice, 2016; Rice et al., 2012). Therefore, a sample of 136 participants
(i.e., 95/0.70) was expected to provide complete data from at least 94 participants.

SUBSTANCE USE AND TECHNOLOGY 56
References
Choi, K.-H., Hudes, E. S., & Steward, W. T. (2008). Social discrimination, concurrent sexual
partnerships, and HIV risk among men who have sex with men in Shanghai, China. AIDS
and Behavior, 12, 71–77. doi:10.1007/s10461-008-9394-0
Delaney, K. P., Kramer, M. R., Waller, L. A., Flanders, W. D., & Sullivan, P. S. (2014). Using a
geolocation social networking application to calculate the population density of sex-
seeking gay men for research and prevention services. Journal of Medical Internet
Research, 16, e249. doi:10.2196/jmir.3523
Eisenberg, M., & Wechsler, H. (2003). Substance use behaviors among college students with
same-sex and opposite-sex experience: Results from a national study. Addictive
Behaviors, 28, 899–913. doi:10.1016/S0306-4603(01)00286-6
Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical
power analysis program for the social, behavioral, and biomedical sciences. Behavior
Research Methods, 39, 175–191. doi:10.3758/BF03193146
Fernández, M. I., Warren, J. C., Varga, L. M., Prado, G., Hernandez, N., & Bowen, G. S. (2007).
Cruising in cyber space: Comparing Internet chat room versus community venues for
recruiting Hispanic men who have sex with men to participate in prevention studies.
Journal of Ethnicity in Substance Abuse, 6, 143–162. doi:10.1300/J233v06n02_09
Ford, W. L., Weiss, G., Kipke, M. D., Ritt-Olson, A., Iverson, E., & Lopez, D. (2009). The
Healthy Young Men’s Study: Sampling methods to recruit a random cohort of young
men who have sex with men. Journal of Gay & Lesbian Social Services, 21, 357–373.
doi:10.1080/10538720802498280
Gibbs, J. J., & Rice, E. (2016). The social context of depression symptomology in sexual
SUBSTANCE USE AND TECHNOLOGY 57
minority male youth: Determinants of depression in a sample of Grindr users. Journal of
Homosexuality, 63, 278–299. doi:10.1080/00918369.2015.1083773
Herek, G. M., Cogan, J. C., Gillis, J. R., & Glunt, E. K. (1998). Correlates of internalized
homophobia in a community sample of lesbians and gay men. Journal of the Gay &
Lesbian Medical Association, 2, 17–25.
Holloway, I. W., Traube, D. E., Rice, E., Schrager, S. M., Palinkas, L. A., Richardson, J., &
Kipke, M. D. (2012). Community and individual factors associated with cigarette
smoking among young men who have sex with men. Journal of Research on
Adolescence, 22, 199–205. doi:10.1111/j.1532-7795.2011.00774.x
Hoyle, R. H., Stephenson, M. T., Palmgreen, P., Lorch, E. P., & Donohew, R. L. (2002).
Reliability and validity of a brief measure of sensation seeking. Personality and
Individual Differences, 32, 401–414. doi:10.1016/S0191-8869(01)00032-0
Kelly, J., Davis, C., & Schlesinger, C. (2015). Substance use by same sex attracted young
people: Prevalence, perceptions and homophobia. Drug and Alcohol Review, 34, 358–
365. doi:10.1111/dar.12158
Kipke, M. D., Weiss, G., Ramirez, M., Dorey, F., Ritt-Olson, A., Iverson, E., & Ford, W. (2007).
Club drug use in Los Angeles among young men who have sex with men. Substance Use
& Misuse, 42, 1723–1743. doi:10.1080/10826080701212261
Kipke, M. D., Weiss, G., & Wong, C. F. (2007). Residential status as a risk factor for drug use
and HIV risk among young men who have sex with men. AIDS and Behavior, 11(Suppl.
2), 56–69. doi:10.1007/s10461-006-9204-5
Kubicek, K., McDavitt, B., Carpineto, J., Weiss, G., Iverson, E. F., & Kipke, M. D. (2007).
Making informed decisions: How attitudes and perceptions affect the use of crystal,
SUBSTANCE USE AND TECHNOLOGY 58
cocaine, and ecstasy among young men who have sex with men. Journal of Drug Issues,
37, 643–672. doi:10.1177/002204260703700308
MacKellar, D. A., Gallagher, K. M., Finlayson, T., Sanchez, T., Lansky, A., & Sullivan, P. S.
(2007). Surveillance of HIV risk and prevention behaviors of men who have sex with
men—a national application of venue-based, time-space sampling. Public Health
Reports, 122(1, Suppl.), 39–47. doi:10.1177/00333549071220S107
McCabe, S. E., Boyd, C., Hughes, T. L., & d’Arcy, H. (2003). Sexual identity and substance use
among undergraduate students. Substance Abuse, 24, 77–91.
doi:10.1023/A:1023768215020
Melchior, L. A., Huba, G. J., Brown, V. B., & Reback, C. J. (1993). A short depression index for
women. Educational and Psychological Measurement, 53, 1117–1125.
doi:10.1177/0013164493053004024
Mutchler, M. G., McKay, T., Candelario, N., Liu, H., Stackhouse, B., Bingham, T., & Ayala, G.
(2011). Sex drugs, peer connections, and HIV: Use and risk among African-American,
Latino, and multiracial young men who have sex with men in Los Angeles and New
York. Journal of Gay & Lesbian Social Services, 23, 271–295.
doi:10.1080/10538720.2011.560100
Newcomb, M. E., Clerkin, E. M., & Mustanski, B. (2011). Sensation seeking moderates the
effects of alcohol and drug use prior to sex on sexual risk in young men who have sex
with men. AIDS and Behavior, 15, 565–575. doi:10.1007/s10461-010-9832-7
Newcomb, M. E., Ryan, D. T., Greene, G. J., Garofalo, R., & Mustanski, B. (2014). Prevalence
and patterns of smoking, alcohol use, and illicit drug use in young men who have sex
with men. Drug and Alcohol Dependence, 141, 65–71.
SUBSTANCE USE AND TECHNOLOGY 59
doi:10.1016/j.drugalcdep.2014.05.005
O’Connell, J. M., Lampinen, T. M., Weber, A. E., Chan, K., Miller, M. L., Schechter, M. T., &
Hogg, R. S. (2004). Sexual risk profile of young men in Vancouver, British Columbia,
who have sex with men and inject drugs. AIDS and Behavior, 8, 17–23.
doi:10.1023/B:AIBE.0000017522.64063.ec
Operario, D., Choi, K.-H., Chu, P. L., McFarland, W., Secura, G. M., Behel, S., … Valleroy, L.
(2006). Prevalence and correlates of substance use among young Asian Pacific Islander
men who have sex with men. Prevention Science, 7, 19–29. doi:10.1007/s11121-005-
0018-x
Rice, E., Holloway, I., Winetrobe, H., Rhoades, H., Barman-Adhikari, A., Gibbs, J., … Dunlap,
S. (2012). Sex risk among young men who have sex with men who use Grindr, a
smartphone geosocial networking application. Journal of AIDS & Clinical Research, S4,
005. doi:10.4172/2155-6113.S4-005
Thiede, H., Valleroy, L. A., MacKellar, D. A., Celentano, D. D., Ford, W. L., Hagan, H., …
Torian, L. V. (2003). Regional patterns and correlates of substance use among young
men who have sex with men in 7 US urban areas. American Journal of Public Health, 93,
1915–1921. doi:10.2105/AJPH.93.11.1915
Traube, D. E., Schrager, S. M., Holloway, I. W., Weiss, G., & Kipke, M. D. (2013).
Environmental risk, social cognition, and drug use among young men who have sex with
men: Longitudinal effects of minority status on health processes and outcomes. Drug and
Alcohol Dependence, 127, 1–7. doi:10.1016/j.drugalcdep.2012.06.007
Vanable, P. A., McKirnan, D. J., & Stokes, J. P. (1998). Identification and involvement with the
Gay Community Scale. In C. M. Davis, W. L. Yarber, R. Bauserman, G. Sheer, & S. L.
SUBSTANCE USE AND TECHNOLOGY 60
Davis (Eds.), Handbook of sexuality-related measures (pp. 407–409). Thousand Oaks,
CA: Sage.
Wang, S. J., Middleton, B., Prosser, L. A., Bardon, C. G., Spurr, C. D., Carchidi, P. J., … Bates,
D. W. (2003). A cost-benefit analysis of electronic medical records in primary care.
American Journal of Medicine, 114, 397–403. doi:10.1016/S0002-9343(03)00057-3

SUBSTANCE USE AND TECHNOLOGY 61
Figure 3.1. Geographic Sampling Frame A

SUBSTANCE USE AND TECHNOLOGY 62
Figure 3.2. Geographic Sampling Frame B (GSF A Present)

SUBSTANCE USE AND TECHNOLOGY 63
Figure 3.3. Geographic Sampling Frame C (GSF B Present)

Sources: Esri, HERE, DeLorme, USGS, Intermap, INCREMENT P, NRCan, Esri Japan, METI, Esri China (Hong Kong), Esri Korea, Esri
(Thailand), MapmyIndia, NGCC, © OpenStreetMap contributors, and the GIS User Community
SUBSTANCE USE AND TECHNOLOGY 64
Figure 3.4. Geographic Sampling Frame B with Grid of 79 Sampling Points

SUBSTANCE USE AND TECHNOLOGY 65
Figure 3.5. Interpolated User Density on Mondays at 10 a.m.

SUBSTANCE USE AND TECHNOLOGY 66
Figure 3.6. Day-Time-Sampling-Locations for Mondays at 10 a.m. in GSF B

SUBSTANCE USE AND TECHNOLOGY 67
Figure 3.7. Day-Time-Sampling-Locations for Mondays at 10 a.m. in GSF C

Sources: Esri, HERE, DeLorme, USGS, Intermap, INCREMENT P, NRCan, Esri Japan, METI, Esri China (Hong Kong), Esri Korea, Esri
(Thailand), MapmyIndia, NGCC, © OpenStreetMap contributors, and the GIS User Community
SUBSTANCE USE AND TECHNOLOGY 68
Chapter 4 (Paper 1): Using a Geosocial Networking Application for Random Sampling of
Young Men Who Have Sex with Men: Development of an Innovative Method
Introduction
Research with hidden populations (i.e., groups for whom no explicit sampling frame
exists, membership is unobservable, or membership in the group may be stigmatizing;
Heckathorn, 1997) can be daunting because there are no simple ways to facilitate study
recruitment. When an individual’s membership in a social group is not observable, how do you
locate and encourage them to participate in research? Without a defined sampling frame (i.e., list
of population individuals or locations where all population individuals are located and the best
times to find them in these locations), there are limited ways to recruit a simple random sample,
which is regarded as the gold standard for research with human subjects (Meyer & Wilson,
2009). This is the case for research on young men who have sex with men (YMSM). This barrier
to research with YMSM is especially troubling because YMSM represent a stigmatized and at-
risk population. Research on YMSM has found that YMSM have some of the highest rates of
substance use (Boyd, McCabe, & d’Arcy, 2003; Gattis, Sacco, & Cunningham-Williams, 2012;
Goldberg, Strutz, Herring, & Halpern, 2013; Hagger-Johnson et al., 2013; Hatzenbuehler,
Corbin, & Fromme, 2008; McCabe, Boyd, Hughes, & d’Arcy, 2003; McCabe, Hughes,
Bostwick, West, & Boyd, 2009) and mental health disparities (i.e., depression symptomology,
suicidal ideation, and suicide attempt; Almeida, Johnson, Corliss, Molnar, & Azrael, 2009;
Cochran, Sullivan, & Mays, 2003; Hatzenbuehler, Hilt, & Nolen-Hoeksema, 2010; Marshal et
al., 2011; Mills et al., 2004) compared to their heterosexual peers.
In response to the need for research to identify factors that contribute to these behavioral
health disparities, scholars have employed different forms of recruitment methodology to
SUBSTANCE USE AND TECHNOLOGY 69
increase accessibility of YMSM. To date, each of these recruitment methodologies has
considerably strengthened what is known about YMSM, but in this era of technological
advancement, they may offer more limitations than strengths. Therefore, the purpose of this
paper is to provide background on the strengths and limitations (e.g., sampling frame definition
and generalizability) of previously used YMSM recruitment methodologies, introduce a novel
method for recruiting a probability sample of YMSM from smartphone apps, and discuss the
strengths and limitations of implementation of this novel method.
Venue-Based Convenience Sample Recruitment
Before the creation of the internet, social research relied heavily on recruitment of
YMSM from venues and locations (e.g., bathhouses, bars, clubs, cruising locations, gay pride
festivals) where they were known to congregate (Gold, Skinner, & Ross, 1994; Kalichman,
Kelly, Morgan, & Rompa, 1997; Ridge, Plummer, & Minichiello, 1994). Even in the 21st
century, researchers have relied on venue-based convenience sampling of YMSM (Boone, Cook,
& Wilson, 2013; Dolezal, Carballo-Diéguez, Nieves-Rosa, & Dı
́ az, 2000; Halkitis & Figueroa,
2013; Kelly, Davis, & Schlesinger, 2015; Mansergh et al., 2008; Mutchler et al., 2011) because
of the ease of accessing men in the population and the low cost of recruitment method
implementation (Meyer & Wilson, 2009). This type of recruitment involves the research team
identifying locations where YMSM congregate and then approaching perceived potential
participants, inquiring about eligibility, and inviting them to participate in the study (Meyer &
Wilson, 2009). Because recruitment occurs at a single event or several time periods at known
areas of high concentration of YMSM (Boone et al., 2013; Dolezal et al., 2000; Halkitis &
Figueroa, 2013; Kelly et al., 2015; Mansergh et al., 2008; Mutchler et al., 2011), the sampling
frame is small and defined as all men at that specific location at that specific time (Pequegnat et
SUBSTANCE USE AND TECHNOLOGY 70
al., 2007). Because samples recruited at these locations are convenience samples, the
generalizability of study results to the full population of YMSM is limited (Schwarcz, Spindler,
Scheer, Valleroy, & Lansky, 2007). Venue-based convenience samples are recruited at select
locations at select times and recruited individuals tend to be more homogenous and arguably less
representative of the full population of YMSM. However, venue-based convenience sampling is
the most widely accepted recruitment methodology used to sample YMSM in urban areas,
because of the low cost of implementation and known venues with high concentrations of
YMSM (Meyer & Wilson, 2009).
Internet-Based Recruitment
With the advent of the internet, YMSM have been able to communicate across vast
distances. The internet has also allowed researchers to find and recruit YMSM to participate in
research. The two primary methods used are chat room or direct message recruitment and
clickable banner ad recruitment (Fernández et al., 2004; Fernández et al., 2007; Horvath, Bowen,
& Williams, 2006; Horvath, Rosser, & Remafedi, 2008; Hospers, Kok, Harterink, & de Zwart,
2005; Pequegnat et al., 2007; Sullivan et al., 2011). Most recently, studies have relied on YMSM
recruited through Facebook (Bauermeister et al., 2015; Grov, Rendina, Jimenez, & Parsons,
2016; Lelutiu-Weinberger et al., 2015; Lorimer, Flowers, Davis, & Frankis, 2016; Merchant et
al., 2017; Thoma, 2017). An advantage of social media recruitment is that banner ads are
targeted at individuals based on their reported interests or personal information. Accessing men
through the internet requires low resources and typically a small research team (Raymond et al.,
2010). Studies that use these methods report achieving the study sample size quickly (Meyer &
Wilson, 2009). Additionally, researchers are not limited by geographic area, and instead are able
to reach men anywhere in the world.
SUBSTANCE USE AND TECHNOLOGY 71
Despite these strengths, internet website recruitment methods have limited viability
because they typically rely on nonprobability samples (Allen, Mansergh, Mimiaga, Holman, &
Herbst, 2017; Fernández et al., 2007; Horvath, Bowen, et al., 2008; Horvath, Rosser, et al., 2008;
Hospers et al., 2005).

Although online website recruitment tends to require less resources and
time than venue-based recruitment, there are significant disadvantages to recruiting a
convenience sample that lacks control related to users’ physical geolocation (i.e., a website user
could be located anywhere in the world; Pequegnat et al., 2007). This is especially true in the
case of targeted banner ads, which may rely on the reported location rather than explicit location.
Many websites also lack the ability to limit individuals to one profile, which makes probability-
based recruitment difficult (Pequegnat et al., 2007). Because individuals may create as many
profiles as they like, there is risk of individuals being double or tripled enrolled in a study
(Pequegnat et al., 2007). This means that the sampling frame of these studies is typically larger
than venue-based studies (e.g., every 18- to 24-year-old man on Facebook) and difficult to define
(Pequegnat et al., 2007). Further, a reliance on self-selection for the study (i.e., clicking on a
banner ad) and self-reported data informing targeted recruitment (e.g., men who report “interest”
in other men) make the findings from these YMSM studies limited in generalizability.
Respondent-Driven Sampling
Respondent-driven sampling (RDS) is a recruitment methodology intended for use with
hidden populations that have strong intergroup network connections. As a more rigorous
snowball method, it relies on seeds (i.e., initially recruited participants) and their network
connections to recruit chains of other eligible participants (Malekinejad et al., 2008; White et al.,
2012). An RDS perspective assumes that there is no sampling frame for a hidden population,
individuals can be recruited because they are socially connected, and these social networks allow
SUBSTANCE USE AND TECHNOLOGY 72
access to the population given no other feasible alternatives (Heckathorn, 1997; Léon, Des
Jarlais, Jauffret-Roustide, & Le Strat, 2016). RDS begins with the recruitment of participant
seeds, who then recruit individuals in their social network to participate in the study
(Heckathorn, 1997). RDS then uses advanced statistical modeling to weight participant data
based on location in the recruited social network (Heckathorn, 1997). RDS has been
implemented successfully with a variety of hard-to-reach populations, including homeless youth
(Gwadz et al., 2010), injection drug users (Frost et al., 2006, Heckathorn, Semaan, Broadhead, &
Hughes, 2002; Léon et al., 2016), and HIV-positive men who have sex with men (MSM;
Charurat et al., 2015; Lauby et al., 2008).
RDS has been promoted as an option for recruitment of general populations of YMSM,
but recent literature reported that this procedure may not be as promising. Shoptaw et al. (2009),
in their study on MSM in Los Angeles, found that using RDS was not an appropriate recruitment
strategy because it led to homogenous recruitment chains of men who were mostly men of color,
low income, and unemployed. Homogeny of socioeconomic status has been found in other
YMSM studies using RDS (Kuhns et al., 2015). Because RDS relies on the germination of initial
seeds into recruitment chains, a study with MSM indicated that germinating seeds are more
likely to report unprotected anal intercourse compared to nongerminating seeds and that men
who use illicit drugs are more likely to actively recruit for the study (Reisner et al., 2010). This
means that in addition to recruiting demographically homogenous samples, RDS procedures may
recruit individuals with higher risk profiles and may systematically exclude YMSM who report
less risk. Additionally, the methodological assumption that YMSM are socially connected
systematically excludes YMSM who may not identify at a sexual minority, do not disclose that
they have sex with men, or have low interest in being connected to other YMSM, therefore
SUBSTANCE USE AND TECHNOLOGY 73
limiting their social connections to other YMSM. Because of the reliance on chains of
recruitment, RDS studies of YMSM report that an extended period (e.g., up to 3 years; Kuhns et
al., 2015) is needed to reach the full study sample size (Kuhns et al., 2015; Lachowsky et al.,
2016). Although RDS may allow access to high-risk individuals who are YMSM, as a sampling
method for recruiting a generalizable sample of YMSM in a timely manner, it presents serious
limitations.
Venue-Based Probability Sample Recruitment
Although many studies using venue-based sampling with YMSM have relied on
convenience samples (Boone et al., 2013; Dolezal et al., 2000; Halkitis & Figueroa, 2013; Kelly
et al., 2015; Mansergh et al., 2008; Mutchler et al., 2011), more rigorous procedures have been
developed that rely on a preliminary enumeration (i.e., counting of potential participants) stage
for identifying both times and locations of YMSM congregation (Kipke, Weiss, Ramirez, et al.,
2007; Kipke, Weiss, & Wong, 2007; Operario et al., 2006; Thiede et al., 2003; Traube, Schrager,
Holloway, Weiss, & Kipke, 2013). Venue-based stratified probability sampling helps ensure that
participants are recruited based on probability rather than convenience (Ford et al., 2009;
MacKellar et al., 2007). These methods begin by identifying all the venues where YMSM can be
found at each period of the day (morning, afternoon, evening, and night; Ford et al., 2009). In
effect, a full study sampling frame is defined, wherein every possible venue (e.g., gay bars,
clubs) and period for recruitment is identified in a geographic area and each venue-day-time
period has the same probability of being selected for clustered recruitment (Ford et al., 2009).
Therefore, the findings from these studies are generalizable to all YMSM who attend venues in
the geographic area of interest.
SUBSTANCE USE AND TECHNOLOGY 74
Despite this enhancement, these recruitment strategies have been criticized for excluding
YMSM who either do not identify as gay or bisexual but engage in same-sex behaviors or have a
limited connection to the gay community (Meyer & Wilson, 2009). Additionally, studies
examining the psychosocial determinants of substance use for YMSM have relied on samples
recruited from gay venues that arguably condone substance use (e.g., gay bars, local gay events;
Kelly et al., 2015; Kipke, Weiss, Ramirez, et al., 2007; Kipke, Weiss, & Wong, 2007; Mutchler
et al., 2011; Operario et al., 2006; Thiede et al., 2003; Traube et al., 2013). This sampling bias is
a critical concern in substance abuse research because it may artificially inflate the reporting of
substance use prevalence in this community. Further, venue-based probability sampling also
requires considerable resources to implement (MacKellar et al., 2007; Meyer & Wilson, 2009).
Introduction of Geosocial Networking Smartphone Apps
Emerging literature has suggested that YMSM are increasingly using technology to meet
other YMSM and relying less on venues to provide social interaction (Grov, 2012; Grov &
Crow, 2012; Zablotska, Holt, & Prestage, 2012). In 2009, geosocial networking apps (GSNAs)
were introduced to the Apple smartphone app market, with the creation of Grindr (Leslie, 2009).
During the past 9 years, these apps have grown in popularity among YMSM, with some apps
boasting 25 million users worldwide (Hornet, 2018a). On most GSNAs, users see a home screen
with a matrix of pictures corresponding with the individual users closest to their geographic
location. Users can then message these individuals, send pictures, exchange other contact
information, or send their exact map location using the integrated message system. YMSM can
browse the profiles of other YMSM in the same geographic region, and because profiles are
ordered based on their relative proximity to the user, YMSM users may click on a profile and see
the distance of other users from their current location. Portions of the YMSM community that
SUBSTANCE USE AND TECHNOLOGY 75
previously attended gay venues specifically to meet other men (and not to use alcohol or other
substances) may no longer use those venues to facilitate connections.
Given this cultural shift and advancement in technology, YMSM researchers have begun
to look to GSNAs as an innovative technology for recruitment. Rice et al. (2012) developed
methods that involved probability-based methods, which by using the GSNA integrated
messenger, randomly approached MSM to participate in a research study. Although these
methods were ultimately successful at recruiting men to participate in social research, changes in
GSNA user agreements have made direct messaging between researchers and users no longer
possible (Grindr, 2017; Hornet, 2018c; Lucid Dreams, 2014a, 2014b; Perry Street Software,
2017).
Since then, studies have recruited convenience samples of MSM from GSNAs. By
partnering with a GSNA, researchers have been successful at purchasing ads to be displayed as
pop-up messages that cover the entire smartphone screen and appear when the user first opens
the application or as a banner ad that appears at the top or bottom of the screen during app use
(Buckingham et al., 2017; Goedel & Duncan, 2016; Goedel, Hagan, et al., 2017; Goedel, Mayer,
Mimiaga, & Duncan, 2017; Goedel, Safren, Mayer, & Duncan, 2017; Grov et al., 2016; Lorimer
et al., 2016; Phillips, Grov, & Mustanski, 2015; Siegler et al., 2015). These methods allow
researchers to quickly connect with large samples of MSM (Buckingham et al., 2017). Because
individuals are required to use their smartphone geolocation, researchers can specify locations
for the display of ads (Buckingham et al., 2017; Goedel, Hagan, et al., 2017; Goedel, Mayer, et
al., 2017; Goedel, Safren, et al., 2017; Grov et al., 2016; Lorimer et al., 2016; Merchant et al.,
2017; Phillips et al., 2015; Siegler et al., 2015).
SUBSTANCE USE AND TECHNOLOGY 76
The sampling frame used in these studies is large and loosely defined, because there is a
general geographic area for ad broadcasts, but the time periods for recruitment are largely
arbitrary or hypothesized as time periods of heavy use (Buckingham et al., 2017; Goedel, Hagan,
et al., 2017; Goedel, Mayer, et al., 2017; Goedel, Safren, et al., 2017; Grov et al., 2016; Lorimer
et al., 2016; Merchant et al., 2017; Phillips et al., 2015; Siegler et al., 2015). Although it is
estimated that about two thirds of MSM use GSNAs (Phillips et al., 2014), because studies using
GSNA recruited convenience samples, the generalizability of the study results is limited.
Currently, venue-based probability sampling is the most feasible probability-based
methodology to sample YMSM because it theoretically builds a full sampling frame, then uses
probability to recruit clusters of men (Meyer & Wilson, 2009). However, recent literature has
indicated that the geographic data embedded in GSNAs can provide information to define a
sampling frame. Delaney, Kramer, Waller, Flanders, and Sullivan (2014) developed a method for
estimating the density of GSNA users in Atlanta, Georgia, using embedded user data. Although
researchers have hypothesized the ability to use these data to inform locations for recruitment, to
date, these methods have not been fully developed nor implemented (Delaney et al., 2014;
Goedel, Brooks, & Duncan, 2016).
Given the impact of technological advancements on how YMSM can now be recruited,
there is a need to investigate the use of GSNAs for probability-based recruitment. Therefore, the
aims of this paper are to: (a) present a tested model for GSNA-based probability sampling, and
(b) discuss the barriers and strengths encountered in the development of these methods.
Methods
Two YMSM samples (one from an app and the other from venues) were recruited
concurrently in Los Angeles, California, from April 2017 through April 2018. The process
SUBSTANCE USE AND TECHNOLOGY 77
outlined here was modeled after the venue-based stratified probability sampling procedures
presented by researchers of the Healthy Young Men’s Study in Los Angeles (Ford et al., 2009).
The GSNA methods were informed by a previous study by this dissertation’s author and
colleagues, also conducted in Los Angeles (Rice et al., 2012). Geographic information systems
methods were also used to build the GSNA sampling frame and were informed by a 2014 study
conducted in Atlanta (Delaney et al., 2014). In this chapter, the venue-based procedures are not
be presented. However, consistent with the stages of venue-based probability sampling (Ford et
al., 2009), the GSNA-based probability sampling methods followed three stages: (1) building of
the sampling frame, (2) enumeration of the study sampling frame, and (3) recruitment of
participants using a monthly sampling calendar. Additionally, two options for determining
locations for sampling are presented: (a) systematic collection and analysis in ArcMap for areas
smaller than 100 square kilometers and (b) a geographic coverage maximization method for
areas larger than 100 square kilometers.
Stage 1: Building a Sampling Frame
The process of determining the sampling frame involved identification of the study
GSNA and then the study geographic sampling area. A community advisory board consisting of
18- to 24-year-old MSM who live in Los Angeles was convened. The board was asked to report
on the most popular GSNA used by YMSM in the geographic sampling frame (GSF). Each
proposed GSNA needed to meet three criteria to be included: (a) ability to filter users by age, (b)
ordering of user profiles by relative proximity to the study smart device, and (c) reporting of the
relative distance of users from the study smart device. The top three GSNAs were contacted by
the principal investigator to inquire about interest in collaborating in the current study. From
SUBSTANCE USE AND TECHNOLOGY 78
these conversations, the study GSNA was selected (i.e., Hornet). Board participants received a
$30 gift card incentive.
The GSF in which recruitment was conducted went through two evolutions. Initially,
GSF A was defined by the boundaries of zip code 90036, a portion of West Hollywood,
California (see Figure 4.1). Upon initial pilot tests of the GSNA-based probability sampling
methods and discussions with the study GSNA (i.e., Hornet), it was determined that the area
covered by GSF A lacked the GSNA user diversity to meet the projected sample size. Therefore,
the area of the study GSF was increased by almost 4 times to create GSF B, which is defined by
census bureau tracts and encompasses a portion of West Hollywood, Hollywood, and Koreatown
(see Figure 4.2). Recruitment for the venue-based sample was completed in GSF B while
methods for recruitment with the GSNA were being pilot tested. Once the methods for GSNA
recruitment were established, it was determined through consultations with the health strategist
of the study GSNA that the GSF should be expanded to maximize number of unique users
reached. As such, the GSF for GSNA-based sample was expanded to the southern portion of Los
Angeles County, with a northern boundary of Burbank, southern boundary of Long Beach,
western boundary of Santa Monica, and eastern boundary of Pomona (see Figure 4.3).
Stage 2: Enumeration of the Sampling Frame
Enumeration as it is conceived in venue-based stratified probability sampling involves
estimating the number of YMSM at all venues and times in the GSF (Ford et al., 2009). In the
same way, enumeration in GSNA-based probability sampling involves identifying where app
users are located in the GSF and how this density is influenced by different periods and days of
the week. Because the GSF of the study evolved during pilot testing, Stage 2 was conducted
using two methods. Option A was implemented in GSF A and GSF B because the size of the
SUBSTANCE USE AND TECHNOLOGY 79
geographic area allowed these procedures, whereas Option B was implemented in GSF C due to
the size of the geographic area.
Option A: Systematic collection and analysis in ArcMap. These following procedures
were used for GSF A and GSF B and informed by procedures that were used to collect user
density data in Atlanta by Delaney and colleagues (2014). Although both GSF A and GSF B
used these procedures, for the purposes of being concise, GSF B is used as the example. Using
the study GSNA, anonymous geospatial observation data were collected from the application
interface. For GSF B, a density map was created by observing the user density data of an 11-by-8
grid of 1-kilometer-separated points (9 points were unnecessary due to the nonuniform nature of
the eastern boundary; see Figure 4.4), constituting 79 points. Observing density data was
accomplished by changing the geographic positioning system (GPS) information supplied to the
study smart device (using a third-party application, “Fake GPS Location”), therefore allowing
the study team to stay in a fixed location. This procedure allows for rapid user density
observation of the study sampling area, which minimizes any potential change in the user density
during the period required to observe all points in the GSF.
After opening the study GSNA, a study team member filtered the visible profiles by
reported age (i.e., 18 to 24). Second, because each profile lists the associated geographic distance
from the study smart device, the study team member scrolled through the listed user profiles and
noted the number of users (k) in a specific distance (j, i.e., 1 mile) of the study smart device.
Geographic user density data was determined by calculating the area of a circle with a radius j,
and then dividing k by that area (density = k/[πj
2
]). For example, if 20 users’ profiles listed their
distance from the study smart device as 1 mile or less, the average user density at that geographic
point is 6.37 users per square mile (i.e., 20/[π1
2
] = 23.87).
SUBSTANCE USE AND TECHNOLOGY 80
The observation of user density data was completed for each sampling-day-time period
(SDTP; four periods × 7 days = 28) for all points of GSF B. The periods used were the midpoints
of each of the periods presented by Ford et al. (2009; i.e., 8 a.m.–12 p.m., 12 p.m.–4 p.m., 4
p.m.–8 p.m., and 8 p.m.–12 a.m.): 10 a.m., 2 p.m., 6 p.m., and 10 p.m. Two items of data were
collected at each geographic grid point: (a) a GPS point and (b) an associated user density value.
This procedure was completed at each grid point 3 times for every SDTP in a 1-month period.
The user density values were then averaged across the three collection times for each grid point
of each SDTP. For example, in GSF B, the user density was collected for all 79 points on
Monday at 10 a.m. on three Mondays, and then averaged across each of the grid points. This
means that Point 35 of 79 on Mondays at 10 a.m. had a density of 6.68, 7.00, and 5.41 on three
Mondays. This density was then averaged (6.37 users per square mile) and converted to square
kilometers to approximate the expected density at that specific point on any Monday at 10 a.m.
In total, 6,636 points of data (i.e., 28 SDTP × 79 points × three observations) were collected to
estimate the user density on every day and during every period in GSF B. Manually collecting 84
sets (28 SDTP × three collection times) of user density data from 79 points took an average of
22.96 minutes (SD = 2.01 minutes) per observation period.
The user density data from GSF B was then imported into ArcMap 10.5. The data file
contained 79 observations (each grid point) with two types of data: (a) a GPS datum converted to
x–y coordinates and (b) a corresponding user density for each SDTP. The data were visualized
for each SDTP using x–y coordinates and converted into an interpolated layer titled
raster_SDTP_1 (i.e., a map layer that averaged the density between data points to create a full
GSF density layer), with a cell size of 50 square meters (see Figure 4.5). For the purposes of
clarity, the SDTP of Monday at 10 a.m. is used to explain the methods used to analyze the user
SUBSTANCE USE AND TECHNOLOGY 81
density data. Using the math algebra function, raster_M10_1 was divided by the number of 50
square meters in 1 square kilometer (i.e., 400), making raster_M10_2. This raster was created to
approximate the number of users at any 50-square-meter section.
Circular focal statistics analyses were then used to calculate the radial distance j required
to capture k users (i.e., 8), referred to as summation of neighborhoods. Starting with the shortest
radial distance j (i.e., 1 km) and increasing distance j (e.g., 1.6 km, 2 km, 3 km, 3.21 km, 4 km,
4.83 km, 5 km) with each new focal statistics, new rasters (i.e., raster_M10 _3.j; e.g.,
raster_M10 _3.1, raster_M10 _3.1.6, raster_M10 _3.2, raster_M10 _3.3, raster_M10 _3.3.21,
etc.) were created until a final raster encompassed GSF B (see Figure 4.6). Each raster indicates
that at any one point in the colored region, the study team would be able to observe at least k
(i.e., 8) users in the j radial distance. Using the reclassify function, each raster was recoded to
make all 1s become 0s, to create raster_M10 _4j. Using the math algebra function, raster_3j+1
was then multiplied by raster_4j. For example, raster_3.2 overlaps some of the same area as
raster_3.1.6, because raster_3.2 contains the geographic area wherein the study team could
observe eight or more users in a 2 km radial distance, which includes areas within 1.6 km. By
multiplying raster_4.1.6 (the recoded raster_3.1.6) by raster_3.2, the new raster (i.e.,
raster_5.2) no longer contained the area in raster_3.1.6 (see Figure 4.7). In the ArcMap window,
each raster_M10_5j was brought to the front of the map and given unique colors. These different
colored sections corresponded with the radial distance required to, on average, observe eight
users on Monday at 10 a.m. Sampling points were determined based on their circular buffer size
(i.e., the distance j it takes to capture k users). The number of sampling points needed for each
SDTP for GSF B was determined by three criteria: point-based circular buffers must (a)
maximize coverage of GSF B, (b) minimize buffer overlap, and (c) minimize coverage of areas
SUBSTANCE USE AND TECHNOLOGY 82
outside the GSF B boundary. Points were placed on the map using trial and error. The editor tool
allowed points to be created and placed on the map in locations to adhere to the previously stated
criteria. Once placed in a raster_SDTP_5j area, the point feature’s data table was opened and a
new variable was created called “distance.” The corresponding distance j of raster_SDTP_5j
where the point was located was then added to the data column for distance. Once the necessary
number of points (adhering to the three criteria) were placed in the map, the buffer tool was used
to visualize the distance achieved by each point. This process continued until the three criteria
were met (see Figure 4.8). The center of each circle in Figure 4.8 represents the GPS coordinates
of the five unique day-time-sampling-locations (DTSLs) for recruitment of YMSM. Each
SDTP’s user density data were processed using the same procedures in ArcMap 10.5. Figure 4.9
summarizes the analysis of user density data in ArcMap 10.5. Once completed, 140 unique
DTSLs had been determined. Each DTSL had a corresponding day and time for recruitment,
GPS point, and buffer distance for recruiting.
Use of Option A procedures means that on average, each DTSL will reach a specified
number of users.
Option B: Geographic coverage maximization. Upon piloting recruitment in GSF B,
the GSF was expanded to GSF C due to a lack of user diversity (i.e., limited unique users). GSF
C covered an area approximately 50 times larger than GSF B. Because of the considerable area,
the process for collecting user density data through a 1-kilometer-spaced grid could not be
achieved in a timely manner (approximately 20 hours). Therefore, five points were chosen as the
locations for recruitment (see Figure 4.10). Similar to the process of placing sampling points in
ArcMap, these five points were chosen using the three criteria, but with a standard buffer area of
8 miles. Once completed, 140 DTSL were identified in GSF C.
SUBSTANCE USE AND TECHNOLOGY 83
Stage 3: Recruitment and Data Collection
Pilot testing of GSNA-based recruitment occurred in GSF B. After pilot testing, it was
determined that recruitment of the GSNA sample required GSF C. A monthly sampling calendar
was used to randomly select both venue-day-time sampling events and DTSLs. Creation of
calendars followed the procedures used in Ford et al. (2009), except for two adaptations:
exclusion of times of large events taking place in the GSF and procedures to limit one
recruitment strategy per sampling period. At the beginning of each recruitment month and before
the monthly sampling calendar was created, the study team researched YMSM events and
holidays in the GSF (referred to as wild cards; Ford et al., 2009) and recorded those that were
likely to be large and therefore change the density of YMSM in the GSF (e.g., lesbian, gay,
bisexual, and transgender pride events). These events were placed in the sampling calendar to
inhibit recruitment during the associated period.
To create the monthly sampling calendar, all venue-day-time sampling events and DTSLs
were loaded into a spreadsheet. Recruitment events were then categorized as weekday day,
weekday night, weekend day, and weekend night to allow for matching between recruitment
methods, because it was hypothesized that different individuals would be located in the GSF at
each of these times. One randomly selected event was placed in the appropriate day and time on
the calendar and a matched random selection was taken from the other recruitment method. If an
event was randomly selected and the corresponding day and time was not available, an additional
random selection was taken from that same recruitment strategy and so on, until an appropriate
selection event was randomly selected. Selection of DTSLs continued until three or four were
placed on the study calendar. During pilot tests in GSF B, 10 DTSLs were used, but the study
GSNA identified that users were being overburdened with messages, so the number of DTSLs
SUBSTANCE USE AND TECHNOLOGY 84
was reduced to three or four per month in GSF C. Once the targeted sample size was achieved
for a recruitment strategy, the corresponding remaining sampling events for that month were
disregarded.
Recruitment for the GSNA sample followed similar procedures outlined by Rice et al.
(2012), except for the use of DTSLs, the monthly sampling calendar for determining DTSLs, and
direct messages sent from the study GSNA company rather than from the study team. Instead of
visiting the specified geographic point, the study GSNA was technologically misled and
provided with GPS coordinates for the DTSL specified on the monthly sampling calendar. As
noted, each DTSL had an associated GPS point and radial distance j (e.g., 8 miles). The
recruitment process began with one study team member opening the GSNA on the study smart
device and making an anonymous user profile. After filtering the visible user profiles by age
eligibility (i.e., 18–24 years old), the study team member recorded all public GSNA user IDs
within 8 miles of the study smart device. Concurrently, the study GSNA (i.e., Hornet) sent out a
preset direct message broadcast in the GSNA to each GSNA user within 8 miles of the DTSL.
The direct message appeared as a personal message from Hornet and remained in the users’
message inbox. The message explained the USC study and incentives for participation, and
provided a link to the GSNA eligibility survey (e.g., “Hey, you appear to be eligible for a USC
study! Do you have 2 minutes to answer a few questions? If you’re eligible and you take the
survey, you can earn up to $50”). Users could then click on a link to the study eligibility survey
and complete the eligibility items. At the end of the eligibility survey, the potential participant
created a unique study ID code using a list of questions about themselves (e.g., provide the last
two letters of your first name, the two digits of your birth month, and the last two digits of your
phone number) and input his contact information (i.e., email address, mobile phone number).
SUBSTANCE USE AND TECHNOLOGY 85
The eligibility data was received by the contact information manager, who determined if
individuals were eligible to take the study survey (i.e., met the study eligibility criteria and had a
unique study ID code indicating they had not previously taken the main study survey). The
contact information manager then provided a list of new eligible unique study IDs to the data
manager, who never had access to the contact information of the participants. The data manager
used the unique study IDs to create unique one-use web links to the main study survey and
provided the contact information manager with these links. The contact information manager
then messaged the potential participant using his preferred mode of contact with instructions to
take the main study survey, his unique study ID, and the unique link to take the main study
survey. The contact information manager never had access to the main study survey data.
Once entering the main study survey website, potential participants were given the
opportunity to read informed consent information. After giving informed consent, participants
were instructed to enter their unique study ID code. The main study survey asked participants to
report on demographic information, substance use characteristics, substance use determinants,
and sexual risk behaviors. Once completing the main study survey, participants were asked to
indicate if they had previously completed the study. This item was asked as a last safeguard
against duplicate data. Once a participant completed the main study survey, the submitted data
were available to the data manager. The data manager provided a list of unique study ID codes
for individuals who completed the main study survey and an associated supplemental survey
unique link to the contact information manager. The contact information manager then sent each
participant a $35 downloadable gift card. In addition, the contact information manager sent each
participant who completed the main study survey a link to the supplemental survey with relevant
SUBSTANCE USE AND TECHNOLOGY 86
instructions. The supplemental survey followed the same procedures as the main study survey,
except that the participant was incentivized with a $15 downloadable gift card.
Barriers and Strengths of GSNA-Based Probability Sampling
Because GSNA-based probability sampling methodology may continue to be developed
and refined, a priority of this chapter was to provide other researchers with knowledge of the
barriers encountered in this study. During the development and implementation of GSNA-based
probability sampling methodologies, two challenges were encountered that forced
reconsideration of the originally conceived procedures and resulted in a strengthened
methodology. These two challenges were navigation of the GSNA partnership and determination
of the geographic sampling area.
Navigation of the GSNA Partnership
When the methods for this study were originally conceived, researchers and advertisers
had the ability to send direct messages to users for study recruitment or to advertise a product.
What was originally intended to provide MSM with a digital environment bounded by
geographic proximity could have easily become an environment overburdened with advertising
of research, services, and products. Because one of the top priorities of a GSNA company is the
user experience, GSNA companies in 2015 decided to update their user agreements to restrict
advertising in the application to messages sent from the GSNA company (Grindr, 2017; Hornet,
2018c; Lucid Dreams, 2014a, 2014b; Perry Street Software, 2017). Therefore, a partnership was
developed with the study GSNA, Hornet. One of the first decisions to be made was the type of
study broadcast advertisement users should receive. Hornet has three types of advertising
options: pop-up ads, banner ads, and direct messages (Hornet, 2018b). Previous literature used
pop-up and banner ads, but what these studies neglected to report is that pop-up and banner
SUBSTANCE USE AND TECHNOLOGY 87
advertisements are only sent to users who do not use the paid service (Buckingham et al., 2017;
Goedel & Duncan, 2016; Goedel, Hagen, et al., 2017; Goedel, Mayer, et al., 2017; Goedel,
Safren, et al., 2017; Grov et al., 2016; Lorimer et al., 2016; Phillips et al., 2015; Siegler et al.,
2015). Most gay GSNAs provide the app service for free with advertisements or a paid
subscription without advertisements. Systematic exclusion of subscribers limits the
generalizability of study results, and therefore the direct message option was chosen because this
procedure also most closely imitates the methods used by this dissertation’s author and
colleagues (Rice et al., 2012). Use of direct messages and inclusion of all Hornet users are a
strength of this study, because these procedures allowed the sampling frame to be defined as all
YMSM who use Hornet in Los Angeles County.
Because all users, including paid users, received the study advertisement, frequency of
recruitment broadcasts became a negotiation. The original procedures modeled after venue-based
probability sampling (Ford et al., 2009) involved 10 recruitment events per month. These
procedures were especially problematic for the GSNA in GSF B, due to concern that users would
receive the broadcast multiple times a month. Even GSF C, the southern portion of Los Angeles
County, still contained an area wherein repeated broadcasts in the same five areas could mean
users would receive the prompt more than once in a month. Therefore, to ease concern and
reduce message burden on the GSNA users, the number of recruitment events was reduced to
three or four in a month. Initial estimates indicated that up to 2 months and 20 recruitment events
would be required to reach the full study sample size. However, this reduction in number of
monthly sampling events meant that the timeline for recruitment was extended. In a larger GSF
(e.g., a state or full country), this would not be an issue, because the likelihood of areas receiving
a repeated broadcast would be reduced. These alterations of the original procedures allowed a
SUBSTANCE USE AND TECHNOLOGY 88
collaborative partnership to develop between the study team and the study GSNA, Hornet,
ultimately a meaningful strength of this research. Community-based participatory research
prioritizes the development of partnerships with community organizations (Rhodes, Malow, &
Jolly, 2010). Although GSNAs may not typically be regarded as community organizations due to
their for-profit business structure, these technologies have become the digital social environment
wherein YMSM interact, and many of these organizations have a commitment to gay men’s
health (e.g., Grindr for Equality, Hornet’s Know Your Status; Grindr, 2018; Hornet, 2018d).
Determination of the Geographic Sampling Frame
The most challenging barrier to implementing the methods of this study was determining
the appropriate size of the GSF. The two enlargements of the GSF during the 7-month pilot
period were a direct result of concern about the number of unique users reached by the
recruitment broadcasts. Essentially, determining the size of and methods for user density
enumeration in the GSF was a process of critically examining current knowledge of the
behaviors of GSNA users and a process of mathematical analyses.
GSF A is an area in West Hollywood that is well known for having the highest
concentration of gay venues. Therefore, it was theorized that when individuals who use the study
GSNA come to the West Hollywood area, they would simultaneously use the application. The
highest concentration of users would be expected in this area. This theorized idea was based on
two assumptions about YMSM and their GSNA use: (a) GSNA users use these apps when they
attend gay venues, and (b) GSNA users attend gay venues. Previous research indicated that
GSNA users use this technology for multiple reasons, including to “kill time,” “make new
friends,” and “connect with the gay community,” and at multiple times throughout the week
(Holloway et al., 2014; Phillips et al., 2014; Rice et al., 2012). The assumption that GSNA users
SUBSTANCE USE AND TECHNOLOGY 89
use this technology while in gay venues was not indicated. In fact, qualitative research indicated
that individuals purposely use GSNAs to connect with other YMSM when they are not in areas
known for having a high concentration of gay venues (Tudor, 2012). Additionally, there is little
to no known research on whether GSNA users also attend gay venues. One study, which
recruited men through RDS procedures, found that 27% of online-recruited MSM versus 6% of
in-person-recruited MSM reported that it is personally “not important” to be connected to and
involved in the gay community (Lachowsky et al., 2016). Although the importance of being
connected to the gay community may only be an indicator of venue attendance, it does suggest
that involvement in what is perceived as gay activities may be lower in men who use the internet
to facilitate connections. Following this logic, the GSF was enlarged by 4 times from GSF A to
GSF B, to encompass a larger area of Los Angeles.
The enumeration and recruitment stages were both piloted in GSF B. Option A, the
systematic collection and analysis of data in ArcMap, was implemented. This method estimated
that an average of eight users would be reached by each recruitment event. Eight users were
chosen as the target number of individuals approached to allow for GSNA recruitment events to
be comparable to the venue-based recruitment methods that were concurrently implemented
(Ford et al., 2009). During the 26 piloted recruitment events during 5 months, the eligibility
survey was completed 44 times (by users of all ages), nine individuals were found to be eligible,
and six completed the survey. Observation data identified that that during 26 recruitment events,
184 unique users (with a reported age between 18 and 24) received the recruitment broadcast.
The mean number of unique users at each event was 7.077 (SD = 4.127), and a one-sample t-test
indicated (t[25] = -1.140, p = .270) the average number of unique users did not significantly vary
from the expected eight users. Although the Option A data collection and analysis procedures
SUBSTANCE USE AND TECHNOLOGY 90
appeared on average to correctly estimate the number of users in each area, the number of 18- to
24-year-old users completing the eligibility survey was rather limited. A systematic review of
studies using GSNA to recruit MSM found that the recruitment-to-survey-completion ratio of
seven studies ranged from 100:2 to 100:74 (Zou & Fan, 2017). Although Option A methods used
in GSF B indicated a ratio comparable to other studies of 100:3, there were concerns about how
many more unique users could be reached in the same geographic area. Discussions between the
GSNA health strategist and this dissertation’s author prompted the enlargement of the GSF to
GSF C. With this enlargement and the use of Option B, the five recruitment locations reached
different numbers of users at each DTSL. Although the number of users was different at each
location, this method still adhered to the assumption that all users in the GSF had the same
probability of being recruited. Further, Option B compared to Option A more closely aligned
with the procedures of venue-based probability sampling (Ford et al., 2009), because this
methodology stated that a minimum of eight potential participants should be accessible at each
venue-day-time recruitment event. In fact, in this study, potential participant estimates per
venue-day-time event ranged from eight to 174 in a 4-hour period, and DTSL estimates for
Option B ranged from nine to 141. Therefore, by adapting the strengths (i.e., building a full
sampling frame, allowing all members of the sampling frame to have the same probability of
being recruited) of venue-based probability sampling to a digital environment, GSNA-based
probability sampling methods strengthened the generalizability of a GSNA-based recruitment
method.
Conclusions
One of the primary aims of this dissertation is to support the evolution and adaptation of
rigorous methodology to fit a digital format. Based on the outlined processes, GSNA-based
SUBSTANCE USE AND TECHNOLOGY 91
probability sampling with user density enumeration procedures can be developed and
implemented. Whereas previous research with GSNA has neglected to consider building a full
sampling frame and therefore recruited using convenience (Buckingham et al., 2017; Goedel &
Duncan, 2016; Goedel, Hagen, et al., 2017; Goedel, Mayer, et al., 2017; Goedel, Safren, et al.,
2017; Grov et al., 2016; Lorimer et al., 2016; Phillips et al., 2015; Siegler et al., 2015), the
procedures developed in this study defined a full sampling frame and made probability sampling
accessible through GSNAs.
Use of GSNA-based probability sampling procedures has several meaningful
implications for YMSM research. First, it allows a theoretically more generalizable sample of
men to be recruited from apps, which strengthens what is known about this population. Although
GSNAs require the use of a smartphone, it is estimated that 94% of 18- to 29-year-olds in the
United States have access to this technology (Pew Research Center, 2018). Additionally, GSNAs
are estimated to be used by two thirds of all MSM (Phillips et al., 2014). These procedures also
lay the groundwork for recruitment of YMSM who live outside urban centers and therefore do
not have access to traditional gay venues. Moreover, because these methods rely on a
geographically bound digital environment rather than a physical (i.e., venues) or geographically
amorphous digital (i.e., websites) environment, targeted recruitment can occur digitally
anywhere in the world (without the need of venues) while limiting potential participants to the
targeted geographic region.
There are several study limitations to note. The sampling frame built in this study was
limited to 18- to 24-year-old Hornet users in the southern portion of Los Angeles County.
Although research has indicated that YMSM are increasingly using GSNAs to find one another
(Grov, 2012; Grov & Crow, 2012; Zablotska et al., 2012), these men spread their use among
SUBSTANCE USE AND TECHNOLOGY 92
multiple apps. By using only one GSNA in recruitment, the current study excluded YMSM who
use other apps. Exclusion of other apps was purposeful, because recent literature indicated that
YMSM report using at least three GSNA concurrently (Goedel & Duncan, 2015). Inclusion of a
second or third GSNA would increase the risk of double enrollment in the study.
The methods tested in this study were implemented in a large urban area, and therefore
very little is known about the application of these methods in rural areas. It is unknown whether
the user density enumeration would be as accurate or achievable in areas of lower population
density. However, due to the limited accessibility of gay venues in rural areas, recruitment
through GSNAs may offer opportunities to include marginalized rural MSM in exploratory and
intervention research.
Future research should also consider the application of these methods at the state and
national levels for recruiting both younger and older MSM. Further, the methods used for
enumerating user density are applicable outside of study recruitment. Because user density data
are geographically bound, they can very easily be linked with state and national datasets (e.g.,
census data) to explore the impact of system-level factors on where MSM reside. To this effect,
one recent study conducted in Vietnam used a GSNA for sexual minority men to count the
number of users who reside in Ho Chi Minh City (Safarnejad, Nga, & Son, 2017). These
calculations were used with survey-level RDS data to estimate the population size of MSM in Ho
Chi Minh City (Safarnejad et al., 2017). The user density procedures developed in the current
study can strengthen the innovative population estimates used in studies like the one conducted
in Vietnam by providing data on where men are located rather than simply a count. This
information can then guide implementation of health interventions with MSM.
SUBSTANCE USE AND TECHNOLOGY 93
The procedures presented in this chapter were developed by building on the work of
methodologists who, using the innovations available to them at the time, strengthened what is
known about YMSM (Delaney et al., 2014; Ford et al., 2009; Rice et al., 2012). Therefore, it is
the author’s hope and belief that the methods presented in this study regarding GSNA-based
probability sampling will continue to evolve to make hidden populations more accessible.

SUBSTANCE USE AND TECHNOLOGY 94
References
Allen, J. E., Mansergh, G., Mimiaga, M. J., Holman, J., & Herbst, J. H. (2017). Mobile phone
and internet use mostly for sex-seeking and associations with sexually transmitted
infections and sample characteristics among Black/African American and
Hispanic/Latino men who have sex with men in 3 US cities. Sexually Transmitted
Diseases, 44, 284–289. doi:10.1097/OLQ.0000000000000590
Almeida, J., Johnson, R. M., Corliss, H. L., Molnar, B. E., & Azrael, D. (2009). Emotional
distress among LGBT youth: The influence of perceived discrimination based on sexual
orientation. Journal of Youth and Adolescence, 38, 1001–1014. doi:10.1007/s10964-009-
9397-9
Bauermeister, J. A., Pingel, E. S., Jadwin-Cakmak, L., Harper, G. W., Horvath, K., Weiss, G., &
Dittus, P. (2015). Acceptability and preliminary efficacy of a tailored online HIV/STI
testing intervention for young men who have sex with men: The Get Connected!
program. AIDS and Behavior, 19, 1860–1874. doi:10.1007/s10461-015-1009-y
Boone, M. R., Cook, S. H., & Wilson, P. (2013). Substance use and sexual risk behavior in HIV-
positive men who have sex with men: An episode-level analysis. AIDS and Behavior, 17,
1883–1887. doi:10.1007/s10461-012-0167-4
Boyd, C. J., McCabe, S. E., & d’Arcy, H. (2003). Ecstasy use among college undergraduates:
Gender, race and sexual identity. Journal of Substance Abuse Treatment, 24, 209–215.
doi:10.1016/S0740-5472(03)00025-4
Buckingham, L., Becher, J., Voytek, C. D., Fiore, D., Dunbar, D., Davis-Vogel, A., … Frank, I.
(2017). Going social: Success in online recruitment of men who have sex with men for
SUBSTANCE USE AND TECHNOLOGY 95
prevention HIV vaccine research. Vaccine, 35, 3498–3505.
doi:10.1016/j.vaccine.2017.05.002
Charurat, M. E., Emmanuel, B., Akolo, C., Keshinro, B., Nowak, R. G., Kennedy, S., …
Blattner, W. (2015). Uptake of treatment as prevention for HIV and continuum of care
among HIV-positive men who have sex with men in Nigeria. Journal of Acquired
Immune Deficiency Syndromes, 68, S114–S123. doi:10.1097/QAI.0000000000000439
Cochran, S. D., Sullivan, J. G., & Mays, V. M. (2003). Prevalence of mental disorders,
psychological distress, and mental health services use among lesbian, gay, and bisexual
adults in the United States. Journal of Consulting and Clinical Psychology, 71, 53–61.
doi:10.1037/0022-006X.71.1.53
Delaney, K. P., Kramer, M. R., Waller, L. A., Flanders, W. D., & Sullivan, P. S. (2014). Using a
geolocation social networking application to calculate the population density of sex-
seeking gay men for research and prevention services. Journal of Medical Internet
Research, 16, e249. doi:10.2196/jmir.3523
Dolezal, C., Carballo-Diéguez, A., Nieves-Rosa, L., & Dı
́ az, F. (2000). Substance use and sexual
risk behavior: Understanding their association among four ethnic groups of Latino men
who have sex with men. Journal of Substance Abuse, 11, 323–336. doi:10.1016/S0899-
3289(00)00030-4
Fernández, M. I., Varga, L. M., Perrino, T., Collazo, J. B., Subiaul, F., Rehbein, A., … Bowen,
G. S. (2004). The Internet as recruitment tool for HIV studies: Viable strategy for
reaching at-risk Hispanic MSM in Miami? AIDS Care, 16, 953–963.
doi:10.1080/09540120412331292480
Fernández, M. I., Warren, J. C., Varga, L. M., Prado, G., Hernandez, N., & Bowen, G. S. (2007).
SUBSTANCE USE AND TECHNOLOGY 96
Cruising in cyber space: Comparing Internet chat room versus community venues for
recruiting Hispanic men who have sex with men to participate in prevention studies.
Journal of Ethnicity in Substance Abuse, 6, 143–162. doi:10.1300/J233v06n02_09
Ford, W. L., Weiss, G., Kipke, M. D., Ritt-Olson, A., Iverson, E., & Lopez, D. (2009). The
Healthy Young Men’s Study: Sampling methods to recruit a random cohort of young
men who have sex with men. Journal of Gay & Lesbian Social Services, 21, 357–373.
doi:10.1080/10538720802498280
Frost, S. D. W., Brouwer, K. C., Cruz, M. A. F., Ramos, R., Ramos, M. E., Lozada, R. M., …
Strathdee, S. A. (2006). Respondent-driven sampling of injection drug users in two U.S.–
Mexico border cities: Recruitment dynamics and impact on estimates of HIV and syphilis
prevalence. Journal of Urban Health, 83(Suppl. 1), 83–97. doi:10.1007/s11524-006-
9104-z
Gattis, M. N., Sacco, P., & Cunningham-Williams, R. M. (2012). Substance use and mental
health disorders among heterosexual identified men and women who have same-sex
partners or same-sex attraction: Results from the National Epidemiological Survey on
Alcohol and Related Conditions. Archives of Sexual Behavior, 41, 1185–1197.
doi:10.1007/s10508-012-9910-1
Goedel, W. C., Brooks, F. A., & Duncan, D. T. (2016). Approaches to sampling gay, bisexual,
and other men who have sex with men from geosocial-networking smartphone
applications: A methodological note. Social Sciences, 5, 51. doi:10.3390/socsci5040051
Goedel, W. C., & Duncan, D. T. (2015). Geosocial-networking app usage patterns of gay,
bisexual, and other men who have sex with men: Survey among users of Grindr, a mobile
dating app. JMIR Public Health and Surveillance, 1, e4. doi:10.2196/publichealth.4353
SUBSTANCE USE AND TECHNOLOGY 97
Goedel, W. C., & Duncan, D. T. (2016). Contextual factors in geosocial-networking smartphone
application use and engagement in condomless anal intercourse among gay, bisexual, and
other men who have sex with men who use Grindr. Sexual Health, 13, 549–554.
doi:10.1071/SH16008
Goedel, W. C., Hagen, D., Halkitis, P. N., Greene, R. E., Griffin-Tomas, M., Brooks, F. A., …
Duncan, D. T. (2017). Post-exposure prophylaxis awareness and use among men who
have sex with men in London who use geosocial-networking smartphone applications.
AIDS Care, 29, 579–586. doi:10.1080/09540121.2016.1259455
Goedel, W. C., Mayer, K. H., Mimiaga, M. J., & Duncan, D. T. (2017). Considerable interest in
pre-exposure prophylaxis uptake among men who have sex with men recruited from a
popular geosocial-networking smartphone application in London. Global Public Health.
Advance online publication. doi:10.1080/17441692.2017.1391859
Goedel, W. C., Safren, S. A., Mayer, K. H., & Duncan, D. T. (2017). Community-level norms
and condomless anal intercourse among gay, bisexual, and other men who have sex with
men who use geosocial-networking smartphone applications in the Deep South. Journal
of HIV/AIDS & Social Services, 16, 382–389. doi:10.1080/15381501.2017.1344599
Gold, R. S., Skinner, M. J., & Ross, M. W. (1994). Unprotected anal intercourse in HIV-infected
and non-HIV-infected gay men. Journal of Sex Research, 31, 59–77.
doi:10.1080/00224499409551730
Goldberg, S., Strutz, K. L., Herring, A. A., & Halpern, C. T. (2013). Risk of substance abuse and
dependence among young adult sexual minority groups using a multidimensional
measure of sexual orientation. Public Health Reports, 128, 144–152.
doi:10.1177/003335491312800304
SUBSTANCE USE AND TECHNOLOGY 98
Grindr. (2017). Grindr terms and conditions of service. Retrieved from
https://www.grindr.com/terms-of-service
Grindr. (2018). Grindr for equality. Retrieved from https://www.grindr.com/community
Grov, C. (2012). HIV risk and substance use in men who have sex with men surveyed in
bathhouses, bars/clubs, and on Craigslist.org: Venue of recruitment matters. AIDS and
Behavior, 16, 807–817. doi:10.1007/s10461-011-9999-6
Grov, C., & Crow, T. (2012). Attitudes about and HIV risk related to the “most common place”
MSM meet their sex partners: Comparing men from bathhouses, bars/clubs, and
Craigslist.org. AIDS Education and Prevention, 24, 102–116.
doi:10.1521/aeap.2012.24.2.102
Grov, C., Rendina, H. J., Jimenez, R., & Parsons, J. T. (2016). Using online settings to identify
gay and bisexual men willing to take or with experience taking PrEP: Implications for
researchers and providers. AIDS Education and Prevention, 28, 378–392.
doi:10.1521/aeap.2016.28.5.378
Gwadz, M. V., Cleland, C. M., Quiles, R., Nish, D., Welch, J., Michaels, L. S., … Leonard, N.
R. (2010). CDC HIV testing guidelines and the rapid and conventional testing practices
of homeless youth. AIDS Education and Prevention, 22, 312–327.
doi:10.1521/aeap.2010.22.4.312
Hagger-Johnson, G., Taibjee, R., Semlyen, J., Fitchie, I., Fish, J., Meads, C., & Varney, J.
(2013). Sexual orientation identity in relation to smoking history and alcohol use at age
18/19: Cross-sectional associations from the Longitudinal Study of Young People in
England (LSYPE). BMJ Open, 3, e002810. doi:10.1136/bmjopen-2013-002810
SUBSTANCE USE AND TECHNOLOGY 99
Halkitis, P. N., & Figueroa, R. P. (2013). Sociodemographic characteristics explain differences
in unprotected sexual behavior among young HIV-negative gay, bisexual, and other
YMSM in New York City. AIDS Patient Care and STDs, 27, 181–190.
doi:10.1089/apc.2012.0415
Hatzenbuehler, M. L., Corbin, W. R., & Fromme, K. (2008). Trajectories and determinants of
alcohol use among LGB young adults and their heterosexual peers: Results from a
prospective study. Developmental Psychology, 44, 81–90. doi:10.1037/0012-
1649.44.1.81
Hatzenbuehler, M. L., Hilt, L. M., & Nolen-Hoeksema, S. (2010). Gender, sexual orientation,
and vulnerability to depression. In J. C. Chrisler & D. R. McCreary (Eds.), Handbook of
gender research in psychology (Vol. 2, pp. 133–151). New York, NY: Springer.
Heckathorn, D. D. (1997). Respondent-driven sampling: A new approach to the study of hidden
populations. Social Problems, 44, 174–199. doi:10.2307/3096941
Heckathorn, D. D., Semaan, S., Broadhead, R. S., & Hughes, J. J. (2002). Extensions of
respondent-driven sampling: A new approach to the study of injection drug users aged
18–25. AIDS and Behavior, 6, 55–67. doi:10.1023/A:1014528612685
Holloway, I. W., Rice, E., Gibbs, J., Winetrobe, H., Dunlap, S., & Rhoades, H. (2014).
Acceptability of smartphone application-based HIV prevention for among young men
who have sex with men. AIDS and Behavior, 18, 285–296. doi:10.1007/s10461-013-
0671-1
Hornet. (2018a). About us. Retrieved from https://love.hornet.com/about/
Hornet. (2018b). Hornet capabilities. Retrieved from
https://love.hornet.com/advertising/overview/
SUBSTANCE USE AND TECHNOLOGY 100
Hornet. (2018c). Terms of service. Retrieved from http://love.hornet.com/terms-of-service/
Hornet. (2018d). Know your status. Retrieved from: https://love.hornet.com/know-your-status/
Horvath, K. J., Bowen, A. M., & Williams, M. L. (2006). Virtual and physical venues as contexts
for HIV risk among rural men who have sex with men. Health Psychology, 25, 237–242.
doi:10.1037/0278-6133.25.2.237
Horvath, K. J., Rosser, B. S., & Remafedi, G. (2008). Sexual risk taking among young Internet-
using men who have sex with men. American Journal of Public Health, 98, 1059–1067.
doi:10.2105/AJPH.2007.111070
Hospers, H. J., Kok, G., Harterink, P., & de Zwart, O. (2005). A new meeting place: Chatting on
the Internet, e-dating and sexual risk behaviour among Dutch men who have sex with
men. AIDS, 19, 1097–1101. doi:10.1097/01.aids.0000174457.08992.62
Kalichman, S. C., Kelly, J. A., Morgan, M., & Rompa, D. (1997). Fatalism, current life
satisfaction, and risk for HIV infection among gay and bisexual men. Journal of
Consulting and Clinical Psychology, 65, 542–546. doi:10.1037/0022-006X.65.4.542
Kelly, J., Davis, C., & Schlesinger, C. (2015). Substance use by same sex attracted young
people: Prevalence, perceptions and homophobia. Drug and Alcohol Review, 34, 358–
365. doi:10.1111/dar.12158
Kipke, M. D., Weiss, G., Ramirez, M., Dorey, F., Ritt-Olson, A., Iverson, E., & Ford, W. (2007).
Club drug use in Los Angeles among young men who have sex with men. Substance Use
& Misuse, 42, 1723–1743. doi:10.1080/10826080701212261
Kipke, M. D., Weiss, G., & Wong, C. F. (2007). Residential status as a risk factor for drug use
and HIV risk among young men who have sex with men. AIDS and Behavior, 11(Suppl.
2), 56–69. doi:10.1007/s10461-006-9204-5
SUBSTANCE USE AND TECHNOLOGY 101
Kuhns, L. M., Kwon, S., Ryan, D. T., Garofalo, R., Phillips, G., II, & Mustanski, B. S. (2015).
Evaluation of respondent-driven sampling in a study of urban young men who have sex
with men. Journal of Urban Health, 92, 151–167. doi:10.1007/s11524-014-9897-0
Lachowsky, N. J., Lal, A., Forrest, J. I., Card, K. G., Cui, Z., Sereda, P., … Hogg, R. S. (2016).
Including online-recruited seeds: A respondent-driven sample of men who have sex with
men. Journal of Medical Internet Research, 18, e51. doi:10.2196/jmir.5258
Lauby, J. L., Millett, G. A., LaPollo, A. B., Bond, L., Murrill, C. S., & Marks, G. (2008). Sexual
risk behaviors of HIV-positive, HIV-negative, and serostatus-unknown Black men who
have sex with men and women. Archives of Sexual Behavior, 37, 708–719.
doi:10.1007/s10508-008-9365-6
Lelutiu-Weinberger, C., Pachankis, J. E., Gamarel, K. E., Surace, A., Golub, S. A., & Parsons, J.
T. (2015). Feasibility, acceptability, and preliminary efficacy of a live-chat social media
intervention to reduce HIV risk among young men who have sex with men. AIDS and
Behavior, 19, 1214–1227. doi:10.1007/s10461-014-0911-z
Léon, L., Des Jarlais, D., Jauffret-Roustide, M., & Le Strat, Y. (2016). Update on respondent-
driven sampling: Theory and practical considerations for studies of persons who inject
drugs. Methodological Innovations, 9. doi:10.1177/2059799116672878
Leslie, J. (2009, October 1). Grindr app makes it easy to be easy. Out & About Nashville.
Retrieved from https://www.outandaboutnashville.com/story/grindr-app-makes-it-easy-
be-easy
Lorimer, K., Flowers, P., Davis, M., & Frankis, J. (2016). Young men who have sex with men’s
use of social and sexual media and sex-risk associations: Cross-sectional, online survey
SUBSTANCE USE AND TECHNOLOGY 102
across four countries. Sexually Transmitted Infections, 92, 371–376.
doi:10.1136/sextrans-2015-052209
Lucid Dreams. (2014a). Application end user license agreement. Retrieved from
https://www.jackd.mobi/#terms
Lucid Dreams. (2014b). Terms of access and use. Retrieved from
https://www.jackd.mobi/#terms
MacKellar, D. A., Gallagher, K. M., Finlayson, T., Sanchez, T., Lansky, A., & Sullivan, P. S.
(2007). Surveillance of HIV risk and prevention behaviors of men who have sex with
men—a national application of venue-based, time-space sampling. Public Health
Reports, 122(1, Suppl.), 39–47. doi:10.1177/00333549071220S107
Malekinejad, M., Johnston, L. G., Kendall, C., Kerr, L. R. F. S., Rifkin, M. R., & Rutherford, G.
W. (2008). Using respondent-driven sampling methodology for HIV biological and
behavioral surveillance in international settings: A systematic review. AIDS and
Behavior, 12(Suppl. 1), 105–130. doi:10.1007/s10461-008-9421-1
Mansergh, G., Flores, S., Koblin, B., Hudson, S., McKirnan, D., Colfax, G. N., & the Project
MIX Study Group. (2008). Alcohol and drug use in the context of anal sex and other
factors associated with sexually transmitted infections: Results from a multi-city study of
high-risk men who have sex with men in the USA. Sexually Transmitted Infections, 84,
509–511. doi:10.1136/sti.2008.031807
Marshal, M. P., Dietz, L. J., Friedman, M. S., Stall, R., Smith, H. A., McGinley, J., … Brent, D.
A. (2011). Suicidality and depression disparities between sexual minority and
heterosexual youth: A meta-analytic review. Journal of Adolescent Health, 49, 115–123.
doi:10.1016/j.jadohealth.2011.02.005
SUBSTANCE USE AND TECHNOLOGY 103
McCabe, S. E., Boyd, C., Hughes, T. L., & d’Arcy, H. (2003). Sexual identity and substance use
among undergraduate students. Substance Abuse, 24, 77–91.
doi:10.1023/A:1023768215020
McCabe, S. E., Hughes, T. L., Bostwick, W. B., West, B. T., & Boyd, C. J. (2009). Sexual
orientation, substance use behaviors and substance dependence in the United States.
Addiction, 104, 1333–1345. doi:10.1111/j.1360-0443.2009.02596.x
Merchant, R. C., Romanoff, J., Clark, M. A., Liu, T., Rosenberger, J. G., Bauermeister, J., &
Mayer, K. H. (2017). Variations in recruitment yield and characteristics of participants
recruited across diverse internet platforms in an HIV testing study of young adult men-
who-have-sex-with-men (YMSM). American Journal of Men’s Health, 11, 1342–1357.
doi:10.1177/1557988317717383
Meyer, I. H., & Wilson, P. A. (2009). Sampling lesbian, gay, and bisexual populations. Journal
of Counseling Psychology, 56, 23–31. doi:10.1037/a0014587
Mills, T. C., Paul, J., Stall, R., Pollack, L., Canchola, J., Chang, Y. J. . . . Catania, J. A. (2004).
Distress and depression in men who have sex with men: The Urban men’s health study.
American Journal of Psychiatry, 161, 278–285. doi:10.1176/appi.ajp.161.2.278
Mutchler, M. G., McKay, T., Candelario, N., Liu, H., Stackhouse, B., Bingham, T., & Ayala, G.
(2011). Sex drugs, peer connections, and HIV: Use and risk among African-American,
Latino, and multiracial young men who have sex with men in Los Angeles and New
York. Journal of Gay & Lesbian Social Services, 23, 271–295.
doi:10.1080/10538720.2011.560100
Operario, D., Choi, K.-H., Chu, P. L., McFarland, W., Secura, G. M., Behel, S., … Valleroy, L.
(2006). Prevalence and correlates of substance use among young Asian Pacific Islander
SUBSTANCE USE AND TECHNOLOGY 104
men who have sex with men. Prevention Science, 7, 19–29. doi:10.1007/s11121-005-
0018-x
Pequegnat, W., Rosser, B. R. S., Bowen, A. M., Bull, S. S., DiClemente, R. J., Bockting, W. O.,
… Konstan, J. (2007). Conducting Internet-based HIV/STD prevention survey research:
Considerations in design and evaluation. AIDS and Behavior, 11, 505–521.
doi:10.1007/s10461-006-9172-9
Perry Street Software. (2017). Terms of service. Retrieved from https://www.scruff.com/tos/
Pew Research Center. (2018). Mobile fact sheet: Who owns cellphones and smartphones.
Retrieved from http://www.pewinternet.org/fact-sheet/mobile/
Phillips, G., II, Grov, C., & Mustanski, B. (2015). Engagement in group sex among geosocial
networking mobile application-using men who have sex with men. Sexual Health, 12,
495–500. doi:10.1071/SH15108
Phillips, G., II, Magnus, M., Kuo, I., Rawls, A., Peterson, J., Jia, Y., … Greenberg, A. E. (2014).
Use of geosocial networking (GSN) mobile phone applications to find men for sex by
men who have sex with men (MSM) in Washington, DC. AIDS and Behavior, 18, 1630–
1637. doi:10.1007/s10461-014-0760-9
Raymond, H. F., Rebchook, G., Curotto, A., Vaudrey, J., Amsden, M., Levine, D., & McFarland,
W. (2010). Comparing internet-based and venue-based methods to sample MSM in the
San Francisco Bay Area. AIDS and Behavior, 14, 218–224. doi:10.1007/s10461-009-
9521-6
Reisner, S. L., Mimiaga, M. J., Johnson, C. V., Bland, S., Case, P., Safren, S. A., & Mayer, K. H.
(2010). What makes a respondent-driven sampling “seed” productive? Example of
SUBSTANCE USE AND TECHNOLOGY 105
finding at-risk Massachusetts men who have sex with men. Journal of Urban Health, 87,
467–479. doi:10.1007/s11524-010-9439-3
Rhodes, S. D., Malow, R. M., & Jolly, C. (2010). Community-based participatory research: A
new and not-so-new approach to HIV/AIDS prevention, care, and treatment. AIDS
Education and Prevention, 22, 173–183. doi:10.1521/aeap.2010.22.3.173
Rice, E., Holloway, I., Winetrobe, H., Rhoades, H., Barman-Adhikari, A., Gibbs, J., … Dunlap,
S. (2012). Sex risk among young men who have sex with men who use Grindr, a
smartphone geosocial networking application. Journal of AIDS & Clinical Research, S4,
005. doi:10.4172/2155-6113.S4-005
Ridge, D. T., Plummer, D. C., & Minichiello, V. (1994). Young gay men and HIV: Running the
risk? AIDS Care, 6, 371–378. doi:10.1080/09540129408258651
Safarnejad, A., Nga, N. T., & Son, V. H. (2017). Population size estimation of men who have sex
with men in Ho Chi Minh City and Nghe An using social app multiplier method. Journal
of Urban Health, 94, 339–349. doi:10.1007/s11524-016-0123-0
Schwarcz, S., Spindler, H., Scheer, S., Valleroy, L., & Lansky, A. (2007). Assessing
representativeness of sampling methods for reaching men who have sex with men: A
direct comparison of results obtained from convenience and probability samples. AIDS
and Behavior, 11, 596. doi:10.1007/s10461-007-9232-9
Shoptaw, S., Weiss, R. E., Munjas, B., Hucks-Ortiz, C., Young, S. D., Larkins, S., … Gorbach,
P. M. (2009). Homonegativity, substance use, sexual risk behaviors, and HIV status in
poor and ethnic men who have sex with men in Los Angeles. Journal of Urban Health,
86(Suppl. 1), 77–92. doi:10.1007/s11524-009-9372-5
SUBSTANCE USE AND TECHNOLOGY 106
Siegler, A. J., Sanchez, T., Sineath, R. C., Grey, J., Kahle, E., & Sullivan, P. S. (2015).
Knowledge and awareness of acute human immunodeficiency virus infection among
mobile app-using men who have sex with men: A missed public health opportunity. Open
Forum Infectious Diseases, 2, ofv016. doi:10.1093/ofid/ofv016
Sullivan, P. S., Khosropour, C. M., Luisi, N., Amsden, M., Coggia, T., Wingood, G. M., &
DiClemente, R. J. (2011). Bias in online recruitment and retention of racial and ethnic
minority men who have sex with men. Journal of Medical Internet Research, 13, e38.
doi:10.2196/jmir.1797
Thiede, H., Valleroy, L. A., MacKellar, D. A., Celentano, D. D., Ford, W. L., Hagan, H., …
Torian, L. V. (2003). Regional patterns and correlates of substance use among young
men who have sex with men in 7 US urban areas. American Journal of Public Health, 93,
1915–1921. doi:10.2105/AJPH.93.11.1915
Thoma, B. C. (2017). Parental monitoring and HIV-related sexual risk behaviors among young
men who have sex with men: Associations with sexual activity and HIV-related sexual
risk behaviors. Journal of Adolescent Health, 61, 348–354.
doi:10.1016/j.jadohealth.2017.03.004
Traube, D. E., Schrager, S. M., Holloway, I. W., Weiss, G., & Kipke, M. D. (2013).
Environmental risk, social cognition, and drug use among young men who have sex with
men: Longitudinal effects of minority status on health processes and outcomes. Drug and
Alcohol Dependence, 127, 1–7. doi:10.1016/j.drugalcdep.2012.06.007
Tudor, M. (2012). Cyberqueer techno-practices: Digital space-making and networking among
Swedish gay men (Doctoral dissertation). Stockholm University, Stockholm, Sweden.
SUBSTANCE USE AND TECHNOLOGY 107
White, R. G., Lansky, A., Goel, S., Wilson, D., Hladik, W., Hakim, A., & Frost, S. D. W. (2012).
Respondent driven sampling—where we are and where should we be going? Sexually
Transmitted Infections, 88, 397–399. doi:10.1136/sextrans-2012-050703
Zablotska, I. B., Holt, M., & Prestage, G. (2012). Changes in gay men’s participation in gay
community life: Implications for HIV surveillance and research. AIDS and Behavior, 16,
669–675. doi:10.1007/s10461-011-9919-9
Zou, H., & Fan, S. (2017). Characteristics of men who have sex with men who use smartphone
geosocial networking applications and implications for HIV interventions: A systematic
review and meta-analysis. Archives of Sexual Behavior, 46, 885–894.
doi:10.1007/s10508-016-0709-3

SUBSTANCE USE AND TECHNOLOGY 108
Figure 4.1. Geographic Sampling Frame A

SUBSTANCE USE AND TECHNOLOGY 109
Figure 4.2. Geographic Sampling Frame B (GSF A Present)

SUBSTANCE USE AND TECHNOLOGY 110
Figure 4.3. Geographic Sampling Frame C (GSF B Present)

Sources: Esri, HERE, DeLorme, USGS, Intermap, INCREMENT P, NRCan, Esri Japan, METI, Esri China (Hong Kong), Esri Korea, Esri
(Thailand), MapmyIndia, NGCC, © OpenStreetMap contributors, and the GIS User Community
SUBSTANCE USE AND TECHNOLOGY 111
Figure 4.4. Geographic Sampling Frame B with Grid of 79 Sampling Points

SUBSTANCE USE AND TECHNOLOGY 112
Figure 4.5. Interpolated User Density on Mondays at 10 a.m.

SUBSTANCE USE AND TECHNOLOGY 113
Figure 4.6. Summation of Neighborhoods
Note. Each item indicates areas in which eight individuals on average could be observed within j distance on
Mondays at 10 a.m. (A = 1.6 km, B = 2 km, C = 3 km).

SUBSTANCE USE AND TECHNOLOGY 114
Figure 4.7. Math Algebra Process of Removing Summation of Neighborhood Overlap
Note. Item A is raster_M10_3.1.6 and Item B is raster_M10_3.2. After math algebra process of multiplication with
the inverse of Item A (raster_M10_4.1.6), raster_M10_5.2 is created (Item C). Item D displays raster_M10_3.1.6
and raster_M10_5.2.

SUBSTANCE USE AND TECHNOLOGY 115
Figure 4.8. Day-Time-Sampling-Locations for Mondays at 10 a.m. in GSF B

SUBSTANCE USE AND TECHNOLOGY 116
Figure 4.9. Steps for Analyzing User Density Data in ArcMap 10.5
2) Visualize x-y coordinates
3) Choose 1 SDTP (Monday 10am), interpolate data, sell
raster_M10_1
4) Math algebra: divide by number of z squares in
raster_M10_2
5) Series of focal statistics, neighborhood summation: round, k users, j distance to full GSF
raster_M10_3.j
raster_M10_3.1 raster_M10_3.1.6 raster_M10_3.2 raster_M10_3.3 raster_M10_3.3.2
raster_M10_4.1.6 raster_M10_4.2 raster_M10_4.1 raster_M10_4.3
6) Reclassify 1 to 0, 0 to 1
3km 3.2km 1km 2km 1.6km
× × × ×
7) Math algebra: Multiply
raster_M10_3.1 raster_M10_5.2 raster_M10_5.3 raster_M10_5.3.2 raster_M10_5.1.6
1) Import user density data file
j =
SUBSTANCE USE AND TECHNOLOGY 117
Figure 4.10. Day-Time-Sampling-Locations for Mondays at 10 a.m. in GSF C

Sources: Esri, HERE, DeLorme, USGS, Intermap, INCREMENT P, NRCan, Esri Japan, METI, Esri China (Hong Kong), Esri Korea, Esri
(Thailand), MapmyIndia, NGCC, © OpenStreetMap contributors, and the GIS User Community
SUBSTANCE USE AND TECHNOLOGY 118
Chapter 5 (Paper 2): Young Men Who Have Sex with Men and Substance Use: A
Comparison of Venue-Based Sampling and Geosocial Networking Application Sampling
Introduction
Research has indicated that young men who have sex with men (YMSM) are at higher
risk of using substances (i.e., alcohol, illicit drugs). Compared to heterosexual men, YMSM have
higher odds of using marijuana (McCabe, Boyd, Hughes, & d’Arcy, 2003; McCabe, Hughes,
Bostwick, West, & Boyd, 2009), ecstasy (Boyd, McCabe, & d’Arcy, 2003; McCabe et al., 2003),
and other hard drugs (McCabe et al., 2009). In addition to higher levels of alcohol consumption
compared to their heterosexual peers (Hagger-Johnson et al., 2013), YMSM also drink to
intoxication more often (Hatzenbuehler, Corbin, & Fromme, 2008) and exhibit higher odds of
developing alcohol dependence (Goldberg, Strutz, Herring, & Halpern, 2013). Further, substance
use has been linked to numerous negative health-related outcomes (e.g., cardiovascular disease,
stroke, cancer, and liver disease; National Institute on Drug Abuse, 2012). For YMSM, the risk
associated with substance use is intensified due to a well-established relationship between
substance use and unprotected anal sex. Binge drinking, marijuana, poppers, methamphetamine,
and cocaine use are each associated with higher odds of unprotected anal sex in this population
(Colfax et al., 2004; Hess et al., 2015; Moeller, Palamar, Halkitis, & Siconolfi, 2014; Parsons,
Lelutiu-Weinberger, Botsko, & Golub, 2013). Although the disparities in YMSM substance use
and their association with sexual risk are clear in the literature, the relationships between
substance use and psychosocial factors, which ultimately inform substance use intervention
development (Kurtz, Stall, Buttram, Surratt, & Chen, 2013), are clouded by issues of sampling
bias.
SUBSTANCE USE AND TECHNOLOGY 119
Studies examining the psychosocial covariates of substance use (e.g., internalized stigma,
experiences of homophobia, depression symptomology, connection with the gay community, and
sensation seeking; Duncan, Hatzenbuehler, & Johnson, 2014; McCabe, Bostwick, Hughes, West,
& Boyd, 2010; Puckett, Newcomb, Garofalo, & Mustanski, 2017; Ryan, Huebner, Diaz, &
Sanchez, 2009; Rosario, Schrimshaw, & Hunter, 2009; Shoptaw et al., 2009; Stall et al., 2001;
Traube, Schrager, Holloway, Weiss, & Kipke, 2013; Trocki, Drabble, & Midanik, 2009; Wong,
Weiss, Ayala, & Kipke, 2010) for YMSM have relied on samples recruited from gay venues that
arguably condone substance use (e.g., gay bars, local gay events; Kelly, Davis, & Schlesinger,
2015; Kipke, Weiss, Ramirez, et al., 2007; Kipke, Weiss, & Wong, 2007; Mutchler et al., 2011;
Operario et al., 2006; Thiede et al., 2003; Traube et al., 2013). Venue-based sampling is the most
widely accepted recruitment methodology used to sample YMSM. This type of recruitment
involves the research team first identifying locations where YMSM congregate and then
approaching perceived potential participants, inquiring about eligibility, and inviting them to
participate in the study (Meyer & Wilson, 2009).
Although many studies using venue-based sampling with YMSM have relied on
convenience samples (Kelly et al., 2015; Mutchler et al., 2011), more rigorous procedures have
been developed that rely on a preliminary enumeration (i.e., counting of potential participants)
stage for identifying both times and locations of YMSM congregation (Kipke, Weiss, Ramirez,
et al., 2007; Kipke, Weiss, & Wong, 2007; MacKellar et al., 2007; Operario et al., 2006; Thiede
et al., 2003; Traube et al., 2013). Venue-based stratified probability sampling helps ensure that
participants are recruited based on probability rather than convenience (Ford et al., 2009).
Despite this enhancement, these recruitment strategies have been criticized for excluding YMSM
who either do not identify as gay or bisexual but engage in same-sex behaviors or have a limited
SUBSTANCE USE AND TECHNOLOGY 120
connection to the gay community (Meyer & Wilson, 2009). This sampling bias is a critical
concern in substance abuse research because it may artificially inflate the reporting of substance
use prevalence in this community and continue a deficit expectation in a marginalized
population. Further, it may cloud our understanding of risk factors for substance use that would
be uncovered in a more representative sample of YMSM.
Geosocial Networking Applications
Recent literature has suggested that YMSM are now using technology to meet other
YMSM and relying less on venues to provide social interaction (Grov, 2012; Grov & Crow,
2012; Zablotska, Holt, & Prestage, 2012). In 2009, the first geosocial networking application
(GSNA) was introduced to the smartphone app market (Leslie, 2009). These technologies use
geolocation data to allow YMSM users to contact and connect in real time with other users. The
format of GSNAs have also made it possible to filter by age and recruit a probability sample of
YMSM. GSNAs like Hornet, which reports having more than 25 million users worldwide
(Hornet, 2018), are estimated to be used by two thirds of men who have sex with men (Phillips et
al., 2014). Although GSNAs require the use of a smart device, it is estimated that 94% of 18- to
29-year-olds in the United States have access to this technology (Pew Research Center, 2018).
GSNAs provide a profile-based home screen environment in which YMSM can browse
the profiles of other YMSM in the same geographic region. Because profiles are ordered based
on their relative proximity to the user, YMSM users may click on a profile and see the distance
of other users from their current location. Whereas YMSM previously had limited access to
socialization, except through gay venues, they can now open a GSNA and easily find other
YMSM. Portions of the YMSM community that previously attended gay venues specifically to
meet other men (and not to use alcohol or other substances) may no longer use those venues to
SUBSTANCE USE AND TECHNOLOGY 121
facilitate connections. Therefore, GSNAs may allow advancement of YMSM recruitment
methods. GSNA procedures have been developed and used in previous studies to recruit both
convenience samples (Buckingham et al., 2017; Duncan et al., 2018; Goedel & Duncan, 2015,
2016; Goedel, Hagen, et al., 2017; Goedel, Halkitis, & Duncan, 2016; Goedel, Safren, Mayer, &
Duncan, 2017; Holloway et al., 2017; Lachowsky et al., 2016; Lorimer, Flowers, Davis, &
Frankis, 2016; Macapagal, Coventry, Puckett, Phillips, & Mustanski, 2016; Rendina, Jimenez,
Grov, Ventuneac, & Parsons, 2014; Siegler et al., 2015) and probability sampling of YMSM
(Gibbs & Rice, 2016; Holloway, Pulsipher, Gibbs, Barman-Adhikari, & Rice, 2015; Holloway et
al., 2014; Rice et al., 2012).
Despite the potential to inform data collection with YMSM, the use of this GSNA in
research has been criticized for being biased toward YMSM who are single, younger, and more
sexually risky (Beymer et al., 2014; Burrell et al., 2012; Choi, Wong, & Fong, 2017; Eaton et al.,
2016; Lehmiller & Ioerger, 2014; Phillips et al., 2014). These criticisms, however, are based on a
limited number of studies that either did not use random sampling procedures (Burrell et al.,
2012) or stratified samples of venue-recruited YMSM based on their GSNA use rather than
sampling using a GSNA (Eaton et al., 2016; Lehmiller & Ioerger, 2014; Phillips et al., 2014). To
date, no study has compared the substance use characteristics and associated psychosocial
determinants of a probability sample recruited through a GSNA to a probability sample of venue-
recruited YMSM.
Therefore, the present study compared the most widely accepted form of recruiting a
probability sample of YMSM (i.e., venue-based stratified probability sampling) to a novel
method that used a location-based GSNA. The main aim of this study was to investigate whether
SUBSTANCE USE AND TECHNOLOGY 122
differences exist in demographics, psychosocial correlates of substance use, and substance use
between venue-recruited and GSNA-recruited probability samples of YMSM.
Methods
A cross-sectional online survey was conducted in Los Angeles County from April 2017
through April 2018 using two forms of recruitment. Of 111 YMSM (aged 18–24 years old), 68
were recruited using venue-based stratified probability sampling procedures and 43 were
recruited using GSNA-based probability sampling procedures. The venue-based stratified
probability sampling methods were informed by the Healthy Young Men’s Study conducted in
Los Angeles (Ford et al., 2009). The GSNA methods were informed by a previous study by this
dissertation’s author and colleagues, also conducted in Los Angeles (Rice et al., 2012).
Geographic information systems methods were also used to build the GSNA sampling frame and
were informed by a 2014 study conducted in Atlanta, Georgia (Delaney, Kramer, Waller,
Flanders, & Sullivan, 2014). Both recruitment methods followed two stages: (1) building of the
sampling frame and enumeration, and (2) recruitment of participants using a monthly sampling
calendar. All stages were completed with a specific geographic sampling frame (GSF; location
for recruitment). The principal area spanned approximately 70 square kilometers, and then this
area was enlarged to an area approximately 50 times larger (the southern portion of Los Angeles
County) to maximize recruitment. All procedures for this study were approved by the
institutional review board of the University of Southern California.
Building the Sampling Frame and Enumeration
Venue-based stratified probability sample. To identify the study sampling frame for
venue-based recruitment, the study team conducted brief interviews with men in public areas in
the GSF and asked interviewees to report known locations for finding YMSM, resulting in a final
SUBSTANCE USE AND TECHNOLOGY 123
list of 41 locations. The study team presented the compiled list of venues to a community
advisory board and asked members to nominate missing venues, eliminate venues believed to be
inappropriate, and provide insight regarding the best times to find YMSM during the four
standard 4-hour sampling periods (8 a.m.–12 p.m., 12 p.m.–4 p.m., 4 p.m.–8 p.m., and 8 p.m.–12
a.m.). Participants received a $30 gift card incentive, and the work resulted in an exhaustive list
of 43 venues currently frequented by YMSM. Investigation of each venue using Google’s
“Popular Times” function, information from each venue’s website, and the information gathered
from the focus group identified 199 periods theoretically high in YMSM attendance (i.e., venue-
day-time period; VDT).
Venue enumeration was conducted during 7 months. The procedures of enumeration
followed those outlined by the Healthy Young Men’s Study (Ford et al., 2009). Type I
enumeration required one study team member to visit each venue during the identified 4-hour
time periods (i.e., 199 VDTs). For 60 minutes, the study team member counted the number of
individuals who appeared to fit the study criteria (i.e., male, aged 18–24). The number counted
was then multiplied by four to approximate a 4-hour time period. Pilot enumerations were also
conducted wherein one study team member counted individuals who appeared to meet the study
criteria and a second study team member asked those individuals if they met the eligibility
criteria (i.e., date of birth, gender, and self-identification as gay, bisexual, or uncertain about
sexual orientation or report of having sex with men). Based on the pilot enumerations, team
members correctly identified eligible individuals 65% of the time, and this number was
multiplied by the 4-hour enumeration to estimate the number of eligible individuals.
An estimated eight individuals in a 4-hour time period was identified as the threshold for
an appropriate VDT based on recommendations from previous studies (Ford et al., 2009;
SUBSTANCE USE AND TECHNOLOGY 124
MacKellar et al., 2007). If a Type I enumeration for a specific VDT resulted in an estimate lower
than eight eligible individuals, then the VDT required a Type II enumeration. Of the 199 Type I
enumerations completed, 144 were determined to be appropriate and 55 required a Type II
enumeration.
Type II enumerations were conducted by two study team members for two 1-hour periods
separated by 1 hour and followed the same procedures as the pilot enumerations. If a VDT was
found to have an estimate less than eight, it was excluded from the sampling frame. Of the 55
Type II enumerations completed, 12 were identified as appropriate and 43 as inappropriate and
excluded as potential VDTs. At the end of Stage 2, 156 VDTs were identified as appropriate for
recruitment.
GSNA probability sample. The community advisory board was also asked to report on
the most popular GSNAs used by YMSM in the GSF. Each proposed GSNA needed to meet
three criteria to be included: (a) ability to filter users by age, (b) ordering of user profiles by
relative proximity to the study smart device, and (c) reporting of the relative distance of users
from the study smart device. The top three GSNAs were contacted by the study’s principal
investigator to inquire about interest in partnership in the current study. From these
conversations, the study GSNA was selected (i.e., Hornet).
The following procedures were used to recruit the majority of the GSNA-based sample
(all but six, who were recruited during pilot testing of the methods). Procedures used to collect
user density data are outlined in Chapter 4 of this dissertation. Similar to the venue-based
procedures, the GSNA-based methods required that users could be recruited from each sampling-
day-time period (SDTP; four periods × 7 days = 28). The periods used were the midpoints of
each of the VDT time periods: 10 a.m., 2 p.m., 6 p.m., and 10 p.m.
SUBSTANCE USE AND TECHNOLOGY 125
Five points in the GSF were chosen as the locations for recruitment to maximize
coverage and allow all the users of the study GSNA to have the same probability of being
approached for recruitment. The number of sampling points needed for each SDTP for the GSF
was determined by three criteria: point-based circular buffers of 8 miles must (a) maximize
coverage of the GSF, (b) minimize buffer overlap, and (c) minimize coverage of areas outside
the GSF boundary. Once completed, 140 day-time-sampling locations (DTSLs) were identified
(five points × 28 SDTPs).
Recruitment and Data Collection
A monthly sampling calendar was used to randomly select both VDTs and DTSLs.
Creation of calendars followed the procedures used in Ford et al. (2009). To create the monthly
sampling calendar, all VDTs and DTSLs were loaded into a spreadsheet. VDTs and DTSLs were
categorized as weekday day, weekday night, weekend day, and weekend night to allow for
matching between recruitment methods, because it was hypothesized that different individuals
would be located in the GSF at each of these times. One randomly selected VDT or DTSL was
placed in the appropriate day and time on the calendar, and a matched random selection was
taken from the other recruitment method. Selection of VDTs and DTSLs continued until 10
VDTs and three or four DTSLs (the study GSNA preferred to limit monthly recruitment periods
to not overburden users) were placed on the study calendar. Once the targeted sample size was
achieved for a recruitment strategy, the corresponding remaining sampling events for that month
were disregarded.
For venue-based recruitment, in accordance with the monthly sampling calendar, two or
three study team members recruited a sample of participants from the specified VDT. While one
study team member counted the number of men who appeared to be eligible, the remaining study
SUBSTANCE USE AND TECHNOLOGY 126
team members individually approached potential participants and asked if they were willing to
answer the eligibility questions to support research. If the potential participant agreed to
participate, he was handed the study iPad and completed the eligibility survey. At the end of the
eligibility survey, the potential participant created a unique study ID code and input his contact
information (i.e., email address, mobile phone number). The potential participant then received a
$5 physical gift card incentive for completing the eligibility survey.
For GSNA-based recruitment, instead of visiting the specified geographic point, the study
GSNA was technologically misled and provided with GPS coordinates for the DTSL specified
on the monthly sampling calendar (using a third-party application, “Fake GPS Location”). This
allowed the study team to stay in a fixed location. As noted, each DTSL had an associated GPS
point and radial distance (e.g., 8 miles). The recruitment process began by one study team
member opening the GSNA on the study smart device and making an anonymous user profile.
After filtering the visible user profiles by age eligibility (i.e., 18–24 years old), the study team
member recorded all public GSNA user IDs within 8 miles of the study smart device.
Concurrently, the study GSNA (i.e., Hornet) sent out a preset direct message broadcast in the
GSNA to each GSNA user within 8 miles of the DTSL. The direct message (e.g., “Hey, you
appear to be eligible for a USC study! Do you have 2 minutes to answer a few questions? If
you’re eligible and you take the survey, you can earn up to $50”) appeared as a personal message
from Hornet and remained in users’ message inbox. Once the potential participant clicked on the
link, he could complete the same eligibility survey previously described. No eligibility incentive
was provided to GSNA-recruited potential participants because pilot tests revealed this
procedure did not increase the rate of screening completion.
SUBSTANCE USE AND TECHNOLOGY 127
The eligibility data for both recruitment methods were received by the contact
information manager, who determined if individuals were eligible to take the study survey (i.e.,
met the study eligibility criteria and had a unique study ID code indicating they had not
previously taken the main study survey). The contact information manager then provided a list of
new eligible unique study IDs to the data manager, who never had access to the contact
information of the participants. The data manager used the unique study IDs to create unique
one-use web links to the main study survey and provided the contact information manager with
these links. The contact information manager then contacted the potential participant using his
preferred mode of contact with instructions to take the main study survey, his unique study ID,
and the unique link to take the main study survey. The contact information manager never had
access to the main study survey data.
Upon entering the main study survey website, potential participants had the opportunity
to read informed consent information. After giving informed consent, participants were
instructed to enter their unique study ID code. The main study survey asked participants to report
on demographic information, substance use characteristics, substance use determinants, and
sexual risk behaviors. After completing the main study survey, participants were asked to
indicate if they had previously completed the study. This item was asked as a safeguard against
duplicate data. Once a participant completed the main study survey, the submitted data were
available to the data manager. The data manager provided a list of unique study ID codes for
individuals who completed the main study survey and unique links to an associated supplemental
survey to the contact information manager. The contact information manager then sent each
participant a downloadable gift card ($25 for venue-recruited participants and $35 for GSNA-
recruited participants). In addition, the contact information manager sent each participant who
SUBSTANCE USE AND TECHNOLOGY 128
completed the main study survey a link to the supplemental survey with relevant instructions.
The supplemental survey followed the same procedures as the main study survey, except each
participant was incentivized with a $15 downloadable gift card.
Measurement
This chapter focuses on comparing both samples by demographics, substance use,
psychosocial covariates of substance use, and sexual risk. Respondents reported their age, sexual
identity, race and ethnicity, relationship status, highest education level (i.e., less than high
school, some college or university, 4-year college or university, master’s or professional degree,
doctorate), employment status (i.e., full-time, part-time, and not employed), household income,
GSNA use in lifetime and prior month, and monthly frequency of attending gay venues.
Participants also answered items about to whom they had revealed their sexual identity status
(e.g., parents, straight friends, LGBTQ friends). Additionally, level of connection with the gay
community was assessed using the 8-item Identification with the Gay Community Scale
(Vanable, McKirnan, & Stokes, 1998),

which has shown to have an acceptable level of internal
consistency (α = .67; Fernández et al., 2007; Holloway et al., 2012).
Substance use was measured with a binary lifetime item and the number of times used
during the previous 30 days (i.e., 0, 1–2, 3–9, 10–19, 20–39, 40 or more) for each substance. In
addition to alcohol consumption, binge drinking (i.e., five or more drinks in 2 hours), and
marijuana use, substance use of three classes of substances were assessed: illicit (i.e.,
methamphetamine, cocaine, heroin, ecstasy, poppers, and LSD), prescription (i.e., pain pills,
tranquillizers, stimulants, muscle relaxers, erectile dysfunction drugs, and steroids), and
emerging (i.e., bath salts, salvias, inhalants, and other synthetics). For the purposes of analysis,
SUBSTANCE USE AND TECHNOLOGY 129
all substance use items for monthly use were dichotomized to use or no use, except for number
of days consuming alcohol and number of days binge drinking.
The psychosocial risk factors measured in this study were based on previous research
(i.e., experiences of homophobia, sensation seeking, depression, internalized homophobia; Kelly
et al., 2015; Kipke, Weiss, Ramirez, et al., 2007; Kipke, Weiss, & Wong, 2007; Mutchler et al.,
2011; Operario et al., 2006; Thiede et al., 2003; Traube et al., 2013). Experiences of homophobia
were assessed using the 6-item Lifetime Experience of Homophobia questionnaire, which has
previously been used with similar samples and had acceptable reliability (α = .69; Choi, Hudes,
& Steward, 2008). The Brief Sensation Seeking Inventory was used to assess sensation seeking
(Hoyle, Stephenson, Palmgreen, Lorch, & Donohew, 2002). This inventory was used with
previous samples of men who have sex with men and had a moderate level of reliability, with a
Cronbach’s alpha ranging from .63 to .68 (Hoyle et al., 2002; Newcomb, Clerkin, & Mustanski,
2011). The 4-item Center for Epidemiologic Studies Depression Scale was used to measure the
level of depression symptomology (Melchior, Huba, Brown, & Reback, 1993). When used with a
YMSM population, this scale exhibited a high level of internal consistency (α = .83; Gibbs &
Rice, 2016). To assess the level of internalized stigma toward a sexual minority identity (i.e.,
internalized homophobia), the 9-item Internalized Homophobia Scale was used (Herek, Cogan,
Gillis, & Glunt, 1998).
Sexual risk items focused on unprotected sex during last sexual encounter (i.e., receptive
anal sex without a condom, insertive anal sex without a condom, and vaginal sex without a
condom), substance use during last sexual encounter, and lifetime HIV testing. Sexual risk items
were dichotomized (e.g., at least one instance of unprotected sex during last sexual encounter,
being under the influence of a substance during last encounter, and ever having had an HIV test).
SUBSTANCE USE AND TECHNOLOGY 130
Analytic Procedure
Analysis was conducted in SPSS version 24 by first comparing the demographic
characteristics, substance use, substance use determinants, and sexual risk characteristics of the
two samples (i.e., t-tests, chi-square analyses). Second, bivariate correlations were completed to
identify relationships between substance use covariates (internalized stigma, depression
symptomology, lifetime experiences of homophobia, sensation seeking, gay community
connection) and monthly substance use (days consuming alcohol, days binge drinking, marijuana
use, and use of any other substance other than marijuana or alcohol). Significant and trending
toward significant correlations were investigated via regressions to examine the association
between each psychosocial covariate and substance use variable, while including recruitment
type as a moderator. A power analysis was conducted to determine the appropriate sample size
needed to investigate differences in substance use between the two subsamples. Given the
current full sample size, a power analysis indicated that the study was powered to identify
differences in use of marijuana, methamphetamine, and poppers.
Results
Both samples were an average of 21 years of age, and no significant difference was found
between the subsamples (see Table 5.1). The app sample largely identified as Latino (47%), with
minorities of Black (16%), multiracial (14%), White (9%), and Asian (14%) participants. A chi-
square analysis indicated a significant difference (χ
2
[4, 108] = 20.335, p < .001) in race and
ethnicity between the two subsamples; the venue sample featured more participants who
identified as White (42%), followed by Latino (31%), multiracial (16%), Asian (9%), and Black
(1%) participants. Most participants reported a sexual orientation of gay—72% of the app sample
and 87% of the venue sample—and a chi-square test indicated a trend toward significance (χ
2
[1,
SUBSTANCE USE AND TECHNOLOGY 131
111] = 3.697, p = .055). Additionally, the subsamples were found to significantly differ in
employment status (i.e., full time or part time) and highest education level (i.e., at least some
college; see Table 5.1). No significant difference was indicated for relationship status or
household income. Participants were asked to report if they were open (i.e., “out”) about their
sexual identity or behavior with their parents, straight friends, and LGBTQ friends. The
subsamples did not differ in the proportion of participants who were out to each group, except
for a significant difference (χ
2
[1, 111] = 4.276, p < .05) in being out to all parents (37% of the
app sample vs. 57% of the venue sample).
Regarding involvement with gay apps and venues, the two samples significantly differed
on two items. Participants were asked to report if they had used a GSNA in the last 30 days;
100% of the app sample reported using a GSNA compared to two thirds of the venue sample,
indicating a significant difference (χ
2
[1, 105] = 15.827, p < .001). Additionally, participants
indicated their frequency of going to a gay bar or club (i.e., never, once a month or less, several
times a month, about once a week, and several times a week or every day). A chi-square analysis
indicated a significant difference (χ
2
[1, 106] = 16.552, p < .001) in monthly attendance at gay
venues between the two subsamples, with 16% of the app sample and 55% of the venue sample
reporting attending venues several times or more a month.
Table 5.1 also provides statistics on the sexual risk behaviors of the sample. No
significant difference was found between the subsamples regarding the proportion of men
reporting unprotected sex or being under the influence of a substance during their last sexual
encounter. Additionally, participants were asked to indicate if they had ever had an HIV test, and
analyses indicated no significant difference between the samples.
SUBSTANCE USE AND TECHNOLOGY 132
Covariates of substance use were also compared between the samples. Five 2-sample t-
tests indicated no significant differences in internalized stigma, depression symptoms, lifetime
experience of homophobia, sensation seeking, or gay community connection. Table 5.2
summarizes these nonsignificant results.
Lifetime substance use was also investigated. Participants were asked to report on their
lifetime use of 19 substances (see Table 5.3). Chi-square tests indicated four significant
differences between the samples and one trend toward significance. For each significant
difference, a higher proportion of the venue-recruited sample reported use of that substance. The
samples significantly differed (χ
2
[1, 107] = 7.592, p < .01) on lifetime marijuana use, with 76%
of the venue sample and 51% of the app sample reporting use. Lifetime use of cocaine was also
significantly higher in the venue sample (χ
2
[1, 107] = 4.111, p < .05); 34% of venue-recruited
respondents reported use of cocaine compared to 16% of the app-recruited sample. Participants
also reported using prescription medications that were not prescribed to them, including pain
pills (χ
2
[1, 107] = 6.739, p < .01), with 19% of the venue sample and 2% of the app sample
reporting lifetime use. Regarding stimulants, significantly (χ
2
[1, 107] = 5.087, p < .05) more
venue-recruited participants (24%) also reported lifetime use compared to the app-recruited
sample (7%). One trend toward significance (χ
2
[1, 107] = 2.762, p = .097) emerged, with 15% of
the venue sample and 5% of the app sample reporting lifetime use of muscle relaxers.
Past 30-day substance use differences were investigated for six of the substance use
variables (i.e., number of days of alcohol consumption, number of days of binge drinking,
marijuana, cocaine, ecstasy, poppers) and any substance use (other than alcohol and marijuana;
see Table 5.4). App-recruited participants reported lower percentages of 30-day use for all
substances tested. Alcohol use was measured by the reported number of days using alcohol and
SUBSTANCE USE AND TECHNOLOGY 133
binge drinking. Chi-square analyses were restricted to variables with an expected cell count of
five in each cell; therefore, five chi-square analyses were conducted (i.e., marijuana, cocaine,
ecstasy, poppers, and any substance other than alcohol or marijuana). Three statistical
differences and one trend toward significance emerged. A two-sample t-test indicated that the
venue sample (M =8.238, SD = 6.734) consumed alcohol on significantly more days each month
(t[95.73] = 2.968, p < .01) compared to the app sample (M = 4.670, SD = 5.370). A significant
difference (χ
2
[1, 107 = 4.410, p < .05) between subsamples was found in the percentage of
participants reporting using poppers in the last 30 days, with 22% of the venue sample and 7% of
the app sample reporting use. Further investigation revealed a significant difference (χ
2
[1, 107] =
3.886, p < .05) between the two samples regarding using any substance (other than marijuana
and alcohol) in the last 30 days; 44% of the venue sample reported substance use, compared to
26% of the app sample. A chi-square test indicated a trend toward significance (χ
2
[1, 107] =
3.384, p = .066) between the proportions of each sample reporting cocaine use, with 16% of the
venue sample and 5% of the app sample reporting use.
Bivariate correlations (see Table 5.5) identified nine significant or trending toward
significant relationships between psychosocial covariates and substance use. Six linear
regression and three binary logistic moderation analyses were conducted to investigate
recruitment type as a moderator in the relationships between substance covariates and the four
different substance use variables (see Table 5.6). One significant finding and one trend toward
significance in moderation emerged. Recruitment type significantly moderated the relationship
between gay community connection and number of days binge drinking in the last month. At low
levels of gay community connection (i.e. 16.71, 1 standard deviation below the mean of 21.33),
app-recruited individuals on average would be expected to report approximately 4.5 days of
SUBSTANCE USE AND TECHNOLOGY 134
binge drinking, whereas at high levels of gay community connection (i.e., 25.96, 1 standard
deviation above the mean of 21.33), app-recruited individuals would be expected to report less
than 1 day of binge drinking (see Figure 5.1), indicating a negative relationship between binge
drinking and gay community connection in the app sample. However, for the venue-recruited
sample, the relationship between gay community connection and binge drinking was positive. At
low levels of gay community connection, the venue sample would be expected to report
approximately 2 days of binge drinking, compared to 4 days of binge drinking at high levels of
gay community connection. A trend toward significance indicated a similar relationship between
gay community connection and number of days drinking alcohol (see Figure 5.2). At low levels
of gay community connection (i.e., 16.71), app-recruited men and venue-recruited men would be
expected to report a similar number of days of alcohol use: 5 days and 6 days, respectively.
However, at high levels of gay community connection (i.e., 25.96), app-recruited men would be
expected to report fewer days drinking alcohol (approximately 4 days) compared to venue-
recruited men (10 days).
Discussion
The aim of this chapter was to investigate potential differences in demographics,
involvement in the gay community, sex risk, substance use covariates, and substance use
between two probability samples of YMSM. Findings indicate that (a) recruitment methodology
has an impact on demographic diversity, (b) app-recruited men do not report higher rates of
sexual risk, (c) sampling procedures influence reported rates of substance use, and (d) the
relationship between covariates of substance use (e.g., gay community connection) and substance
use may differ based on where YMSM are recruited to participate in research.
SUBSTANCE USE AND TECHNOLOGY 135
The app-recruited sample was more diverse in terms of race and ethnicity (i.e., lower
proportion identifying as non-Hispanic White), sexual orientation, education, and employment
status. Additionally, a lower proportion of app-recruited men reported being out about their
sexual behavior to all of their parents. Because apps support anonymity by allowing users to
limit the amount of personal information shared, YMSM who use these technologies have more
control in advertising and disclosing their sexuality compared to attending a public gay venue.
Only 16% of the app sample reported regular attendance at gay venues. Despite lower
engagement with gay venues, which has historically been a crude measure of connection to the
gay community (Vanable et al., 1998), analyses indicated no difference in gay community
connection between the two samples. Therefore, GSNAs may allow sexual minority men who
prefer more privacy to continue to feel connected to the gay community.
Findings indicate that although the men recruited through app-based probability sampling
appeared to be more diverse, less out about their identity or sexual behavior, and less engaged in
gay venues, they reported no difference in sexual risk. Previous literature has found that using
apps is a risk for unprotected sex; however, these men were recruited from gay venues and then
stratified by app use (Eaton et al., 2016; Lehmiller & Ioerger, 2014; Phillips et al., 2014). When
individuals are recruited in person from gay venues, the app-using men identified in these studies
represent men who use both venues and apps. Identifying these men as being examples of app
users misrepresents their behaviors. Individuals who use both venues and apps may have a
different risk profile compared to only app users. Further, venue-based studies systematically
exclude YMSM who do not attend venues on a regular basis and instead use apps to find sex
partners and connect with the gay community. Based on the findings from the current study,
approximately 84% of app-using YMSM would be systematically excluded from venue-based
SUBSTANCE USE AND TECHNOLOGY 136
research, whereas approximately two thirds of venue-recruited men could be recruited via apps,
which is comparable to previous literature (Phillips et al., 2014). Where YMSM are found,
whether digitally or in person, should be a primary consideration in decisions regarding
recruitment methodology.
Because this is the first known study to compare a venue-based probability sample with a
GSNA-based probability sample, significant differences in substance use indeed suggest venue-
based recruitment procedures may artificially inflate substance use among YMSM. A
significantly larger proportion of the venue sample reported ever using marijuana, cocaine, pain
medication, and stimulants. The YMSM recruited through venues also reported almost twice the
number of days consuming alcohol during the past month compared to the app sample. Results
suggest that recruiting YMSM from locations that promote alcohol and substance use may have a
notable impact on the risk profiles of the men recruited, providing evidence that recruitment
methods can have a considerable impact on the rates of substance use reported among YMSM.
Choosing to recruit exclusively from locations where higher rates of substance use are expected
could inflate substance use statistical findings and potentially further stigmatize YMSM.
Understanding differences in substance use covariates is essential for designing and
implementing interventions to reduce substance misuse among YMSM (Kurtz et al., 2013).
Although no differences in substance use covariates (i.e., internalized stigma, experiences of
homophobia, depression symptomology, connection with the gay community, and sensation
seeking) emerged between the samples, several of the relationships between the psychosocial
covariates and substance use differed across the samples. Gay community connection is a
measure of a cognitive and enacted connection to a corporate sexual minority male community
(Gibbs & Rice, 2016). This connection has been found to protect against depression
SUBSTANCE USE AND TECHNOLOGY 137
symptomology (Gibbs & Rice, 2016), but previous literature has indicated a positive relationship
between gay community connection and alcohol use (Newcomb et al., 2011), which the current
study confirmed with regard to the venue-based sample. However, the opposite relationship was
found with the app-based sample. We found that being more connected to the gay community
was associated with lower rates of 30-day alcohol consumption and binge drinking in the app
sample. We hypothesize several possible reasons for this finding. First, results indicate that the
venue-recruited sample attended gay bars and clubs more often than the app-recruited sample.
Because these venues promote use of alcohol, it is logical that the more connected the venue-
recruited sample is to the gay community (i.e., socializing in gay bars and clubs), the more these
men are concurrently drinking alcohol. However, this relationship was not as clear for the app-
recruited sample. In a 2016 study, Feinstein and Newcomb investigated the substance use
motives of YMSM to understand how they might affect substance use (e.g., to cope with stress
or enhance an experience). Findings indicated that both motives have a significant impact on
alcohol use (Feinstein & Newcomb, 2016). For the app-recruited sample in the current study, it
could be that low levels of gay community connection mean that these men are more isolated and
therefore use alcohol to cope. Notably, the app-based sample’s connection with the gay
community protected against alcohol use. This finding suggests that app-using YMSM
experience a different type of connection with the gay community that is separate from attending
gay bars and clubs. Because the scale for gay community connection used in this study was
created prior to the introduction of GSNAs, the construct of gay community connection may
need to be revisited and reconstructed (Vanable et al., 1998).
There are several study limitations to note. The sample in this study was recruited from a
large metropolitan area, and therefore the results may not be generalizable to YMSM living in
SUBSTANCE USE AND TECHNOLOGY 138
rural areas. Additionally, the methods in this study used one GSNA for recruitment. Because
GSNAs for sexual minority men may cater to different populations (e.g., people of color, older
adults), there may be demographic differences among men who use different GSNAs
(Mastroyiannis, 2018). However, the population of Los Angeles is majority Latino or Hispanic
(U.S. Census Bureau, 2016), and the study GSNA, Hornet, was selected because the community
advisory board identified it as an app heavily used by Latino men. Despite the informed selection
of the study GSNA, replicating the present study with an alternate GSNA for recruitment could
yield different results. This study is also limited by statistical power, because the sample size was
chosen to identify differences in three specific substances (i.e., marijuana, methamphetamines,
and poppers). Despite the limited sample size, statistical differences emerged in several
substances. Additionally, future studies can use the presented statistics to determine the power
needed to find statistically significant differences between venue-based and app-based samples.
Although this study has several limitations, the methods developed and implemented in
the study, in addition to the findings, have several implications. First, GSNA-based probability
sampling can allow research on YMSM to be expanded outside of urban centers. Sexual minority
men in rural areas of the United States and in developing countries, where same-sex behavior is
often criminalized or heavily stigmatized, have limited access to venues. The novel recruitment
methods presented in this study can make these populations of YMSM accessible for research.
Therefore, future research using GSNA-based probability sampling can investigate the
psychosocial covariates of behavioral health issues in YMSM living in remote or service-
impoverished areas. Although the methods in this study were used for cross-sectional study
recruitment, these methods can also be used in longitudinal and intervention research recruitment
and implementation.
SUBSTANCE USE AND TECHNOLOGY 139
As noted, the construct of gay community connection had a different impact on alcohol
consumption for an app-recruited sample. Further research is needed to understand gay
community connection for YMSM who engage digitally with the gay community. The findings
from this study also inform recruitment with YMSM. Although a venue-recruited sample may
present with increased substance use, an app-recruited sample might present with more diversity
and limited outness. Due to the nature of this study, conclusions cannot be drawn regarding the
generalizability of either sample. Future research should compare national samples of men drawn
from either recruitment strategy to statistics in nationally representative samples (e.g., Youth
Risk Behavior Survey, National Longitudinal Study of Adolescent to Adult Health) and weight
the data accordingly. Although the GSNA-based methods used in this study are replicable at a
national level, venue-based probability sampling procedures applied at a national level would
involve a significant cost and be difficult to implement (Meyer & Wilson, 2009).
With the introduction of GSNAs, how YMSM interact and advertise their connection to
the gay community has changed. For research methods to remain relevant, they must evolve and
adapt to this change. The methods and findings presented in this study encourage this evolution.

SUBSTANCE USE AND TECHNOLOGY 140
References
Beymer, M. R., Weiss, R. E., Bolan, R. K., Rudy, E. T., Bourque, L. B., Rodriguez, J. P., &
Morisky, D. E. (2014). Sex on demand: Geosocial networking phone apps and risk of
sexually transmitted infections among a cross-sectional sample of men who have sex with
men in Los Angeles County. Sexually Transmitted Infections, 90, 567–572.
doi:10.1136/sextrans-2013-051494
Boyd, C. J., McCabe, S. E., & d’Arcy, H. (2003). Ecstasy use among college undergraduates:
Gender, race and sexual identity. Journal of Substance Abuse Treatment, 24, 209–215.
doi:10.1016/S0740-5472(03)00025-4
Buckingham, L., Becher, J., Voytek, C. D., Fiore, D., Dunbar, D., Davis-Vogel, A., … Frank, I.
(2017). Going social: Success in online recruitment of men who have sex with men for
prevention HIV vaccine research. Vaccine, 35, 3498–3505.
doi:10.1016/j.vaccine.2017.05.002
Burrell, E. R., Pines, H. A., Robbie, E., Coleman, L., Murphy, R. D., Hess, K. L., … Gorbach, P.
M. (2012). Use of the location-based social networking application Grindr as a
recruitment tool in rectal microbicide development research. AIDS and Behavior, 16,
1816–1820. doi:10.1007/s10461-012-0277-z
Choi, K. H., Hudes, E. S., & Steward, W. T. (2008). Social discrimination, concurrent sexual
partnerships, and HIV risk among men who have sex with men in Shanghai, China. AIDS
and Behavior, 12, 71–77. doi:10.1007/s10461-008-9394-0
Choi, E. P. H., Wong, J. Y. H., & Fong, D. Y. T. (2017). The use of social networking
applications of smartphone and associated sexual risks in lesbian, gay, bisexual, and
SUBSTANCE USE AND TECHNOLOGY 141
transgender populations: A systematic review. AIDS care, 29, 145–155.
doi.org/10.1080/09540121.2016.1211606
Colfax, G., Vittinghoff, E., Husnik, M. J., McKirnan, D., Buchbinder, S., Koblin, B., …
EXPLORE Study Team. (2004). Substance use and sexual risk: A participant- and
episode-level analysis among a cohort of men who have sex with men. American Journal
of Epidemiology, 159, 1002–1012. doi:10.1093/aje/kwh135
Delaney, K. P., Kramer, M. R., Waller, L. A., Flanders, W. D., & Sullivan, P. S. (2014). Using a
geolocation social networking application to calculate the population density of sex-
seeking gay men for research and prevention services. Journal of Medical Internet
Research, 16, e249. doi:10.2196/jmir.3523
Duncan, D. T., Goedel, W. C., Stults, C. B., Brady, W. J., Brooks, F. A., Blakely, J. S., & Hagen,
D. (2018). A study of intimate partner violence, substance abuse, and sexual risk
behaviors among gay, bisexual, and other men who have sex with men in a sample of
geosocial-networking smartphone application users. American Journal of Men’s Health,
12, 292–301. doi:10.1177/1557988316631964
Duncan, D. T., Hatzenbuehler, M. L., & Johnson, R. M. (2014). Neighborhood-level LGBT hate
crimes and current illicit drug use among sexual minority youth. Drug and Alcohol
Dependence, 135, 65–70. doi:10.1016/j.drugalcdep.2013.11.001
Eaton, L. A., Maksut, J. L., Gamarel, K. E., Siembida, E. J., Driffin, D. D., & Baldwin, R.
(2016). Online sex partner meeting venues as a risk factor for testing HIV positive among
a community-based sample of black men who have sex with men. Sexually transmitted
diseases, 43, 360–364. doi: 10.1097/OLQ.0000000000000454
SUBSTANCE USE AND TECHNOLOGY 142
Feinstein, B. A., & Newcomb, M. E. (2016). The role of substance use motives in the
associations between minority stressors and substance use problems among young men
who have sex with men. Psychology of Sexual Orientation and Gender Diversity, 3, 357–
366. doi:10.1037/sgd0000185
Fernández, M. I., Warren, J. C., Varga, L. M., Prado, G., Hernandez, N., & Bowen, G. S. (2007).
Cruising in cyber space: Comparing Internet chat room versus community venues for
recruiting Hispanic men who have sex with men to participate in prevention studies.
Journal of Ethnicity in Substance Abuse, 6, 143–162. doi:10.1300/J233v06n02_09
Ford, W. L., Weiss, G., Kipke, M. D., Ritt-Olson, A., Iverson, E., & Lopez, D. (2009). The
Healthy Young Men’s Study: Sampling methods to recruit a random cohort of young
men who have sex with men. Journal of Gay & Lesbian Social Services, 21, 357–373.
doi:10.1080/10538720802498280
Gibbs, J. J., & Rice, E. (2016). The social context of depression symptomology in sexual
minority male youth: Determinants of depression in a sample of Grindr users. Journal of
Homosexuality, 63, 278–299. doi:10.1080/00918369.2015.1083773
Goedel, W. C., & Duncan, D. T. (2015). Geosocial-networking app usage patterns of gay,
bisexual, and other men who have sex with men: Survey among users of Grindr, a mobile
dating app. JMIR Public Health and Surveillance, 1, e4. doi:10.2196/publichealth.4353
Goedel, W. C., & Duncan, D. T. (2016). Contextual factors in geosocial-networking smartphone
application use and engagement in condomless anal intercourse among gay, bisexual, and
other men who have sex with men who use Grindr. Sexual Health, 13, 549–554.
doi:10.1071/SH16008
SUBSTANCE USE AND TECHNOLOGY 143
Goedel, W. C., Hagen, D., Halkitis, P. N., Greene, R. E., Griffin-Tomas, M., Brooks, F. A., …
Duncan, D. T. (2017). Post-exposure prophylaxis awareness and use among men who
have sex with men in London who use geosocial-networking smartphone applications.
AIDS Care, 29, 579–586. doi:10.1080/09540121.2016.1259455
Goedel, W. C., Halkitis, P. N., & Duncan, D. T. (2016). Behavior- and partner-based HIV risk
perception and sexual risk behaviors in men who have sex with men (MSM) who use
geosocial-networking smartphone applications in New York City. Journal of Urban
Health, 93, 400–406. doi:10.1007/s11524-016-0043-z
Goedel, W. C., Safren, S. A., Mayer, K. H., & Duncan, D. T. (2017). Community-level norms
and condomless anal intercourse among gay, bisexual, and other men who have sex with
men who use geosocial-networking smartphone applications in the Deep South. Journal
of HIV/AIDS & Social Services, 16, 382–389. doi:10.1080/15381501.2017.1344599
Goldberg, S., Strutz, K. L., Herring, A. A., & Halpern, C. T. (2013). Risk of substance abuse and
dependence among young adult sexual minority groups using a multidimensional
measure of sexual orientation. Public Health Reports, 128, 144–152.
doi:10.1177/003335491312800304
Grov, C. (2012). HIV risk and substance use in men who have sex with men surveyed in
bathhouses, bars/clubs, and on Craigslist.org: Venue of recruitment matters. AIDS and
Behavior, 16, 807–817. doi:10.1007/s10461-011-9999-6
Grov, C., & Crow, T. (2012). Attitudes about and HIV risk related to the “most common place”
MSM meet their sex partners: Comparing men from bathhouses, bars/clubs, and
Craigslist.org. AIDS Education and Prevention, 24, 102–116.
doi:10.1521/aeap.2012.24.2.102
SUBSTANCE USE AND TECHNOLOGY 144
Hagger-Johnson, G., Taibjee, R., Semlyen, J., Fitchie, I., Fish, J., Meads, C., & Varney, J.
(2013). Sexual orientation identity in relation to smoking history and alcohol use at age
18/19: Cross-sectional associations from the Longitudinal Study of Young People in
England (LSYPE). BMJ Open, 3, e002810. doi:10.1136/bmjopen-2013-002810
Hatzenbuehler, M. L., Corbin, W. R., & Fromme, K. (2008). Trajectories and determinants of
alcohol use among LGB young adults and their heterosexual peers: Results from a
prospective study. Developmental Psychology, 44, 81–90. doi:10.1037/0012-
1649.44.1.81
Herek, G. M., Cogan, J. C., Gillis, J. R., & Glunt, E. K. (1998). Correlates of internalized
homophobia in a community sample of lesbians and gay men. Journal of the Gay &
Lesbian Medical Association, 2, 17–25.
Hess, K. L., Chavez, P. R., Kanny, D., DiNenno, E., Lansky, A., & Paz-Bailey, G. (2015). Binge
drinking and risky sexual behavior among HIV-negative and unknown HIV status men
who have sex with men, 20 US cities. Drug and Alcohol Dependence, 147, 46–52.
doi:10.1016/j.drugalcdep.2014.12.013
Holloway, I. W., Pulsipher, C. A., Gibbs, J., Barman-Adhikari, A., & Rice, E. (2015). Network
influences on HIV risk behaviors of gay, bisexual and other men who have sex with men
using geosocial networking applications. AIDS and Behavior, 19(Suppl. 2), 112–122.
doi:10.1007/s10461-014-0989-3
Holloway, I. W., Rice, E., Gibbs, J., Winetrobe, H., Dunlap, S., & Rhoades, H. (2014).
Acceptability of smartphone application-based HIV prevention for among young men
who have sex with men. AIDS and Behavior, 18, 285–296. doi:10.1007/s10461-013-
0671-1
SUBSTANCE USE AND TECHNOLOGY 145
Holloway, I. W., Tan, D., Gildner, J. L., Beougher, S. C., Pulsipher, C., Montoya, J. A., …
Leibowitz, A. (2017). Facilitators and barriers to pre-exposure prophylaxis willingness
among young men who have sex with men who use geosocial networking applications in
California. AIDS Patient Care and STDs, 31, 517–527. doi:10.1089/apc.2017.0082
Holloway, I. W., Traube, D. E., Rice, E., Schrager, S. M., Palinkas, L. A., Richardson, J., &
Kipke, M. D. (2012). Community and individual factors associated with cigarette
smoking among young men who have sex with men. Journal of Research on
Adolescence, 22, 199–205. doi:10.1111/j.1532-7795.2011.00774.x
Hornet. (2018). About us. Retrieved from https://love.hornet.com/about/
Hoyle, R. H., Stephenson, M. T., Palmgreen, P., Lorch, E. P., & Donohew, R. L. (2002).
Reliability and validity of a brief measure of sensation seeking. Personality and
Individual Differences, 32, 401–414. doi:10.1016/S0191-8869(01)00032-0
Kelly, J., Davis, C., & Schlesinger, C. (2015). Substance use by same sex attracted young
people: Prevalence, perceptions and homophobia. Drug and Alcohol Review, 34, 358–
365. doi:10.1111/dar.12158
Kipke, M. D., Weiss, G., Ramirez, M., Dorey, F., Ritt-Olson, A., Iverson, E., & Ford, W. (2007).
Club drug use in Los Angeles among young men who have sex with men. Substance Use
& Misuse, 42, 1723–1743. doi:10.1080/10826080701212261
Kipke, M. D., Weiss, G., & Wong, C. F. (2007). Residential status as a risk factor for drug use
and HIV risk among young men who have sex with men. AIDS and Behavior, 11(Suppl.
2), 56–69. doi:10.1007/s10461-006-9204-5
SUBSTANCE USE AND TECHNOLOGY 146
Kurtz, S. P., Stall, R. D., Buttram, M. E., Surratt, H. L., & Chen, M. (2013). A randomized trial
of a behavioral intervention for high risk substance-using MSM. AIDS and Behavior, 17,
2914–2926. doi:10.1007/s10461-013-0531-z
Lachowsky, N. J., Lal, A., Forrest, J. I., Card, K. G., Cui, Z., Sereda, P., … Hogg, R. S. (2016).
Including online-recruited seeds: A respondent-driven sample of men who have sex with
men. Journal of Medical Internet Research, 18, e51. doi:10.2196/jmir.5258
Lehmiller, J. J., & Ioerger, M. (2014). Social networking smartphone applications and sexual
health outcomes among men who have sex with men. PLOS One, 9, e86603.
doi:10.1371/journal.pone.0086603
Leslie, J. (2009, October 1). Grindr app makes it easy to be easy. Out & About Nashville.
Retrieved from https://www.outandaboutnashville.com/story/grindr-app-makes-it-easy-
be-easy
Lorimer, K., Flowers, P., Davis, M., & Frankis, J. (2016). Young men who have sex with men’s
use of social and sexual media and sex-risk associations: Cross-sectional, online survey
across four countries. Sexually Transmitted Infections, 92, 371–376.
doi:10.1136/sextrans-2015-052209
Macapagal, K., Coventry, R., Puckett, J. A., Phillips, G., II, & Mustanski, B. (2016). Geosocial
networking app use among men who have sex with men in serious romantic relationships.
Archives of Sexual Behavior, 45, 1513–1524. doi:10.1007/s10508-016-0698-2
MacKellar, D. A., Gallagher, K. M., Finlayson, T., Sanchez, T., Lansky, A., & Sullivan, P. S.
(2007). Surveillance of HIV risk and prevention behaviors of men who have sex with
men—a national application of venue-based, time-space sampling. Public Health
Reports, 122(1, Suppl.), 39–47. doi:10.1177/00333549071220S107
SUBSTANCE USE AND TECHNOLOGY 147
Mastroyiannis, A. (March 5, 2018). Gay dating apps: A comprehensive guide to Jack’d, Ginrd,
Hornet, Scruff, and the rest. Pink News. Retreived from:
https://www.pinknews.co.uk/2018/03/05/best-gay-dating-apps-jackd-grindr-hornet-
scruff/
McCabe, S. E., Hughes, T. L., Bostwick, W. B., West, B. T., & Boyd, C. J. (2009). Sexual
orientation, substance use behaviors and substance dependence in the United States.
Addiction, 104, 1333–1345. doi:10.1111/j.1360-0443.2009.02596.x
McCabe, S. E., Bostwick, W. B., Hughes, T. L., West, B. T., & Boyd, C. J. (2010). The
relationship between discrimination and substance use disorders among lesbian, gay, and
bisexual adults in the United States. American Journal of Public Health, 100, 1946–1952.
doi:10.2105/AJPH.2009.163147
McCabe, S. E., Boyd, C., Hughes, T. L., & d’Arcy, H. (2003). Sexual identity and substance use
among undergraduate students. Substance Abuse, 24, 77–91.
doi:10.1023/A:1023768215020
Melchior, L. A., Huba, G. J., Brown, V. B., & Reback, C. J. (1993). A short depression index for
women. Educational and Psychological Measurement, 53, 1117–1125.
doi:10.1177/0013164493053004024
Meyer, I. H., & Wilson, P. A. (2009). Sampling lesbian, gay, and bisexual populations. Journal
of Counseling Psychology, 56, 23–31. doi:10.1037/a0014587
Moeller, R. W., Palamar, J. J., Halkitis, P. N., & Siconolfi, D. E. (2014). An episodic analysis of
substance use and risky sexual behavior in a racially diverse sample of young men who
have sex with men. Journal of Gay & Lesbian Social Services, 26, 168–185.
doi:10.1080/10538720.2014.891450
SUBSTANCE USE AND TECHNOLOGY 148
Mutchler, M. G., McKay, T., Candelario, N., Liu, H., Stackhouse, B., Bingham, T., & Ayala, G.
(2011). Sex drugs, peer connections, and HIV: Use and risk among African-American,
Latino, and multiracial young men who have sex with men in Los Angeles and New
York. Journal of Gay & Lesbian Social Services, 23, 271–295.
doi:10.1080/10538720.2011.560100
National Institute on Drug Abuse. (2012). Health consequences of drug abuse. Retrieved from
http://www.drugabuse.gov/related-topics/medical-consequences-drug-abuse
Newcomb, M. E., Clerkin, E. M., & Mustanski, B. (2011). Sensation seeking moderates the
effects of alcohol and drug use prior to sex on sexual risk in young men who have sex
with men. AIDS and Behavior, 15, 565–575. doi:10.1007/s10461-010-9832-7
Operario, D., Choi, K.-H., Chu, P. L., McFarland, W., Secura, G. M., Behel, S., … Valleroy, L.
(2006). Prevalence and correlates of substance use among young Asian Pacific Islander
men who have sex with men. Prevention Science, 7, 19–29. doi:10.1007/s11121-005-
0018-x
Parsons, J. T., Lelutiu-Weinberger, C., Botsko, M., & Golub, S. A. (2013). Predictors of day-
level sexual risk for young gay and bisexual men. AIDS and Behavior, 17, 1465–1477.
doi:10.1007/s10461-012-0206-1
Pew Research Center. (2018). Mobile fact sheet: Who owns cellphones and smartphones.
Retrieved from http://www.pewinternet.org/fact-sheet/mobile/
Phillips, G., II, Magnus, M., Kuo, I., Rawls, A., Peterson, J., Jia, Y., … Greenberg, A. E. (2014).
Use of geosocial networking (GSN) mobile phone applications to find men for sex by
men who have sex with men (MSM) in Washington, DC. AIDS and Behavior, 18, 1630–
1637. doi:10.1007/s10461-014-0760-9
SUBSTANCE USE AND TECHNOLOGY 149
Puckett, J. A., Newcomb, M. E., Garofalo, R., & Mustanski, B. (2017). Examining the conditions
under which internalized homophobia is associated with substance use and condomless
sex in young MSM: The moderating role of impulsivity. Annals of Behavioral Medicine,
51, 567–577. doi:10.1007/s12160-017-9878-0
Rendina, H. J., Jimenez, R. H., Grov, C., Ventuneac, A., & Parsons, J. T. (2014). Patterns of
lifetime and recent HIV testing among men who have sex with men in New York City
who use Grindr. AIDS and Behavior, 18, 41–49. doi:10.1007/s10461-013-0573-2
Rice, E., Holloway, I., Winetrobe, H., Rhoades, H., Barman-Adhikari, A., Gibbs, J., … Dunlap,
S. (2012). Sex risk among young men who have sex with men who use Grindr, a
smartphone geosocial networking application. Journal of AIDS & Clinical Research, S4,
005. doi:10.4172/2155-6113.S4-005
Rosario, M., Schrimshaw, E. W., & Hunter, J. (2006). A model of sexual risk behaviors among
young gay and bisexual men: Longitudinal associations of mental health, substance
abuse, sexual abuse, and the coming-out process. AIDS Education and Prevention, 18,
444–460. doi:10.1521/aeap.2006.18.5.444
Ryan, C., Huebner, D., Diaz, R. M., & Sanchez, J. (2009). Family rejection as a predictor of
negative health outcomes in White and Latino lesbian, gay, and bisexual young adults.
Pediatrics, 123, 346–352. doi:10.1542/peds.2007-3524
Shoptaw, S., Weiss, R. E., Munjas, B., Hucks-Ortiz, C., Young, S. D., Larkins, S., … Gorbach,
P. M. (2009). Homonegativity, substance use, sexual risk behaviors, and HIV status in
poor and ethnic men who have sex with men in Los Angeles. Journal of Urban Health,
86(Suppl. 1), 77–92. doi:10.1007/s11524-009-9372-5
SUBSTANCE USE AND TECHNOLOGY 150
Siegler, A. J., Sanchez, T., Sineath, R. C., Grey, J., Kahle, E., & Sullivan, P. S. (2015).
Knowledge and awareness of acute human immunodeficiency virus infection among
mobile app-using men who have sex with men: A missed public health opportunity. Open
Forum Infectious Diseases, 2, ofv016. doi:10.1093/ofid/ofv016
Stall, R., Paul, J. P., Greenwood, G., Pollack, L. M., Bein, E., Crosby, G. M., ... & Catania, J. A.
(2001). Alcohol use, drug use and alcohol-related problems among men who have sex
with men: the Urban Men’s Health Study. Addiction, 96(11), 1589-1601.
Thiede, H., Valleroy, L. A., MacKellar, D. A., Celentano, D. D., Ford, W. L., Hagan, H., …
Torian, L. V. (2003). Regional patterns and correlates of substance use among young
men who have sex with men in 7 US urban areas. American Journal of Public Health, 93,
1915–1921. doi:10.2105/AJPH.93.11.1915
Traube, D. E., Schrager, S. M., Holloway, I. W., Weiss, G., & Kipke, M. D. (2013).
Environmental risk, social cognition, and drug use among young men who have sex with
men: Longitudinal effects of minority status on health processes and outcomes. Drug and
Alcohol Dependence, 127, 1–7. doi:10.1016/j.drugalcdep.2012.06.007
Trocki, K. F., Drabble, L. A., & Midanik, L. T. (2009). Tobacco, marijuana, and sensation
seeking: Comparisons across gay, lesbian, bisexual, and heterosexual groups. Psychology
of Addictive Behaviors, 23, 620–631. doi:10.1037/a0017334
U.S. Census Bureau. (2016). Los Angeles, CA: Race and ethnicity. Retrieved from
https://censusreporter.org/profiles/16000US0644000-los-angeles-ca/
Vanable, P. A., McKirnan, D. J., & Stokes, J. P. (1998). Identification and involvement with the
Gay Community Scale. In C. M. Davis, W. L. Yarber, R. Bauserman, G. Sheer, & S. L.
Davis (Eds.), Handbook of sexuality-related measures (pp. 407–409). Thousand Oaks,
SUBSTANCE USE AND TECHNOLOGY 151
CA: Sage.
Wong, C. F., Weiss, G., Ayala, G., & Kipke, M. D. (2010). Harassment, discrimination, violence
and illicit drug use among young men who have sex with men. AIDS Education and
Prevention, 22, 286–298. doi:10.1521/aeap.2010.22.4.286
Zablotska, I. B., Holt, M., & Prestage, G. (2012). Changes in gay men’s participation in gay
community life: Implications for HIV surveillance and research. AIDS and Behavior, 16,
669–675. doi:10.1007/s10461-011-9919-9

SUBSTANCE USE AND TECHNOLOGY 152
Table 5.1. Demographic Differences between Samples

App Venue

(n = 43) (n = 68)

M or n SD or % M or n SD or % χ
2
or t (df)
Age 21.024 2.089 21.471 1.714 1.220 (108)
Race and ethnicity

African American or Black 7 16.28 1 1.47 20.335***
Hispanic or Latino 20 46.51 21 30.88

White 4 9.30 29 42.65

Asian 6 13.95 6 8.82

Multiracial 6 13.95 11 16.18

Sexual orientation (gay vs. other)

3.697†
Gay 31 72.09 59 86.76

Bisexual 10 23.26 5 7.35

Pansexual 0 0.00 4 5.88

Queer 1 2.33 0 0.00

Other 1 2.33 0 0.00

Education (at least some college) 27 62.79 58 85.29 7.437**
Employed (part or full time) 24 55.81 61 89.71 16.869***
Household income ($50,000 or
above)
7 16.28 17 25.76 1.362
In a relationship 9 20.93 20 29.41 0.982
Out about sexual identity or behavior

To all parents 16 37.21 39 57.35 4.276*
To straight friends 34 79.07 56 82.35 0.185
To LGBTQ friends 37 86.05 58 85.29 0.012
Lifetime GSNA use 41 100.00 66 97.06 0.329
30-day GSNA use 41 100.00 44 66.67 15.827***
Gay venue attendance (several times
a month or more)
7 16.28 37 55.22 16.552***
Never 14 32.56 3 4.48

Once a month or less 22 51.16 27 40.30

Several times a month 5 11.63 20 29.85

About once a week 2 4.65 11 16.42

Several times a week or every day 0 0.00 6 8.96

Sexual risk

Unprotected sex during last
encounter
12 27.91 22 32.35 0.245
Under the influence during last
encounter
10 23.81 13 20.97 0.117
Never had HIV test 4 9.30 8 11.94 0.188
†p < .10. *p < .05. **p < .01. ***p < .001.

SUBSTANCE USE AND TECHNOLOGY 153
Table 5.2. Substance Use Covariate Differences between Samples

App Venue
(n = 39) (n = 68)
M SD M SD t df p
Internalized stigma (homophobia) 19.191 7.385 18.076 7.824 0.738 106 .462
Depressive symptoms 6.954 3.062 7.939 3.450 1.523 107 .131
Lifetime homophobia experiences 12.372 4.123 12.046 3.335 0.455 107 .650
Sensation seeking 28.095 6.092 29.394 5.852 1.107 106 .271
Gay community connection 20.698 3.943 21.761 5.003 1.178 108 .241

SUBSTANCE USE AND TECHNOLOGY 154
Table 5.3. Lifetime Substance Use Differences between Samples

App Venue
(n = 39) (n = 68)

n % n % χ
2

Alcohol 40 93.02 63 92.65 0.006
Bath salts 0 0.00 0 0.00 --
Marijuana 22 51.16 52 76.47 7.592**
Cocaine 7 16.28 23 33.82 4.111*
Ecstasy 9 20.93 18 26.47 0.439
Erectile dysfunction drugs 4 9.30 3 4.41 1.066
Heroin 0 0.00 0 0.00 --
Inhalants 5 11.63 8 11.76 0.000
Ketamine 2 4.65 4 5.88 0.027
LSD 3 6.98 8 11.76 0.676
Methamphetamines (speed) 1 2.33 3 4.41 0.330
Muscle relaxers (e.g., Valium) 2 4.65 10 14.71 2.762†
Psychedelic mushrooms 4 9.30 11 16.18 1.065
Pain pills 1 2.33 13 19.12 6.739**
Poppers (e.g., amyl nitrate) 13 30.23 25 36.76 0.499
Salvias 2 4.65 3 4.41 0.001
Sedatives or sleeping pills 5 11.63 12 17.65 0.736
Steroids 1 2.33 0 0.00 1.596
Stimulants 3 6.98 16 23.53 5.087*
†p < .10. *p < .05. **p < .01.

SUBSTANCE USE AND TECHNOLOGY 155
Table 5.4. Differences in Substance Use during the Prior Month between Samples

App Venue

(n = 39) (n = 68)

M or n SD or % M or n SD or % χ
2
or t (df)
Days consuming alcohol 4.670 5.370 8.238 6.734 2.968 (95.73)**
Days binge drinking 2.900 5.551 3.079 4.081 0.189 (101)
Substance other than marijuana 11 25.58 30 44.12 3.886*
Bath salts 0 0.00 0 0.00 ----
Marijuana 15 34.88 33 48.53 1.999
Cocaine 2 4.65 11 16.18 3.384†
Ecstasy 3 6.98 15 10.00 1.522
Erectile dysfunction drugs 2 4.65 0 0.00 ----
Heroin 0 0.00 0 0.00 ----
Inhalants 1 2.33 1 1.47 ----
Ketamine 1 2.33 0 0.00 ----
LSD 0 0.00 1 1.47 ----
Methamphetamines 0 0.00 0 0.00 ----
Muscle relaxers 0 0.00 1 1.47 ----
Psychedelic mushrooms 1 2.33 1 1.47 ----
Pain pills 0 0.00 1 1.47 ----
Poppers 3 6.98 15 22.06 4.410*
Salvias 0 0.00 0 0.00 ----
Sedatives or sleeping pills 1 2.33 2 2.94 ----
Steroids 1 2.33 0 0.00 ----
Stimulants 0 0.00 3 4.41 ----
†p < .10. *p < .05. **p < .01.

SUBSTANCE USE AND TECHNOLOGY 156
Table 5.5. Correlations between Study Variables

1 2 3 4 5 6 7 8
1. Internalized stigma

2. Depression symptomology .184†

3. Lifetime homophobia
experiences
.233* .232*

4. Sensation seeking .086 .177† .108

5. Gay community connection -.244* -.066 .181† -.137

6. Days consuming alcohol in
last 30 days
-.242* .109 .134 .183† .214*

7. Days binge drinking in last
30 days
-.117 .060 .161 .342*** .016 .649***

8. Marijuana use in last 30 days .150 .236* .030 .373*** -.021 .227* .186†

9. Other substance use in last
30 days
.098 .014 .004 .243* -.041 .148 .100 .462***
†p < .10. *p < .05. **p < .01. ***p < .001.

SUBSTANCE USE AND TECHNOLOGY 157
Table 5.6. Relationship between Covariates and Substance Use during the Prior Month Moderated by Recruitment Type

Alcohol Binge Drinking Marijuana Other Substance
Variables b SE t b SE t OR 95% CI OR 95% CI
Internalized stigma -0.200 0.107 -1.859† -0.063 0.084 -0.751

Recruitment type -3.578 3.355 -1.067 0.506 2.621 0.193

Interaction 0.031 0.169 0.185 -0.028 0.132 -0.210

Depressive symptoms

1.151† 0.989, 1.339

Recruitment type

0.667 0.081, 5.134

Interaction

0.992 0.772, 1.291

Sensation seeking 0.067 0.131 0.513 0.228 0.096 2.367* 1.225** 1.091, 1.376 1.088† 0.994, 1.191
Recruitment type -10.274 6.081 -1.690† -2.703 4.462 -0.606 28.805 0.235, 3,525.392 0.370 0.003, 39.320
Interaction 0.251 0.208 1.207 0.102 0.153 0.667 0.879 0.748, 1.033 1.007 0.862, 1.176
Gay community connection 0.399 0.151 2.650** 0.178 0.115 1.543

Recruitment type 8.327 6.139 1.356 12.384 4.706 2.631*

Interaction -0.547 0.289 -1.881† -0.600 0.221 -2.714**
Note. For recruitment type, 0 = venue, 1 = app.
†p < .10. *p < .05. **p < .01.

SUBSTANCE USE AND TECHNOLOGY 158
Figure 5.1. Moderation of GCC and Number of Days Binge Drinking by Recruitment Type

4.52
0.61
2.16
3.81
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
Low GCC High GCC
Days of Binge Drinking
App Venue
GCC = Gay Community Connection
SUBSTANCE USE AND TECHNOLOGY 159
Figure 5.2. Moderation of GCC and Number of Days Drinking Alcohol by Recruitment Type

5.25
3.88
6.06
9.75
0.00
2.00
4.00
6.00
8.00
10.00
12.00
Low GCC High GCC
Days of Drinking Alcohol
App Venue
GCC = Gay Community Connection
SUBSTANCE USE AND TECHNOLOGY 160
Chapter 6 (Paper 3): Venue-Based versus Geosocial Networking Application-Based
Recruitment of Young Men Who Have Sex with Men: An Examination of Feasibility
Introduction
Probability-based recruitment of young men who have sex with men (YMSM) can be
considerably difficult because the sampling frame is largely unknown (Heckathorn, 1997). To
respond to this barrier, researchers have historically relied on locations (i.e., gay venues, bars,
and clubs) of known attendance to aid in recruitment of this hidden population (Gold, Skinner, &
Ross, 1994; Kalichman, Kelly, Morgan, & Rompa, 1997; Ridge, Plummer, & Minichiello,
1994). Although numerous studies have recruited YMSM from venues using convenience
sampling, other researchers have developed rigorous methods that systematically identify
locations and periods of recruitment to allow a probability-based sample to be recruited. Venue-
based stratified probability sampling was developed primarily to recruit more generalizable
samples of YMSM, by identifying all locations where and times when YMSM congregate (Ford
et al., 2009). These methods include random selection of appropriate 4-hour venue-day-time
periods (VDTs) to recruit clusters of YMSM for research. A large study team and numerous
hours of work are required to implement these methods because each potential venue must be
observed at all times during every day of the week. Although venue-based probability sampling
is systematic in nature and probability based, it has been criticized for being largely inaccessible
to researchers due to the high cost and labor-intensive nature of implementation (Meyer &
Wilson, 2009). Therefore, the majority of research on this population has relied on convenience
samples recruited in person at venues (Boone, Cook, & Wilson, 2013; Dolezal, Carballo-
Diéguez, Nieves-Rosa, & Dı
́ az, 2000; Halkitis & Figueroa, 2013; Kelly, Davis, & Schlesinger,
2015; Mansergh et al., 2008; Mutchler et al., 2011) or through websites (Fernández et al., 2004;
SUBSTANCE USE AND TECHNOLOGY 161
Fernández et al., 2007; Horvath, Bowen, & Williams, 2006; Horvath, Rosser, & Remafedi, 2008;
Hospers, Kok, Harterink, & de Zwart, 2005; Pequegnat et al., 2007; Sullivan et al., 2011).
New innovations in technology during the past decade may hold unique opportunities for
sampling. For example, recent literature has suggested that YMSM are increasingly using
technology to meet other YMSM and relying less on venues to provide social interaction (Grov,
2012; Grov & Crow, 2012; Zablotska, Holt, & Prestage, 2012). In 2009, geosocial networking
apps (GSNAs) were introduced to the Apple smartphone app market, starting with the creation of
Grindr (Leslie, 2009). During the past 9 years, these apps have grown in popularity among
YMSM, with some apps boasting 25 million users worldwide (Hornet, 2018). On most GSNAs,
users see a home screen with a matrix of pictures corresponding with the individual users closest
to their geographic location. Users can then message these individuals, send pictures, exchange
other contact information, or send their exact map location using the integrated message system.
YMSM can browse the profiles of other YMSM in the same geographic region, and because
profiles are ordered based on their relative proximity to the user, YMSM users may click on a
profile and see the distance of other users from their current location.
Given this cultural shift and advancement in technology, YMSM researchers have begun
to look to GSNAs to aid in study recruitment. Studies have recruited both probability (Gibbs &
Rice, 2016; Rice at al., 2011) and convenience samples (Buckingham et al., 2017; Goedel &
Duncan, 2016; Goedel, Hagen, et al., 2017; Goedel, Mayer, Mimiaga, & Duncan, 2017; Goedel,
Safren, Mayer, & Duncan, 2017; Grov, Rendina, Jimenez, & Parsons, 2016; Lorimer, Flowers,
Davis, & Frankis, 2016; Phillips, Grov, & Mustanski, 2015; Siegler et al., 2015) of men who
have sex with men from GSNAs. App-based methods allow researchers to quickly connect with
large samples of this population (Buckingham et al., 2017). Additionally, because users are
SUBSTANCE USE AND TECHNOLOGY 162
required to reveal their smartphone geolocation, researchers can specify locations for the display
of recruitment advertisements (Buckingham et al., 2017; Goedel, Hagen, et al., 2017; Goedel,
Mayer, et al., 2017; Goedel, Safren, et al., 2017; Grov et al., 2016; Lorimer et al., 2016).
The literature also has suggested that app-based recruitment procedures may be more cost
efficient than venue-based procedures (Rice et al., 2012). Additionally, app-based studies
reported reaching the targeted study sample size in a short amount of time (Buckingham et al.,
2017). These indications are important, because previous research has found that certain YMSM
recruitment methods (e.g., respondent-driven sampling) can require an extended data collection
period (i.e., up to 3 years; Kuhns et al., 2015; Lachowsky et al., 2016). The odds of a significant
historic event are increased during an extended data collection process and threaten internal
validity (Schaie, 1983). Although app-based recruitment shows promise in reducing cost and
limiting the length of time required for recruitment, the recruitment efficiency of these methods
is varied (Zou & Fan, 2017). A recent systematic review and meta-analysis of studies using
GSNAs to recruit men who have sex with men revealed that the ratio of approached potential
participants to those who completed the study varied widely, from 100:2 to 100:74 (Zou & Fan,
2017). This wide variation in reported recruitment efficiency may be due to a lack of consensus
among researchers regarding how to accurately measure this metric (e.g., users who click the ad
versus users in the area where the ad is passively broadcast). Because this technology and its use
in research is new, very little is known about the utility of GSNA-based recruitment compared to
other methods.
To date, no known study has compared the feasibility of implementing venue-based
probability sampling to GSNA-based probability sampling. Therefore, the aim of this chapter is
to examine and compare the recruitment efficiency (i.e., length of time to complete the study,
SUBSTANCE USE AND TECHNOLOGY 163
number of recruitment periods required, progression of potential participants through the study
process) and cost efficiency (i.e., study team work hours, financial resources) of venue-based and
GSNA-based probability recruitment of YMSM.
Methods
A cross-sectional survey was conducted in Los Angeles County from April 2017 through
April 2018 using two forms of recruitment. Of 111 YMSM (aged 18–24 years old) recruited, 68
were recruited using venue-based stratified probability sampling procedures and 43 using
GSNA-based probability sampling procedures. The venue-based stratified probability sampling
methods were informed by the Healthy Young Men’s Study conducted in Los Angeles (Ford et
al., 2009). The GSNA methods were informed by a previous study by this dissertation’s author
and colleagues, also conducted in Los Angeles (Rice et al., 2012). Geographic information
systems methods were also used to build the GSNA sampling frame and were informed by a
2014 study conducted in Atlanta, Georgia (Delaney et al., 2014). Both recruitment methods
followed two stages: recruitment preparation and recruitment implementation. Both stages were
completed with a specific geographic sampling frame (GSF, or location for recruitment). The
principal area spanned approximately 70 square kilometers; this area was later enlarged to an
area approximately 50 times larger (the southern portion of Los Angeles County) to maximize
recruitment. During each stage of the study, all hours of work and costs were recorded and
tracked by the principal investigator and research assistants. All procedures for this study were
approved by the institutional review board of the University of Southern California.
Recruitment Preparation
Venue-based stratified probability sample. To identify the study sampling frame for
venue-based recruitment, the study team conducted brief interviews with men in public areas in
SUBSTANCE USE AND TECHNOLOGY 164
the GSF and asked interviewees to report known locations for finding YMSM, resulting in a final
list of 41. The study team presented the compiled list of venues to a community advisory board
and asked members to nominate missing venues, eliminate venues that they believed were
inappropriate, and provide insight regarding the best times to find YMSM during the four
standard 4-hour sampling periods (8 a.m.–12 p.m., 12 p.m.–4 p.m., 4 p.m.–8 p.m., and 8 p.m.–12
a.m.). Participants received a $30 gift card incentive, and the work resulted in an exhaustive list
of 43 venues currently frequented by YMSM. Investigation of each venue using Google’s
“Popular Times” function, information from each venue’s website, and the information gathered
from the focus group identified 199 periods theoretically high in YMSM attendance (i.e., VDTs).
Venue enumeration was conducted during a 6-month period. The procedures of
enumeration followed those outlined by the Healthy Young Men’s Study (Ford et al., 2009).
Type I enumeration required one study team member to visit each venue during the identified 4-
hour periods (i.e., 199 VDTs). For 60 minutes, the study team member counted the number of
individuals who appeared to fit the study criteria (i.e., male, aged 18–24). The number counted
was then multiplied by 4 to approximate a 4-hour period. Pilot enumerations were also
conducted wherein one study team member counted individuals who appeared to meet the study
criteria and a second study team member asked those individuals if they met the eligibility
criteria (i.e., date of birth, gender, and self-identification as gay, bisexual, or uncertain about
sexual orientation or report of having sex with men). Based on the pilot enumerations, team
members correctly identified eligible individuals 65% of the time, and this number was
multiplied by the 4-hour enumeration to estimate the number of eligible individuals.
An estimated eight individuals in a 4-hour period was identified as the threshold for an
appropriate VDT based on recommendations from previous studies (Ford et al., 2009; MacKellar
SUBSTANCE USE AND TECHNOLOGY 165
et al., 2007). If a Type I enumeration for a specific VDT resulted in an estimate lower than eight
eligible individuals, then the VDT required a type II enumeration. Of the 199 type 1
enumerations completed, 144 were determined to be appropriate and 55 required a type II
enumeration.
Type II enumerations were conducted by two study team members for two 1-hour periods
separated by 1 hour and followed the same procedures as the pilot enumerations. If a VDT was
found to have an estimate less than eight, it was excluded from the sampling frame. Of the 55
type II enumerations completed, 12 were identified as appropriate and 43 as inappropriate and
excluded as potential VDTs. At the end of Stage 2, 156 VDTs were identified as appropriate for
recruitment.
GSNA probability sample. The community advisory board was also asked to report on
the most popular GSNAs used by YMSM in the GSF. Each proposed GSNA needed to meet
three criteria to be included: (a) ability to filter users by age, (b) ordering of user profiles by
relative proximity to the study smart device, and (c) reporting of the relative distance of users
from the study smart device. The top three GSNAs were contacted by the study’s principal
investigator to inquire about interest in collaborating. From these conversations, the study GSNA
was selected (i.e., Hornet).
The following procedures were used to recruit most of the GSNA-based sample (all but
six individuals, who were recruited during pilot testing of the methods). Procedures used to
collect user density data are outlined in Chapter 4 of this dissertation. Similar to the venue-based
procedures, the GSNA-based methods required that users could be recruited from each sampling-
day-time period (SDTP; four periods × 7 days = 28). The periods used were the midpoints of
each VDT: 10 a.m., 2 p.m., 6 p.m., and 10 p.m.
SUBSTANCE USE AND TECHNOLOGY 166
Five points in the GSF were chosen as the locations for recruitment to maximize
coverage and allow for all the users of the study GSNA to have the same probability of being
approached for recruitment. The number of sampling points needed for each SDTP in the GSF
was determined by three criteria: point-based circular buffers of 8 miles that (a) maximized
coverage of the GSF, (b) minimized buffer overlap, and (c) minimized coverage of areas outside
the GSF boundary. Once completed, these procedures identified 140 day-time-sampling
locations, or DTSLs (five points × 28 SDTPs).
Recruitment Implementation
A monthly sampling calendar was used to randomly select both VDTs and DTSLs.
Creation of calendars followed the procedures used in Ford et al. (2009). To create the monthly
sampling calendar, all VDTs and DTSLs were loaded into a spreadsheet. VDTs and DTSLs were
categorized as weekday day, weekday night, weekend day, and weekend night to allow for
matching between recruitment methods, because it was hypothesized that different individuals
would be located in the GSF at each of these times. One randomly selected VDT or DTSL was
placed in the appropriate day and time on the calendar and a matched random selection was
taken from the other recruitment method. Selection of VDTs and DTSLs continued
approximately 10 VDTs and three or four DTSLs (the study GSNA preferred to limit monthly
recruitment periods to avoid overburdening users) were placed on the study calendar. Once the
targeted sample size was achieved for a recruitment strategy, the corresponding remaining
sampling events for that month were disregarded.
For venue-based recruitment, in accordance with the monthly sampling calendar, two or
three study team members recruited a sample of participants from the specified VDT. While one
study team member counted the number of men who appeared to be eligible, the remaining study
SUBSTANCE USE AND TECHNOLOGY 167
team members individually approached potential participants and asked if they were willing to
answer the eligibility questions to support research. If the potential participant agreed to
participate, he completed the eligibility survey on a study iPad. At the end of the eligibility
survey, the potential participant created a unique study ID code and input his contact information
(i.e., email address, mobile phone number). The potential participant received a $5 physical gift
card incentive for completing the eligibility survey, because pilot testing of the methods revealed
that this procedure encouraged the venue-recruited individuals to complete the eligibility survey.
For GSNA-based recruitment, instead of visiting the specified geographic point, the study
GSNA was technologically misled and provided with GPS coordinates for the DTSL specified
on the monthly sampling calendar (using a third-party application, “Fake GPS Location”). This
allowed the study team to stay in a fixed location. As noted, each DTSL had an associated GPS
point and radial distance (i.e., 8 miles). The recruitment process began by one study team
member opening the GSNA on the study smart device and making an anonymous user profile.
After filtering the visible user profiles by age eligibility (i.e., 18–24 years old), the study team
member recorded all public GSNA user IDs within 8 miles of the study smart device.
Concurrently, the study GSNA (i.e., Hornet) sent out a preset direct message broadcast in the
GSNA to each GSNA user within 8 miles of the DTSL. The direct message (e.g., “Hey, you
appear to be eligible for a USC study! Do you have 2 minutes to answer a few questions? If
you’re eligible and you take the survey, you can earn up to $50”) appeared as a personal message
from Hornet and remained in users’ message inbox. Once potential participants clicked on the
link, they could complete the same eligibility survey previously described. No eligibility
incentive was provided to GSNA-recruited potential participants because pilot tests revealed this
procedure did not increase the rate of screening completion.
SUBSTANCE USE AND TECHNOLOGY 168
The eligibility data for both recruitment methods was received by the contact information
manager, who determined if individuals were eligible to take the study survey (i.e., met the study
eligibility criteria and had a unique study ID code indicating they had not previously taken the
main study survey). The contact information manager then provided a list of new eligible unique
study IDs to the data manager, who never had access to the contact information of the
participants. The data manager used the unique study IDs to create unique one-use web links to
the main study survey and provided the contact information manager with these links. The
contact information manager then sent the potential participant, using his preferred mode of
contact, instructions to take the main study survey, his unique study ID, and the unique link to
take the main study survey. The contact information manager never had access to the main study
survey data. Additionally, both the contact information manager and data manager kept a record
of all hours of work completed for each recruitment method.
Upon entering the main study survey website, potential participants had the opportunity
to read informed consent information. After giving informed consent, participants were
instructed to enter their unique study ID code. The main study survey asked participants to report
demographic information, substance use characteristics, substance use determinants, and sexual
risk behaviors. Once completing the main study survey, participants were asked to indicate if
they had previously completed the study. This item was asked as a safeguard against duplicate
data. Once a participant completed the main study survey, the submitted data were available to
the data manager. The data manager provided a list of unique study ID codes of individuals who
completed the main study survey and a unique link to an associated supplemental survey to the
contact information manager. The contact information manager then sent each participant a
downloadable gift card ($25 for a venue-recruited participant and $35 for a GSNA-recruited
SUBSTANCE USE AND TECHNOLOGY 169
participant). In addition, the contact information manager messaged each participant who
completed the main study survey a link to the supplemental survey with relevant instructions.
The supplemental survey followed the same procedures as the main study survey, except that the
participant was incentivized with a $15 downloadable gift card.
Measurement
During the research process, a detailed log of all recruitment periods, study personnel
work hours, and costs was created. Recruitment periods for venue-based recruitment were
tracked by the study personnel in the field, and a deidentified spreadsheet was used by the data
manager to track the progression of all potential participants through the study process (i.e.,
eligible to participate, participated in the main survey, participated in the supplemental survey).
GSNA recruitment periods were tracked by recording all unique GSNA users in the geographic
area during the recruitment broadcast. This served as the number of unique users “approached”
via the GSNA during each recruitment period. Similar to the venue-based tracking procedures,
all GSNA-based potential participants identified as eligible were included in a tracking
spreadsheet, which was updated with their research progression throughout the research process.
All work hours and costs were recorded in one of four categories: recruitment
preparation, recruitment implementation, software, and hardware (Wang et al., 2003).
Recruitment preparation involved all work hours and costs that preceded recruitment of study
participants. These items fit in six categories: doctoral candidate administrative work (e.g.,
organizing study team, identifying and assigning weekly enumeration, institutional review board
applications, study protocol development, analysis of user density data); doctoral candidate
fieldwork (e.g., meetings with venue management, collection of GSNA user density data,
meetings with the study GSNA); student or temporary worker fieldwork (e.g., venue
SUBSTANCE USE AND TECHNOLOGY 170
enumerations, pilot enumerations, brief interviews); training of students or temporary workers;
community advisory board meetings (e.g., study personnel work hours, key informant
incentives); and pilot testing of methods (preliminary recruitment periods used to identify
improvements in methods).
Recruitment implementation activities were defined by six categories: doctoral candidate
administrative work (e.g., data management, planning recruitment periods); doctoral candidate
fieldwork (e.g., observing GSNA users during recruitment periods); student or temporary worker
fieldwork (i.e., venue-based recruitment periods); student or temporary worker administrative
work (e.g., contact information management); advertisements (i.e., GSNA broadcast
advertisements); and study incentives (i.e., main survey, supplemental survey). All student
workers and temporary workers were paid at the same rate of $15 per hour; however, because of
administrative costs, direct costs varied by worker. Specifically, student workers, while
university was in session, cost the project $15.00 per hour, but during the summer and winter
break periods, they cost the project $20.03 per hour. Temporary workers earned an income of
$15.00 per hour, but including fringe benefits, they cost the study project $18.89 per hour.
Additionally, although the doctoral candidate received a stipend throughout the administration of
the study, all work hours conducted by the doctoral candidate were recorded at a cost of $25.00
per hour to simulate the cost of a graduate-level program director.
Software costs included a 2-year license for ArcGIS to analyze user density data from the
study GSNA. Hardware costs included four study iPads used for venue-based recruitment,
associated data plans, and one android-based smartphone used to collect GSNA data.
SUBSTANCE USE AND TECHNOLOGY 171
Analytic Procedure
To address recruitment efficiency, recruitment periods were treated as individual
observations. Recruitment periods were grouped by method and compared using two-sample t-
tests for the number of individuals observed (unique and total number, including repeats), who
completed the eligibility survey, deemed eligible, who completed the main survey, and who
completed the supplemental survey. Additionally, the venue-based methods informing this study
used enumerations to identify locations where an estimated eight eligible individuals would be
observed (Ford et al., 2009). Therefore, to test whether these estimates were accurate,
recruitment periods from each method were compared based on whether eight or more
potentially eligible participants were observed, using two chi-square tests to identify differences.
A direct comparison of costs was conducted using previously described cost subgroups
(Wang et al., 2003). Additionally, the recruitment methods were compared based their cost per
participant (i.e., main survey completed), because a different number of participants was
recruited for each method. The findings from the cost analysis and recruitment efficiency
analysis were then combined to estimate the cost per participant for each recruitment method,
based on the final sample size. All final cost estimates are provided in 2018 U.S. dollars.
Results
GSNA-based and venue-based recruitment implementation was completed in 18 and 59
recruitment periods, respectively. App-based recruitment occurred during 5 months, whereas
venue-based recruitment occurred during a 7-month period. During this time, potential study
participants were observed either in person around venues identified as locations where YMSM
congregate or while using the study GSNA, Hornet. Potential participants were tracked
throughout the study process (observed, approached, completed eligibility survey, deemed
SUBSTANCE USE AND TECHNOLOGY 172
eligible, completed main survey, and completed supplemental survey). Figure 6.1 summarizes
the flow of potential participants through the study process. Percentages in black indicate the
percentage of individuals at the previous stage who moved on to the subsequent stage (e.g., of
those found eligible in the venue-based recruitment, 38% completed the main survey). Red
percentages indicate the percentage of the originally identified potential participants (i.e.,
approached for venue-based and unique users for GSNA-based) at each stage (e.g.,
approximately 6% of unique users recruited through the GSNA completed the main survey).
Six two-sample t-tests and two chi-square analyses were conducted to test for differences
between the recruitment periods (see Table 6.1). Recruitment periods from the GSNA method
had significantly (t[16.89] = 3.197, p < .01) more observed potential participants (M = 46)
compared to venue-based recruitment periods (M = 13). GSNA-based recruitment periods also
had significantly (t[17.20] = 2.574, p < .05) more unique potential participants (M = 35)
compared to recruitment periods from the venue-based method (M =13). This same result
(t[18.15] = 3.206, p < .01) was found regarding the number of individuals who completed the
eligibility survey; an average of 19 individuals completed the eligibility survey for each app-
based recruitment period, compared to six who completed the eligibility survey for each venue-
based recruitment period. No significant differences in the number of individuals found to be
eligible, completing the main survey, or completing the supplemental survey was indicated
between the two recruitment period types. Two chi-square analyses indicated a significant
difference between the proportion of recruitment periods in which eight or more individuals were
observed (χ
2
[1, 76] = 10.734, p < .01) and in which eight or more unique individuals were
observed (i.e., not repeats; χ
2
[1, 76] = 4.856, p < .05).
SUBSTANCE USE AND TECHNOLOGY 173
Total time (i.e., study team work hours) required to complete each method was tracked
and tabulated (see Table 6.2), indicating that 1,225 work hours were needed to complete venue-
based recruitment (i.e., preparation and implementation), compared to 238 hours for GSNA-
based recruitment (19%). The recruitment preparation stage of venue-based recruitment relied
heavily on student and temporary workers, accounting for 480 of the 669 necessary hours.
Compared to venue-based recruitment preparation, GSNA-based recruitment (144 hours)
required 21% of the work hours. Fewer research work hours were spent on the implementation of
each recruitment strategy (556 for venue-based recruitment and 94 for app-based recruitment).
Further, a team of six was necessary to complete venue-based recruitment methods, compared to
two study team members committed to the GSNA-based recruitment methods.
A cost analysis of the two recruitment methods indicated meaningful differences in the
financial resources necessary to complete each strategy (see Table 6.3). Costs were separated
into four categories: recruitment preparation, recruitment implementation, software, and
hardware (Wang et al., 2003). Recruitment preparation for the venue-based procedures
($13,013.75) cost approximately 3 times more than the app-based procedures ($4,042.92). The
majority of this cost difference was due to the $8,835 used to fund student and temporary
workers while they completed the 199 type I and 55 type II enumerations. Recruitment
implementation was almost 4 times more expensive for the venue-based procedures ($14,619.79)
compared the app-based procedures ($4,025.00). However, a proportion of this difference was
due to a larger number of venue-recruited participants (n = 68) completing the study survey,
compared to GSNA-recruited participants who completed the final survey (n = 43). When the
survey incentive was removed, the venue-based procedures ($10,229.79) were more than 4 times
costlier compared to GSNA-based procedures ($2,250.00). Overall, the app-based procedures
SUBSTANCE USE AND TECHNOLOGY 174
cost approximately $6,791.92 to complete and the venue-based procedures cost $26,051.54 to
complete. Further, when sample size was considered, an GSNA-based participant cost $157.95,
more than 2 times less than a venue-recruited participant ($383.11).
Using costs and recruitment efficiency statistics, estimation of the costs per participant
per recruitment method were approximated for study sample sizes ranging from 30 to 1,000
individuals (see Figure 6.2). The cost per venue-recruited participant was calculated using a
formula that accounted for sample size, implementation preparation without pilot testing
($12.074.29), eligibility incentives ($5), rate of those assessed for eligibility versus who
completed the study (328/68 = 4.824), participant incentives ($25), and cost per participant for
study implementation, excluding incentives ($150.44; Costvenue = [$12,074.29 + (n × 4.824 × $5)
+ (n × $25) + (n × $150.44)] ÷ n). Similarly, cost per GSNA-recruited participant was calculated
using a formula that accounted for sample size, implementation preparation without pilot testing
($2,887.50), broadcast advertisements (estimated at $1,500), a $10 higher participant incentive
($35), and cost per participant for study implementation, excluding incentives ($47.30; CostGSNA
= [$2,887.50 + $1,500 + (n × $35) + (n × $47.30)] ÷ n).
For a sample size of 200, the GSNA recruitment procedures were estimated to cost
$99.85 per participant, whereas a venue-recruited participant was estimated to cost $247.85.
Figure 6.2 shows that the cost per participant begins to level off at a sample size of 600
participants (venue = $219.68, app = $89.61) and remains steady through 1,000 (venue =
$211.63, app = $86.68). Estimates indicate that a venue-recruited participant would cost between
2.44 and 2.63 times more than an GSNA-recruited participant at study sample sizes up to 1,000.
SUBSTANCE USE AND TECHNOLOGY 175
Discussion
The primary aim of this chapter was to compare the feasibility of implementing venue-
based and GSNA-based probability sampling. By comparing recruitment efficiency and
analyzing study costs, several meaningful findings emerged: (a) app and venue recruitment
periods vary in their recruitment efficiency, (b) in the current study, venue-based recruitment
was costlier and more labor intensive, and (c) at all sample sizes, venue-based recruitment is
expected to be at least twice as expensive as GSNA-based recruitment.
The ability for each recruitment method to successfully recruit potential participants who
complete the study is not clearly stronger for one method over the other. As seen in Figure 6.1,
venue-based recruitment appeared to more successfully recruit a larger proportion of eligible
potential participants (24% of total approached) compared to the app-based recruitment (11% of
total unique users observed). However, app-recruited individuals were more likely to participate
in the study once found eligible. Venue-based recruitment procedures may increase screening of
potential participants, whereas app-based recruitment procedures might facilitate completion of
the study. It is possible that individuals recruited in person were more likely to complete the
eligibility survey because of issues of social desirability, given that they were personally
recruited face to face. Previous literature has found similar findings when comparing face-to-face
recruitment to telephone recruitment (Holbrook, Green, & Krosnick, 2003). Once venue-
recruited individuals were requested to complete the study survey online, these individuals may
have experienced less pressure to appear socially desirable, because the study personnel
interactions were less personalized. This contrasts with GSNA-recruited potential participants
who chose to complete the eligibility survey and main study survey given the same level of
social pressure (i.e., online social message). These findings are consistent with online survey
SUBSTANCE USE AND TECHNOLOGY 176
research showing little evidence for social desirability bias among general populations
(Lindhjem & Navud, 2011).
Findings from the financial and hourly cost analysis provide evidence for GSNA-based
recruitment as a cost-efficient method for recruiting probability samples of YMSM. Venue-based
recruitment required 3 times the number of work hours to recruit a participant compared to
GSNA-based recruitment. Further, the venue-based procedures implemented in this study
required a study team 3 times larger than the app-based procedures. Beyond the number of hours
required to complete each method, the venue-based procedures were more than twice as
expensive as the GSNA-based procedures. GSNA-based recruitment procedures appeared largely
less burdensome and easier to implement, using less resources and smaller research teams.
The cost of each recruitment method has meaningful implications for research on
YMSM. Estimates from the National Institutes of Health (2017) indicated that from 2017 to
2018, the funding committed to research on sexual and gender minority people will be cut by
approximately 23%. When the most rigorous method for probability-based recruitment of
YMSM has a high cost, given the dearth of federal funding, researchers have two options: (a)
complete several expensive and lengthy probability-based YMSM studies, or (b) complete
numerous low-cost studies using samples of YMSM that may have limited generalizability to the
population. Based on the estimates presented in Figure 6.2, researchers appear to have a third
alternative, because GSNA-based probability sampling can cut costs of probability-based
recruitment by more than half.
There are several study limitations to note. Although the cost analysis presented in this
chapter is a detailed account of all the costs associated with the research process of completing
each recruitment methodology, other costs were excluded (e.g., administrative costs associated
SUBSTANCE USE AND TECHNOLOGY 177
with managing the grant funds at the institutional level, reimbursement processing, conference
travel to present study results, monthly stipend to the doctoral candidate). These costs were not
tracked because (a) they are typical costs associated with any primary data collection, and (b) the
purpose of this study was to compare the cost of the research process for each recruitment
methodology. Therefore, any future study using these cost estimates should expect additional
costs outside of direct research costs. Moreover, although the doctoral candidate in this study
received a monthly stipend and therefore was not paid at an hourly wage, for the purposes of the
cost analysis, all work completed by the doctoral candidate was recorded at a rate of $25 an hour
to approximate the cost of a project manager (i.e., graduate professional). Depending on the
institution and cost of living in the area, this hourly cost could differ. Beyond limitations related
to how costs were determined, this study was conducted in a large urban area. Therefore,
application of these methods and expected costs may vary based on population density. Although
venue-based probability sampling procedures could be considerably influenced by
implementation in a rural context (because there are limited venues for YMSM in rural areas),
the procedures and costs associated with GSNA-based probability sampling procedures are
expected to be consistent.
An additional limitation of this study relates to how app advertisements were broadcast.
All users of the study GSNA received the study broadcast advertisement in the selected area at
the specific recruitment time. This procedure allowed for all users, in addition to the 18- to 24-
year-old reported users, to take the eligibility survey. Although this procedure was inclusive in
that it allowed 18- to 24-year-olds who chose to not disclose their age in the GSNA public
profile to participate, it also made meaningful comparisons of the number of potential
participants who completed the eligibility survey not possible. Therefore, because app users at
SUBSTANCE USE AND TECHNOLOGY 178
any age could complete the eligibility survey, drawing conclusions from the finding that
significantly more app-recruited individuals completed the eligibility survey compared to venue-
recruited individuals is discouraged.
Despite these limitations, findings from this study inform several directions for future
research. First, because the current study was implemented in a large urban area in the United
States, GSNA-recruitment procedures should be investigated for their feasibility in rural areas of
the United States and developing countries. Resources in service-impoverished areas (e.g., rural
areas and developing countries) tend to be limited. Therefore, GSNA-based recruitment may
reduce the cost to study YMSM who live in these marginalized areas. Additional research is also
needed to determine factors (e.g., study GSNA selection, type of advertisement, location of
recruitment) that contribute to the wide variety of reported recruitment ratios (Zou & Fan, 2017).
Last, although recruitment of cross-sectional study samples has been shown to be feasible in the
current study and previous studies using GSNAs for recruitment (Buckingham et al., 2017;
Goedel & Duncan, 2016; Goedel, Hagen, et al., 2017; Goedel, Mayer, et al., 2017; Goedel,
Safren, et al., 2017; Grov et al., 2016; Lorimer et al., 2016; Phillips, Grov, & Mustanski, 2015;
Siegler et al., 2015), little is known about the application of these methods to recruit YMSM for
longitudinal and intervention research. Research investigating the feasibility of GSNA methods
for recruiting YMSM for intervention and longitudinal research is needed.
Although the aim of this chapter was to investigate costs of two different recruitment
methods, the overarching goal of this study was to have an impact on the health of YMSM. By
reducing costs and increasing accessibility of probability-based methods, more rigorous research
can be conducted on the factors that contribute to YMSM health outcomes.

SUBSTANCE USE AND TECHNOLOGY 179
References
Boone, M. R., Cook, S. H., & Wilson, P. (2013). Substance use and sexual risk behavior in HIV-
positive men who have sex with men: An episode-level analysis. AIDS and Behavior, 17,
1883–1887. doi:10.1007/s10461-012-0167-4
Buckingham, L., Becher, J., Voytek, C. D., Fiore, D., Dunbar, D., Davis-Vogel, A., … Frank, I.
(2017). Going social: Success in online recruitment of men who have sex with men for
prevention HIV vaccine research. Vaccine, 35, 3498–3505.
doi:10.1016/j.vaccine.2017.05.002
Delaney, K. P., Kramer, M. R., Waller, L. A., Flanders, W. D., & Sullivan, P. S. (2014). Using a
geolocation social networking application to calculate the population density of sex-
seeking gay men for research and prevention services. Journal of Medical Internet
Research, 16, e249. doi:10.2196/jmir.3523
Dolezal, C., Carballo-Diéguez, A., Nieves-Rosa, L., & Dı
́ az, F. (2000). Substance use and sexual
risk behavior: Understanding their association among four ethnic groups of Latino men
who have sex with men. Journal of Substance Abuse, 11, 323–336. doi:10.1016/S0899-
3289(00)00030-4
Fernández, M. I., Varga, L. M., Perrino, T., Collazo, J. B., Subiaul, F., Rehbein, A., … Bowen,
G. S. (2004). The Internet as recruitment tool for HIV studies: Viable strategy for
reaching at-risk Hispanic MSM in Miami? AIDS Care, 16, 953–963.
doi:10.1080/09540120412331292480
Fernández, M. I., Warren, J. C., Varga, L. M., Prado, G., Hernandez, N., & Bowen, G. S. (2007).
Cruising in cyber space: Comparing Internet chat room versus community venues for
recruiting Hispanic men who have sex with men to participate in prevention studies.
SUBSTANCE USE AND TECHNOLOGY 180
Journal of Ethnicity in Substance Abuse, 6, 143–162. doi:10.1300/J233v06n02_09
Ford, W. L., Weiss, G., Kipke, M. D., Ritt-Olson, A., Iverson, E., & Lopez, D. (2009). The
Healthy Young Men’s Study: Sampling methods to recruit a random cohort of young
men who have sex with men. Journal of Gay & Lesbian Social Services, 21, 357–373.
doi:10.1080/10538720802498280
Gibbs, J. J., & Rice, E. (2016). The social context of depression symptomology in sexual
minority male youth: Determinants of depression in a sample of Grindr users. Journal of
Homosexuality, 63, 278–299. doi:10.1080/00918369.2015.1083773
Goedel, W. C., & Duncan, D. T. (2016). Contextual factors in geosocial-networking smartphone
application use and engagement in condomless anal intercourse among gay, bisexual, and
other men who have sex with men who use Grindr. Sexual Health, 13, 549–554.
doi:10.1071/SH16008
Goedel, W. C., Hagen, D., Halkitis, P. N., Greene, R. E., Griffin-Tomas, M., Brooks, F. A., …
Duncan, D. T. (2017). Post-exposure prophylaxis awareness and use among men who
have sex with men in London who use geosocial-networking smartphone applications.
AIDS Care, 29, 579–586. doi:10.1080/09540121.2016.1259455
Goedel, W. C., Mayer, K. H., Mimiaga, M. J., & Duncan, D. T. (2017). Considerable interest in
pre-exposure prophylaxis uptake among men who have sex with men recruited from a
popular geosocial-networking smartphone application in London. Global Public Health.
Advance online publication. doi:10.1080/17441692.2017.1391859
Goedel, W. C., Safren, S. A., Mayer, K. H., & Duncan, D. T. (2017). Community-level norms
and condomless anal intercourse among gay, bisexual, and other men who have sex with
SUBSTANCE USE AND TECHNOLOGY 181
men who use geosocial-networking smartphone applications in the Deep South. Journal
of HIV/AIDS & Social Services, 16, 382–389. doi:10.1080/15381501.2017.1344599
Gold, R. S., Skinner, M. J., & Ross, M. W. (1994). Unprotected anal intercourse in HIV-infected
and non-HIV-infected gay men. Journal of Sex Research, 31, 59–77.
doi:10.1080/00224499409551730
Grov, C. (2012). HIV risk and substance use in men who have sex with men surveyed in
bathhouses, bars/clubs, and on Craigslist.org: Venue of recruitment matters. AIDS and
Behavior, 16, 807–817. doi:10.1007/s10461-011-9999-6
Grov, C., & Crow, T. (2012). Attitudes about and HIV risk related to the “most common place”
MSM meet their sex partners: Comparing men from bathhouses, bars/clubs, and
Craigslist.org. AIDS Education and Prevention, 24, 102–116.
doi:10.1521/aeap.2012.24.2.102
Grov, C., Rendina, H. J., Jimenez, R., & Parsons, J. T. (2016). Using online settings to identify
gay and bisexual men willing to take or with experience taking PrEP: Implications for
researchers and providers. AIDS Education and Prevention, 28, 378–392.
doi:10.1521/aeap.2016.28.5.378
Halkitis, P. N., & Figueroa, R. P. (2013). Sociodemographic characteristics explain differences
in unprotected sexual behavior among young HIV-negative gay, bisexual, and other
YMSM in New York City. AIDS Patient Care and STDs, 27, 181–190.
doi:10.1089/apc.2012.0415
Heckathorn, D. D. (1997). Respondent-driven sampling: A new approach to the study of hidden
populations. Social Problems, 44, 174–199. doi:10.2307/3096941
SUBSTANCE USE AND TECHNOLOGY 182
Holbrook, A. L., Green, M. C., & Krosnick, J. A. (2003). Telephone versus face-to-face
interviewing of national probability samples with long questionnaires: Comparisons of
respondent satisficing and social desirability response bias. Public Opinion Quarterly, 67,
79–125. doi:10.1086/346010
Hornet. (2018). About us. Retrieved from https://love.hornet.com/about/
Horvath, K. J., Bowen, A. M., & Williams, M. L. (2006). Virtual and physical venues as contexts
for HIV risk among rural men who have sex with men. Health Psychology, 25, 237–242.
doi:10.1037/0278-6133.25.2.237
Horvath, K. J., Rosser, B. S., & Remafedi, G. (2008). Sexual risk taking among young Internet-
using men who have sex with men. American Journal of Public Health, 98, 1059–1067.
doi:10.2105/AJPH.2007.111070
Hospers, H. J., Kok, G., Harterink, P., & de Zwart, O. (2005). A new meeting place: Chatting on
the Internet, e-dating and sexual risk behaviour among Dutch men who have sex with
men. AIDS, 19, 1097–1101. doi:10.1097/01.aids.0000174457.08992.62
Kalichman, S. C., Kelly, J. A., Morgan, M., & Rompa, D. (1997). Fatalism, current life
satisfaction, and risk for HIV infection among gay and bisexual men. Journal of
Consulting and Clinical Psychology, 65, 542–546. doi:10.1037/0022-006X.65.4.542
Kelly, J., Davis, C., & Schlesinger, C. (2015). Substance use by same sex attracted young
people: Prevalence, perceptions and homophobia. Drug and Alcohol Review, 34, 358–
365. doi:10.1111/dar.12158
Kuhns, L. M., Kwon, S., Ryan, D. T., Garofalo, R., Phillips, G., II, & Mustanski, B. S. (2015).
Evaluation of respondent-driven sampling in a study of urban young men who have sex
with men. Journal of Urban Health, 92, 151–167. doi:10.1007/s11524-014-9897-0
SUBSTANCE USE AND TECHNOLOGY 183
Lachowsky, N. J., Lal, A., Forrest, J. I., Card, K. G., Cui, Z., Sereda, P., … Hogg, R. S. (2016).
Including online-recruited seeds: A respondent-driven sample of men who have sex with
men. Journal of Medical Internet Research, 18, e51. doi:10.2196/jmir.5258
Leslie, J. (2009, October 1). Grindr app makes it easy to be easy. Out & About Nashville.
Retrieved from https://www.outandaboutnashville.com/story/grindr-app-makes-it-easy-
be-easy
Lindhjem, H., & Navrud, S. (2011). Are Internet surveys an alternative to face-to-face interviews
in contingent valuation? Ecological Economics, 70, 1628–1637.
doi:10.1016/j.ecolecon.2011.04.002
Lorimer, K., Flowers, P., Davis, M., & Frankis, J. (2016). Young men who have sex with men’s
use of social and sexual media and sex-risk associations: Cross-sectional, online survey
across four countries. Sexually Transmitted Infections, 92, 371–376.
doi:10.1136/sextrans-2015-052209
MacKellar, D. A., Gallagher, K. M., Finlayson, T., Sanchez, T., Lansky, A., & Sullivan, P. S.
(2007). Surveillance of HIV risk and prevention behaviors of men who have sex with
men—a national application of venue-based, time-space sampling. Public Health
Reports, 122(1, Suppl.), 39–47. doi:10.1177/00333549071220S107
Mansergh, G., Flores, S., Koblin, B., Hudson, S., McKirnan, D., Colfax, G. N., & the Project
MIX Study Group. (2008). Alcohol and drug use in the context of anal sex and other
factors associated with sexually transmitted infections: Results from a multi-city study of
high-risk men who have sex with men in the USA. Sexually Transmitted Infections, 84,
509–511. doi:10.1136/sti.2008.031807
Meyer, I. H., & Wilson, P. A. (2009). Sampling lesbian, gay, and bisexual populations. Journal
SUBSTANCE USE AND TECHNOLOGY 184
of Counseling Psychology, 56, 23–31. doi:10.1037/a0014587
Mutchler, M. G., McKay, T., Candelario, N., Liu, H., Stackhouse, B., Bingham, T., & Ayala, G.
(2011). Sex drugs, peer connections, and HIV: Use and risk among African-American,
Latino, and multiracial young men who have sex with men in Los Angeles and New
York. Journal of Gay & Lesbian Social Services, 23, 271–295.
doi:10.1080/10538720.2011.560100
National Institutes of Health. (2017). NIH categorical spending: Estimates of funding for
variation research, condition, and disease categories. Retrieved from
https://report.nih.gov/categorical_spending.aspx
Pequegnat, W., Rosser, B. R. S., Bowen, A. M., Bull, S. S., DiClemente, R. J., Bockting, W. O.,
… Konstan, J. (2007). Conducting Internet-based HIV/STD prevention survey research:
Considerations in design and evaluation. AIDS and Behavior, 11, 505–521.
doi:10.1007/s10461-006-9172-9
Phillips, G., II, Grov, C., & Mustanski, B. (2015). Engagement in group sex among geosocial
networking mobile application-using men who have sex with men. Sexual Health, 12,
495–500. doi:10.1071/SH15108
Rice, E., Holloway, I., Winetrobe, H., Rhoades, H., Barman-Adhikari, A., Gibbs, J., … Dunlap,
S. (2012). Sex risk among young men who have sex with men who use Grindr, a
smartphone geosocial networking application. Journal of AIDS & Clinical Research, S4,
005. doi:10.4172/2155-6113.S4-005
Ridge, D. T., Plummer, D. C., & Minichiello, V. (1994). Young gay men and HIV: Running the
risk? AIDS Care, 6, 371–378. doi:10.1080/09540129408258651
SUBSTANCE USE AND TECHNOLOGY 185
Schaie, K. W. (Ed.). (1983). Longitudinal studies of adult psychological development. New
York, NY: Guilford Press.
Siegler, A. J., Sanchez, T., Sineath, R. C., Grey, J., Kahle, E., & Sullivan, P. S. (2015).
Knowledge and awareness of acute human immunodeficiency virus infection among
mobile app-using men who have sex with men: A missed public health opportunity. Open
Forum Infectious Diseases, 2, ofv016. doi:10.1093/ofid/ofv016
Sullivan, P. S., Khosropour, C. M., Luisi, N., Amsden, M., Coggia, T., Wingood, G. M., &
DiClemente, R. J. (2011). Bias in online recruitment and retention of racial and ethnic
minority men who have sex with men. Journal of Medical Internet Research, 13, e38.
doi:10.2196/jmir.1797
Wang, S. J., Middleton, B., Prosser, L. A., Bardon, C. G., Spurr, C. D., Carchidi, P. J., … Bates,
D. W. (2003). A cost-benefit analysis of electronic medical records in primary care.
American Journal of Medicine, 114, 397–403. doi:10.1016/S0002-9343(03)00057-3
Zablotska, I. B., Holt, M., & Prestage, G. (2012). Changes in gay men’s participation in gay
community life: Implications for HIV surveillance and research. AIDS and Behavior, 16,
669–675. doi:10.1007/s10461-011-9919-9
Zou, H., & Fan, S. (2017). Characteristics of men who have sex with men who use smartphone
geosocial networking applications and implications for HIV interventions: A systematic
review and meta-analysis. Archives of Sexual Behavior, 46, 885–894.
doi:10.1007/s10508-016-0709-3
SUBSTANCE USE AND TECHNOLOGY 186
Table 6.1. Recruitment Period Statistics

App Venue

(n = 18) (n = 59)

n or M % or SD n or M % or SD t(df) or χ
2

Periods per month 3.6 -- 8.43 -- --
Months to complete recruitment 5 -- 7 -- --
Observed potential participants 46.06 41.43 13.49 12.808 3.197(16.89)**
Observed unique potential
participants
34.56 34.53 12.69 12.378 2.574(17.2)*
Completed eligibility survey
a
18.78 17.21 5.56 5.688 3.206(18.146)**
Eligible for participation 4.39 3.26 3.02 4.028 1.318(75)
Completed main survey 2.06 1.924 1.15 1.827 1.813(75)
Completed supplemental survey 1.78 1.629 0.92 1.633 1.963(75)
Eight or more potential
participants
17 100.00 34 57.63 10.734**
Eight or more unique potential
participants
14 82.35 31 52.54 4.856*
a
Includes app users of all ages (3,043).
*p < .05. **p < .01.

SUBSTANCE USE AND TECHNOLOGY 187
Table 6.2. Summary of Recruitment Method Work Hours

App Venue App vs. Venue
a

Recruitment preparation

Doctoral candidate administration 51 83 61.45
Doctoral candidate fieldwork 48 12 400.00
Student or temporary worker fieldwork 0 480

Training students or temporary workers 0 27

Community advisory board 10 10 100.00
Pilot testing 35 57 61.40
Preparation subtotal 144 669 21.47
Recruitment implementation

Doctoral candidate administration 28 42 66.67
Doctoral candidate fieldwork 6 0

Student or temporary worker fieldwork 0 415

Student or temporary worker administration 60 99 60.61
Implementation subtotal 94 556 16.91
Total cost 238 1,225 19.40
Work hours per participant 6 18 30.67
Note. Sample sizes: venue n = 68, app n = 43.
a
Figures indicate percentage of time spent in app-based recruitment relative to venue-based
recruitment.

SUBSTANCE USE AND TECHNOLOGY 188
Table 6.3. Summary of Recruitment Method Costs

App Venue App vs. Venue
a

Recruitment preparation

Doctoral candidate administration $1,275.00 $2,075.00 61.45
Doctoral candidate fieldwork $1,200.00 $300.00 400.00
Student or temporary worker fieldwork $0.00 $8,835.29

Training students or temporary workers $0.00 $451.50

Community advisory board $412.50 $412.50 100.00
Pilot testing $1,155.42 $939.46 122.99
Preparation subtotal $4,042.92 $13,013.75 31.07
Recruitment implementation

Doctoral candidate administration $700.00 $1,050.00 66.67
Doctoral candidate fieldwork $150.00 $0.00

Student or temporary worker fieldwork $0.00 $7,425.85

Student or temporary worker administration $900.00 $1,753.94 51.31
Advertisements $500.00 $0.00

Incentives $1,775.00 $4,390.00 40.43
Implementation subtotal $4,025.00 $14,619.79 27.53
Software $200.00 $0.00

Hardware $299.00 $2,808.00 10.65
Total cost $6,791.92 $26,051.54 26.07
Cost per participant $157.95 $383.11 41.23
Note. Sample sizes: venue n = 68, app n = 43.
a
Figures indicate percentage of cost spent on app-based recruitment relative to venue-based recruitment.

SUBSTANCE USE AND TECHNOLOGY 189
Figure 6.1. Recruitment Flow Chart for Venue-Based and App-Based Recruitment

Observed
(n = 796)
Approached
(n = 749)
Duplicates (n = 47)
Completed Eligibility
(n = 328)
Declined (n = 312)
Unable to make contact (n = 111)
Eligible
(n = 178)
Ineligible
(n = 150)
Completed Main
Survey
(n = 68)
Completed
Supplemental Survey
(n = 54)
94%
44%
54%
38%
79%
24%
9%
7%
44%
Observed
(n = 783)
Unique Users
(n = 589)
Duplicates (n = 193)
Completed Eligibility
(n = 338)*
Eligible
(n = 62)
Ineligible
(n = 259)*
Completed Main
Survey
(n = 37)
Completed
Supplemental Survey
(n = 32)
75%
60%
86%
11%
6%
5%
*Of 3,043 GSNA users not limited by reported age
Venue-Based Recruitment App-Based Recruitment
SUBSTANCE USE AND TECHNOLOGY 190
Figure 6.2. Cost per Participant by Recruitment Method based on Study Sample Size

30 50 100 150 200 250 300 400 500 600 700 800 900 1000
App $228.55 $170.05 $126.17 $111.55 $104.23 $99.85 $96.92 $93.27 $91.07 $89.61 $88.57 $87.78 $87.17 $86.68
Venue $602.03 $441.04 $320.30 $280.05 $259.93 $247.85 $239.80 $229.74 $223.70 $219.68 $216.80 $214.65 $212.97 $211.63
$0.00
$100.00
$200.00
$300.00
$400.00
$500.00
$600.00
$700.00
Cost in US Dollars (US$)
Sample Size N
Figure 6.2: Cost per Participant per Recruitment Method based on the Study Sample
Size (N)
Figure Assumptions: Standard recruitment prep. without pilot (venue=$12,074.29, App=$2,887.50); Venue eligibility incentive $5 with rate of 4.824
assessed per participant; standard app broadcast advertisement cost of $1,500; different survey incentive venue=$25 and app=$35, standard
recruitment implementation cost without advertisement or incentives (venue=$150.44, app=$47.30 per participant)
SUBSTANCE USE AND TECHNOLOGY 191
Chapter 7: Conclusions
The overarching goal of this dissertation was to test a novel method for recruiting young
men who have sex with men (YMSM) to participate in research: geosocial networking
application (GSNA)-based probability sampling. This investigation included three tests: (a)
whether the methods can be developed, (b) how a sample recruited using these methods
compares to a sample recruited using the most rigorous conventional method, and (c) how this
novel recruitment methodology influences the cost and recruitment efficiency of YMSM
probability sampling. The findings from these three investigations taken together present five
implications for the types of YMSM who can be recruited using GSNA, the nature of health risk
for YMSM, the accessibility of YMSM, the recruitment and cost efficiency of research with
YMSM, and the ethics of YMSM recruitment.
No significant differences in sexual risk and psychosocial measures suggests a similar
sample can be achieved using a GSNA compared to venue-based procedures. Although there
were significant differences in demographics between the two samples of men recruited, these
differences may be largely due to the diversity and range of individuals who can be recruited
using each methodology. GSNA-based probability sampling appears to recruit more diverse
populations of YMSM, including men who are more likely to be people of color, less “out” about
their sexual orientation, and more varied in their education level and employment status. There
was also no indication that men recruited through the apps are more likely to be involved in risky
sex practices compared to venue-recruited samples. This finding provides counterevidence for
studies of venue-recruited YMSM that indicated a higher probability of sexual risk for those who
use apps (Eaton et al., 2016; Lehmiller & Ioerger, 2014; Phillips et al., 2014). Because the
higher-risk men identified in these studies were both venue and app users, it is reasonable that
SUBSTANCE USE AND TECHNOLOGY 192
individuals who are using both apps and venues to meet sexual partners represent a higher risk
profile. These research questions can be further investigated using samples of men recruited from
GSNAs.
Further, this study revealed that the most relevant and timely research question is not “Do
apps lend themselves to risk?” (as many research studies have previously investigated; Eaton et
al., 2016; Lehmiller & Ioerger, 2014; Phillips et al., 2014), but rather “How has the accessibility
of apps changed the way we understand risk among YMSM?” For example, whereas previous
research identified gay community connection as a risk factor for increased alcohol consumption
(Stall et al., 2001), the data from the current study indicate that for GSNA-recruited men, gay
community connection is a protective factor against both higher rates alcohol consumption and
binge drinking. The connections formed and sense of knowing that peers are accessible in any
location may change how gay community connection affects substance risk. Further, connecting
with other sexual minority men outside of a bar environment may provide this community with a
mechanism for severing the cultural norm of socializing while using substances, a difficulty
noted by sexual minority men pursuing sobriety (Matthews, Lorah, & Fenton, 2006; Trussler,
Perchal, & Barker, 2000). This venue culture of substance use may also explain why the venue-
recruited sample reported using alcohol and other substances at higher rates compared to the
GSNA-recruited sample.
Evidence from this study indicates that YMSM are using gay venues less and instead
using GSNAs to connect. Based on reports from the venue sample, at least two thirds of venue-
attending YMSM would be accessible via GSNAs, confirming previous findings from a study
implemented in Washington, DC, with venue-attending men who have sex with men (MSM;
Phillips et al., 2014). However, this estimate from Washington, DC, was based on data collected
SUBSTANCE USE AND TECHNOLOGY 193
from YMSM who attend venues. By excluding YMSM who only use apps and do not attend
venues, it is likely that this finding systematically underestimated the proportion of YMSM who
use GSNAs on a regular basis.
However, a conservative estimate and a liberal estimate can be determined using the data
from the current study. Assuming that the Hornet-recruited sample is largely representative of
samples recruited from all major gay GSNAs and that the venue sample and the GSNA sample
represent YMSM as a whole, an estimate of the number of YMSM who use GSNAs can be
calculated. Findings indicate that 33% of GNSA-recruited YMSM never go to gay bars and clubs
and 51% report going to gay bars and clubs once or less a month. These statistics serve as the
conservative and liberal thresholds of the estimate; of GSNA-recruited YMSM either
conservatively, 33% (a) do not and 67% (b) do attend venues on a regular basis, or liberally,
84% (c; 33%+51%) do not and 16% (d) do attend venues on a regular basis. Given these
conservative (67% attend venues) and liberal (16% attend venues) estimates, two calculations
can be made. Results indicate that 67% (e) of the venue-recruited YMSM use GSNAs on a
regular basis, which means 33% (f) of the venue-recruited sample can only be located in or near
gay venues. Therefore, 67% of the venue sample and 67% (conservative estimate) or 16%
(liberal estimate) of the app sample theoretically represent the same men, those who both use
apps and attend venues. To determine the proportion of the YMSM population that only uses
venues, only uses apps, and uses both venues and apps, the denominator for each sample
proportion needs to be the same. Therefore the conservative estimate for the population that only
uses venues is 25% (i.e., proportion = [(f ÷ e) × (b ÷ 1) ÷ [(f ÷ e) × (b ÷ 1) + b + a] or [(.33 ÷
.67) × .67] ÷ [(.33 ÷ .67) × .67 + .67 + .33]); that only uses apps is 25% (i.e., proportion = [a ÷ [(f
÷ e) × (b ÷ 1) + b + a] or [.33 ÷ [(.33 ÷ .67) × .67 + .67 + .33]); and that uses both apps and
SUBSTANCE USE AND TECHNOLOGY 194
venues is 50% (i.e., proportion = [e ÷ [(f ÷ e) × (b ÷ 1) + b + a] or [.67 ÷ [(.33 ÷ .67) × .67 + .67
+ .33]). The liberal estimate indicates that 7% only use venues (i.e., proportion = [(f ÷ e) × (d ÷
1) ÷ [(f ÷ e) × (d ÷ 1) + d + c] or [(.33 ÷ .67) × .16] ÷ [(.33 ÷ .67) × .16 + .16 + .84]); 78% only
use apps (i.e., proportion = [c ÷ [(f ÷ e) × (d ÷ 1) + d + c] or [.84 ÷ [(.33 ÷ .67) × .16 + .16 +
.84]); and 15% use both apps and venues (i.e., proportion = [d ÷ [(f ÷ e) × (d ÷ 1) + d + c] or [.16
÷ [(.33 ÷ .67) × .16 + .16 + .84]). Table 7.1 summarizes these population estimates.
Based on the findings from this study, between 75% (i.e., 25% + 50%) and 93% (i.e.,
78% + 15%) of YMSM are estimated to use GSNAs on a regular basis, whereas an estimated
25% to 78% only use GSNAs and do not attend gay bars or clubs. Understanding where men can
be located should guide recruitment method selection. Future research can improve these
estimates by including items on venue attendance and app use in survey research.
Based on these estimates, GSNA-based recruitment can reach a majority of the YMSM
population and is more cost efficient than venue-based procedures. Further, it is more
straightforward and less burdensome to enlarge the geographic sampling frame (GSF) of a
GSNA-recruiting study compared to a venue-recruiting study, which requires additional
enumeration periods. For larger GSFs, GSNA-based methods are more feasible to implement. At
the national and state levels, where venue-based sampling may not be possible, GSNA-based
methods appear to be feasible. Although the application of these methods in larger areas is a
strength, at a neighborhood level, GSNA-based sampling may not be as feasible as venue-based
recruitment. The pilot testing of the GSNA methods revealed that although the original GSF
(GSF A) was in an area with a high density of gay venues, user density did not seem to be
similarly high. Therefore, for studies in which YMSM from a specific neighborhood or small
geographic area are being targeted, venue-based procedures may be more appropriate.
SUBSTANCE USE AND TECHNOLOGY 195
Two ethical issues became apparent during the current study. First, venue-based
recruitment relies on the study personnel to accurately identify YMSM using visual cues.
Although the team for the current study was able to correctly identify eligible participants 54%
of the time (excluding those who the study team was not able to approach or who declined to
complete the eligibility survey), the study was not able to track the number of individuals who
the study team misidentified as being ineligible and were therefore not approached to take part in
the study. During recruitment, the study team was instructed to approach individuals who
appeared to fit the age constraints of the study and at no point was encouraged to determine
eligibility based on whether the individual appeared to be a man who has sex with men. If study
team members were instructed to target men who appeared to be sexual minority men, they
would be relying on stereotypical presentations of sexuality that would systematically exclude
sexual minority men who do not present in this way. Although the study team was instructed to
limit observations to age, it is reasonable to assume that sexual orientation presentation may have
guided determinations of individuals’ study eligibility. An appropriate or ethical way of training
study personnel members to accurately identify sexual minority men is difficult to imagine.
Fundamentally, reliance on human perception to identify sexual orientation, or even age, is
flawed and a considerable weakness of venue-based recruitment. However, this same weakness
is a strength of GSNA-recruitment. Reliance on the age information provided in a public profile
is not only reasonably more reliable, but ethically less problematic. Additionally, when men are
recruited on applications that largely cater to MSM, the study team is not required to use visual
data to determine eligibility. Instead, in the current study, eligibility for the GSNA-recruited
sample was exclusively determined by answers to the eligibility survey.
SUBSTANCE USE AND TECHNOLOGY 196
The second ethical issue relates to publishing substance use findings that potentially
further stigmatize a population. The current study found that substance use estimates are higher
in a venue-recruited sample compared to a GSNA-recruited sample. Although conclusions about
the generalizability of one sample compared to the other are discouraged, these results suggest
that recruiting samples of YMSM from venues that cater to substance use systematically
excludes YMSM who use substances at lower rates. A study of a venue-based probability sample
of 528 YMSM living in Los Angeles, much like the subsample from the current study, found that
YMSM reported consuming alcohol on an average of 8 days during the prior 30 days (Wong,
Kipke, & Weiss, 2008). This finding is confirmed by the results of the venue-based sample from
the current study (8 days on average). However, the current study also found that this statistic is
twice the rate of alcohol consumption for YMSM recruited through GSNAs (4 days on average).
Generalizing the substance use characteristics of YMSM recruited exclusively from venues to
the entire population of YMSM is not only irresponsible but socially unjust. Reporting inflated
substance use characteristics as if they represent the full population of YMSM continues a deficit
expectation. Therefore, a concerted effort must be made to clearly state the limitations of
generalizability and frame findings in the context of potential methodological bias.
Although this study introduces a novel method for recruiting YMSM to participate in
research and highlights the impact of recruitment method on the demographic and behavioral
characteristics of the study sample, there are several limitations to note. The methods tested in
this study were implemented in a large urban area, and therefore very little is known about the
application of these methods in rural areas. Specifically, it is unknown whether the user density
enumeration would be as accurate or achievable in areas of lower population density. However,
due to the limited accessibility of gay venues in rural areas, recruitment through GSNAs may
SUBSTANCE USE AND TECHNOLOGY 197
offer opportunities to include marginalized rural MSM in exploratory and intervention research.
In addition, due to the cross-sectional nature of this study, very little is known about the
feasibility of implementing GSNA-based recruitment in longitudinal research. However, given
that a higher percentage of the GSNA-recruited sample completed the supplemental survey
(administered within a week of the main study) compared to the venue-recruited sample, there is
evidence that follow-up surveying with GSNA-participants may be feasible. Finally, although
this sample is likely to be representative of YMSM who can be recruited from venues and
Hornet, there are likely YMSM who do not use apps and also do not attend venues. Further,
although research has indicated that YMSM are increasingly using GSNAs to find one another
(Grov, 2012; Grov & Crow, 2012; Zablotska, Holt, & Prestage, 2012), these men spread their
use among multiple apps. By using only one GSNA in recruitment, the current study
systematically excluded YMSM who use other apps. To date, very little is known about the
demographics and behavior profiles of YMSM recruited through different GSNAs
(Mastroyiannis, 2018). Therefore, the study findings should be interpreted in the context of these
limitations to generalizability.
The findings from this study can inform future research recruitment and targeted
intervention with YMSM. Although a venue-recruited sample may present with increased
substance use, an app-recruited sample might present with more diversity and limited outness.
Due to the nature of this study, conclusions cannot be drawn regarding the generalizability of
either sample. Future research should compare national samples of men drawn from either
recruitment strategy to statistics in nationally representative samples (e.g., Youth Risk Behavior
Survey, National Longitudinal Study of Adolescent to Adult Health) and weight the data
accordingly. Although the GSNA-based methods used in this study are replicable at a national
SUBSTANCE USE AND TECHNOLOGY 198
level, venue-based probability sampling procedures applied at a national level would be
expensive and difficult to implement (Meyer & Wilson, 2009). Additionally, research is needed
to determine the feasibility of using GSNA recruitment methods to study YMSM in rural areas
and developing countries. Application of these methods in service- and community resource-
impoverished areas could meaningfully improve what is known about the most marginalized
sexual minority populations. Last, the methods used for enumerating user density are applicable
outside of study recruitment. Because user density data is geographically bound, it can very
easily be linked with state and national datasets (e.g., census data) to explore the impact of
system-level factors on where MSM reside. This linked data can then help guide targeted health
intervention and implementation with MSM.
With the introduction of GSNAs, the way YMSM interact with and advertise their
connection to the gay community has changed. For research methods to remain relevant, they
must evolve and adapt to this change without sacrificing rigor. The methods and findings
presented in this dissertation not only encourage this evolution, but significantly reduce the costs
of recruitment implementation. By reducing cost and increasing accessibility of probability-
based methods, more rigorous research can be conducted to determine the factors that contribute
to YMSM health disparities.

SUBSTANCE USE AND TECHNOLOGY 199
References
Eaton, L. A., Maksut, J. L., Gamarel, K. E., Siembida, E. J., Driffin, D. D., & Baldwin, R.
(2016). Online sex partner meeting venues as a risk factor for testing HIV positive among
a community-based sample of black men who have sex with men. Sexually transmitted
diseases, 43, 360–364. doi: 10.1097/OLQ.0000000000000454
Grov, C. (2012). HIV risk and substance use in men who have sex with men surveyed in
bathhouses, bars/clubs, and on Craigslist.org: Venue of recruitment matters. AIDS and
Behavior, 16, 807–817. doi:10.1007/s10461-011-9999-6
Grov, C., & Crow, T. (2012). Attitudes about and HIV risk related to the “most common place”
MSM meet their sex partners: Comparing men from bathhouses, bars/clubs, and
Craigslist.org. AIDS Education and Prevention, 24, 102–116.
doi:10.1521/aeap.2012.24.2.102
Lehmiller, J. J., & Ioerger, M. (2014). Social networking smartphone applications and sexual
health outcomes among men who have sex with men. PLOS One, 9, e86603.
doi:10.1371/journal.pone.0086603
Mastroyiannis, A. (March 5, 2018). Gay dating apps: A comprehensive guide to Jack’d, Ginrd,
Hornet, Scruff, and the rest. Pink News. Retreived from:
https://www.pinknews.co.uk/2018/03/05/best-gay-dating-apps-jackd-grindr-hornet-
scruff/
Matthews, C. R., Lorah, P., & Fenton, J. (2006). Treatment experiences of gays and lesbians in
recovery from addiction: A qualitative inquiry. Journal of Mental Health Counseling, 28,
111–132. doi:10.17744/mehc.28.2.9m35re2aj9l28j47
Meyer, I. H., & Wilson, P. A. (2009). Sampling lesbian, gay, and bisexual populations. Journal
SUBSTANCE USE AND TECHNOLOGY 200
of Counseling Psychology, 56, 23–31. doi:10.1037/a0014587
Phillips, G., II, Magnus, M., Kuo, I., Rawls, A., Peterson, J., Jia, Y., … Greenberg, A. E. (2014).
Use of geosocial networking (GSN) mobile phone applications to find men for sex by
men who have sex with men (MSM) in Washington, DC. AIDS and Behavior, 18, 1630–
1637. doi:10.1007/s10461-014-0760-9
Stall, R., Paul, J. P., Greenwood, G., Pollack, L. M., Bein, E., Crosby, G. M., ... & Catania, J. A.
(2001). Alcohol use, drug use and alcohol-related problems among men who have sex
with men: the Urban Men’s Health Study. Addiction, 96(11), 1589-1601.
Trussler, T., Perchal, P., & Barker, A. (2000). ‘Between what is said and what is done’: Cultural
constructs and young gay men’s HIV vulnerability. Psychology, Health & Medicine, 5,
295–306. doi:10.1080/713690193
Wong, C. F., Kipke, M. D., & Weiss, G. (2008). Risk factors for alcohol use, frequent use, and
binge drinking among young men who have sex with men. Addictive Behaviors, 33,
1012–1020. doi:10.1016/j.addbeh.2008.03.008
Zablotska, I. B., Holt, M., & Prestage, G. (2012). Changes in gay men’s participation in gay
community life: Implications for HIV surveillance and research. AIDS and Behavior, 16,
669–675. doi:10.1007/s10461-011-9919-9

SUBSTANCE USE AND TECHNOLOGY 201
Table 7.1. Population Estimates of App and Gay Venue Use among YMSM

Conservative Liberal
% %
Venue sample

Only use venues 33 (f) 33 (f)
Use apps and venues 67 (e) 67 (e)
App sample

Only use apps 33 (a) 84 (c)
Use apps and venues 67 (b) 16 (d)
YMSM population estimates

Only use venues 25 7
Only use apps 25 78
Use apps and venues 50 15
Use venues 75 22
Use apps 75 93
Note. Letters in parentheses correspond with formulas in the text.

Abstract (if available)

Abstract Young men who have sex with men (YMSM

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Social network and contextual influences on substance use and HIV risk behavior among young men who have sex with men

PDF

Impact of change in sexual identity on mental health risks among sexual minority adolescents

PDF

Sex talks: an examination of young Black women's social networks, sexual health communication, and HIV prevention behaviors

PDF

Homelessness and substance use treatment: using multiple methods to understand risks, consequences, and unmet treatment needs among young adults

PDF

Growing up with Alphas: how animation marketers can survive the next generation

PDF

Complicated relationships: the prevalence and correlates of sexual assault, dating violence victimization, and minority stress among sexual minority adolescents throughout the United States

PDF

Tobacco use change among formerly homeless supportive housing residents: socioecological barriers and facilitators to cessation

PDF

A mixed-methods examination of transgender and gender nonbinary students’schooling experiences: implications for developing affirmative schools

PDF

Stigma-free pregnancy: a recruitment and retention strategy for healthcare systems to engage pregnant women with substance use disorder in collaborative care

PDF

WISER women’s program: well-being innovation with support and education for resilience—a homelessness prevention intervention

PDF

Main, moderating, and mediating effects of social network compositions on suicidal ideation among North Korean refugees in South Korea

PDF

Contextualizing social network influences on substance use among high risk adolescents

PDF

Distress tolerance and mindfulness disposition: associations with substance use during adolescence and emerging adulthood

PDF

Disclosure of lesbian, gay, bisexual, and transgender (LGBT) identity among U.S. service members after repeal of LGBT military service bans

PDF

Interagency collaboration: cultivating resources and strengthening substance use service delivery in child welfare

PDF

Social network engagement and HIV risk among homeless former foster youth

PDF

A school for implementing arts and technology: an innovation study

PDF

Preventing exploitation of people with intellectual and developmental disabilities (I/DD) healthy and unhealthy relationships educational workshops sustaining innovation: healthy relationships matter

PDF

Myths and taboos: deconstructing patriarchy through intimate partner violence awareness

PDF

SCALE UP: an integrated wellness framework for schools

Asset Metadata

Creator Gibbs, Jeremy J. (author)

Core Title Substance use and technology: testing an innovative method for recruitment of young men who have sex with men

School School of Social Work

Degree Doctor of Philosophy

Degree Program Social Work

Publication Date 07/13/2018

Defense Date 05/07/2018

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag geosocial networking applications, methodology,OAI-PMH Harvest,sexual minorities,substance use,Technology,YMSM,young adult

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Goldbach, Jeremy (committee chair), Echols, Alice (committee member), Rice, Eric (committee member), Traube, Dorian (committee member)

Creator Email jeremyjg@usc.edu,jeremyjgibbs@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-16310

Unique identifier UC11671458

Identifier etd-GibbsJerem-6397.pdf (filename),usctheses-c89-16310 (legacy record id)

Legacy Identifier etd-GibbsJerem-6397.pdf

Dmrecord 16310

Document Type Dissertation

Format application/pdf (imt)

Rights Gibbs, Jeremy J.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA