University of Southern California Dissertations and Theses

“Under the radar”: the relative effectiveness of gay-vague, gay-explicit, and control anti-smoking messages for targeted and nontargeted audiences

(USC Thesis Other)

“Under the radar”: the relative effectiveness of gay-vague, gay-explicit, and control anti-smoking messages for targeted and nontargeted audiences

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

“UNDER THE RADAR”: THE RELATIVE EFFECTIVENESS OF GAY-VAGUE,
GAY-EXPLICIT, AND CONTROL ANTI-SMOKING MESSAGES FOR TARGETED
AND NONTARGETED AUDIENCES

by

Linda Alexander

__________________________________________________________________

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
(COMMUNICATION)

December 2009

Copyright 2009 Linda Alexander

ii
TABLE OF CONTENTS

List of Tables iii

List of Figures iv

Abstract v

Chapter One 1

Chapter Two 18

References 56

iii
LIST OF TABLES

Table 1: First Appearances of Cigarette Ads in Gay Publications in the 1990s 4
Table 2: Results By Sample and Comparison With Control 31
Table 2.1: Comparisons Between Gay Vague and Gay Explicit 35
Messages By Sample

iv
LIST OF FIGURES

Figure 1: Parliament Lights ad that appeared in the May 1999, issue of 7
Out magazine

Figure 2: Symbol Recognition 28

v
ABSTRACT

Smoking kills over 440,000 individuals annually. Lesbian, gay, bisexual, and
transgendered (LGBT) individuals, compared to non-LGBT individuals, are especially
likely to smoke. Anti-smoking messages that are effective for LGBT individuals, but that
also reach non-LGBT individuals would be desirable: But, to date, no research has
specifically examined this. Based on theoretical foundations developed by psychologists
and marketing researchers, two studies (one each for LGBT and non-LGBT) were
conducted. In Study 1, using a within subjects design, 61 LGBT individuals viewed eight
messages that differed in whether they specifically targeted LGBT individuals (in an
explicit or vague way) or did not (control). The dependent measures were the perception
that a message was intended for the individual, message effectiveness, and the likelihood
of sharing anti-smoking concerns with others after exposure to a message. To avoid
order effects, a balanced Latin square design was used such that each message appeared
before and after every other message. Study 2 used the same design and stimulus
materials as in Study 1: However, the 209 participants were heterosexuals. Analyses
revealed that homosexuals and heterosexuals differ in their patterns of responses to these
messages. An intriguing finding was that when both LGBT and non-LGBT audiences are
exposed to the same health message, “gay-vague” messages that provide symbols but no
explicit text might best facilitate LGBT receptivity and mitigate non-LGBT reactivity.
Based on these findings, public health campaigns that target a minority population may
be able to reach non-target as well as target populations by using less explicit messages.

1
CHAPTER ONE

Although smoking kills more than 440,000 people in the U. S. every year (CDC,
“Tobacco-Related Mortality”), research indicates that lesbian, gay, bisexual, and
transgendered (LGBT) individuals smoke more than the general population in the U.S.
(Gruskin, et al., 2007; Ryan, et al., 2001; Stall et al., 1999). And among Californians, a
recent survey conducted by the California Department of Health Services revealed that
LGBT members’ smoking rates are double that of the state’s general population (30.4%
vs. 15.4%). Moreover, in terms of a gender breakdown, LGBT women smoke three
times more than the women in California’s general population (32.5% vs. 11.9%) and
27.4% of LGBT men smoke, compared to 19.1 % of general population of men. While
these statistics are disturbing, perhaps the most alarming finding is that almost half of
LGBT youth (18-24 year olds) smoke, which is 2.5 times more than the general
population in this age range (43.7% vs. 16.6%).

Targeting the LGBT Community With Pro-Tobacco and Anti-Tobacco Messages

The Tobacco Companies and the LGBT Community: An Historical Perspective
While we cannot say for sure why so many LGBT people are smokers, it should
be noted that based on the tobacco industry’s own internal documents made publicly
available as a result of the 1998 Master Settlement Agreement between tobacco
companies and attorney generals of 46 states (Tobacco Master Settlement Agreement
Summary), tobacco companies have actively courted members of the LGBT community

2
through the years by forging political alliances with the community, supporting gay
causes and events, and advertising in gay media (see Alexander, 2005, for a review).
Although a Philip Morris company spokesman once said: “Gays and lesbians
have not been part of our research and have not been a consideration” (Wilke, 2001),
there is evidence to the contrary. In the past, Philip Morris
1
has used Direct Male, a gay-
oriented marketing group, to collect survey information from smokers because, according
to a Philip Morris representative, “It was an inexpensive way to reach adult males—with
a high response rate” (Horovitz, 1993). And in February 1994, Philip Morris hired
Guiles & Associates to conduct market research on gay male smokers in San Francisco.
The firm produced a report entitled: “Exploratory Qualitative Research: Benson &
Hedges in the Gay Market.”
2
Moreover, between 1995 and 1997, R. J. Reynolds laid out
a plan to market Camel cigarettes to subcultures in San Francisco. The rather unfortunate
name assigned to this endeavor was Project Scum (“subculture urban market”).
3
It was
also clear from a 1997 Phillip Morris internal memo that the company was well aware
that “a large percentage of Gays and Lesbians are smokers [and]…. are good prospects

1
Recently, the parent company of the Philip Morris companies went through a name
change; it is now called Altria Group, Inc.. In this paper, however, the name “Philip
Morris” will be used.

2
File document. Guiles & Associates report for Philip Morris regarding February 10,
1994, research on the gay market. www.pmdocs.com/ DOCID= 2047277153/75.

3
File document. R. J. Reynold’s proposed marketing plan entitled “Project Scum.”
www.rjrtdocs.com/ Two versions: 1995: RJR DOCID=518021121/1129. 1997: RJR
DOCID: 517060265/0272.

3
for the Benson & Hedges brand”
4
and had begun to act on this knowledge. By placing
ads for Special Kings cigarettes in Genre magazine in 1992, Philip Morris became the
first tobacco company to advertise in a gay publication (Costantino, “Gays Teens and
Smoking”). Other tobacco companies soon followed suit (see Table 1).

4
File document. Philip Morris interoffice correspondence dated October 9, 1997 from
Bert Johnson to Yvette Robinson and May Nivar regarding Community Event Marketing
(CEM) for gays and lesbians. www.pmdocs.com/ DOCID=2071145104.

4

Table 1:
First Appearances of Cigarette Ads in Gay Publications in the 1990s
________________________________________________________________________

Date Brand Parent Company
________________________________________________________________________

Fall 1992 Benson & Hedges (Special Kings) Philip Morris
May 1994 Parliament Philip Morris
November 1994 Benson & Hedges* Philip Morris
February 1996 Moonlight Tobacco R.J. Reynolds
March 1996 Red Kamel Brown & Williamson
April 1996 Carlton R.J. Reynolds
September 1996 Merit Philip Morris
November 1996 Lucky Strike Brown & Williamson
May 1997 Kool Brown & Williamson
June 1997 Kamel R.J. Reynolds
June 1998 Camel Lights R.J. Reynolds

Note. From Wilke (2001). Wilke states that the first Benson & Hedges appearance in a
gay publication took place in 1994. His original list did not include the 1992 Special
Kings ad in Genre.

5
Anti-smoking Ads Targeting the LGBT Community are Few and Their Effects Are
Unknown
While tobacco companies are trying to encourage members of the LGBT
community to use their products, fewer and less well-funded groups are trying to
intervene to dissuade members of the LGBT community from smoking. For example, the
American Legacy Organization has created anti-smoking media campaigns that target
gays and lesbians (americanlegacy.org)
Nevertheless, these ads are few and far between and their effects are unknown.
The relative lack of research is understandable. Even though members of the gay
community smoke disproportionately more than everyone else, they are not a particularly
large subpopulation, numbering between 13 to 15 million Americans (Yin, 2003).
Moreover, it is very costly to make health campaigns happen that might specifically
target and reach gay and lesbian smokers.
There is an additional problem. When faced with a message that explicitly targets
a minority group, those who are not group members might either ignore the message
(“it’s not for people like me”), or in the worst-case scenario, have a negative response,
such as a backlash against the sponsoring organization. This has been the case for
companies that have marketed products to gays and lesbians in years past (Johnson,
1999).

6
Flying Under the Radar: Can We Use Less Explicitly Targeted Messages That Reach the
Desired Minority Audience But Do Not “Turn Off” the Mainstream Audience?
The primary aim of the current research was to address the challenges of creating
health messages that will resonate with a target audience without alienating members of
the non-target audience. One way to do this is to make the act of targeting a particular
group less explicit. It has been a fairly common practice for companies who want gay
dollars to use what some have called “gay-vague” ads in order to avoid a potential
backlash from heterosexuals. “Gay-vague” or “gay window” ads are recognized by
members of the LGBT community as having subtle gay content which heterosexuals tend
to miss (Alsop, 1999; Commercial Closet; Sender, 1999). An example of this is to
feature crowd scenes made up of attractive people who are not explicitly paired up,
thereby making it impossible to determine their true relationships to each other.
According to consultant William Waybourn of Windows Communications, a firm that
works with national corporations in advertising to the LGBT community, there is a real
plus in highlighting ambiguous relationships, such that when there are two men and a
woman in an ad, LGBT viewers see a gay couple with a friend, while non-LGBT viewers
only see a group of three people (Rothman, 1999). Figure 1 is an example of this
scenario.

7

Figure 1: Parliament Lights ad that appeared in the May 1999, issue of Out magazine

8

Subtle gay-oriented cues can also be in the form of text. In terms of gay-vague
health messages, a recent anti-smoking campaign features the tag line: “Learn how to
quit smoking without being scared straight. [As if you could be]” (Last Drag LA). This
suggests to LGBT individuals that being gay is not a choice, so you could never be
“straight” (heterosexual). Similarly, the city of San Diego used the line “Come out and
be yourself” in a promotion to attract LGBT visitors (Hansen, 2004), presumably to not
anger those who aren’t gay-friendly.
Benetton’s director of communications for North America, Peter Fressola, says
that another way in which marketers attempt to reach a gay audience is to use gay-
oriented images and symbols in ads, such as a rainbow motif or pink triangles. For
example, a travel-service in Lebanon is very careful about how it markets itself in a
conservative Muslim climate. Their advertising is mostly online, and their ads are quite
subtle; the word “gay” is never used. Instead, the ads feature a line of rainbow colors,
which is a very well known gay cultural symbol (Fassihi, 2007).
But even when ads contain only subtle gay cues, they might work. Sender’s
(1999) focus group research revealed that gay, lesbian, and bisexual participants
interpreted these kinds of ads as gay more often than the heterosexual participants in her
study. This is in line with Grier and Brumbaugh’s (1999) finding that those in their
study’s sample who were not gay or lesbian had difficulty in decoding symbols
associated with the LGBT community.

9
The problem for marketers is that if ads target LGBT members too explicitly, it
can result in a backlash, as mentioned earlier. There are those out there who are very anti-
gay, and their response to ads that are perceived to be targeting the LGBT community is
very different from those who are gay-friendly. For example, Bhat, Leigh, and Wardlow
(1998) found that when their participants were shown ads with gay and lesbian romantic
couples, those who were highly prejudiced toward LGBT people were more likely to
dislike the ads compared to those who were low in LGBT prejudice.

Current Work: Relevant Theory and Past Systematic Research
The current research had two goals. First, to assess what types of messages might
be most likely to be positively evaluated by members of the LGBT community. Second,
to assess what messages, that the non-LBGT community might see, that would not be
likely to “backfire” and/or adversely impact the smoking attitudes of non-LGBT
members. The theory and research most relevant to these goals is reviewed below.
Distinctiveness Theory
Distinctiveness theory was initially developed and empirically tested from the late
‘50s through the early ‘80s. The basis for distinctiveness theory is the idea that people
tend to selectively perceive things about a particular object that are unpredictable,
unusual or distinctive. What is not distinctive is ignored because it is more cognitively
efficient to do so (McGuire, 1999). At the outset, the focus of this perspective was on
self-perception, such that what makes you stand out in a crowd is what you feel most
defines you. This leads to describing yourself based on what is most notable, or

10
distinctive, about you. According to McGuire and Padawer-Singer (1976), “each person
is his or her own most complex stimulus object….[and] we notice any aspect (or
dimension) of ourselves to the extent that our characteristic on that dimension is peculiar
in our social milieu” (p. 744). Over the years, research on distinctiveness theory has
examined the effect of one’s gender and ethnicity (McGuire & Padawer-Singer;
McGuire, McGuire, Child, & Fujioka, 1978), and more recently, one’s sexual orientation
(e.g., Aaker, Brumbaugh, & Grier, 2000).
At some point early on, research attention turned to the perception of objects other
than the self (e.g., McGuire & Padawer-Singer, 1976), and in recent years, distinctiveness
theory has proven to be very useful in predicting how consumers will respond to various
marketing efforts. The key point is that the distinctiveness of the person who sees an ad
influences how the ad is interpreted and evaluated. The general idea is that minority
members tend to make positive evaluations of ads that make their social identity salient in
some way (e.g., using culturally-relevant cues in ads). The assumption is that when an ad
targets a minority consumer by highlighting the quality that makes him or her distinctive
(e.g., including African American models in ads that target African Americans), it
resonates with the minority consumer, leading to favorable target market effects (Aaker
et al, 2000; Deshpande & Stayman, 1994; Forehand, Reed (II), & Deshpande, 2002; Grier
& Brumbaugh, 2004). Similar effects, using a distinctiveness theory perspective, have
been found outside of the United States (e.g., Martin, Kwai-Choi Lee & Yang, 2004).
Recognition of in-group cultural cues. Every cultural group has symbols that are
meaningful to its group members. These cultural symbols are often unnoticed or

11
unrecognized by those who aren’t members of the in-group. For example, Heinz ketchup
containers have had a kosher “Circle U” symbol on the label for over eighty years
(heinz.com), but the average non-Jewish consumer is probably not aware of the symbol
and/or does not recognize its meaning. In their research, Grier and Brumbaugh (1999)
discovered that participants’ ability to “decode” the cultural cues in ads was based on in-
group membership. Majority members who were exposed to minority-targeted ads
containing symbols associated with the minority group found it difficult to interpret the
ads and decode the cues. In contrast, minority members were able to “get” the ads and
recognize the cues, which led to a positive evaluation of these minority-targeted ads.
With regard to the LGBT community in particular, Oakenfull (2004) suggested that
companies wanting to be positively evaluated by LGBT members and to avoid negative
evaluations by non-LGBT individuals should explore the effectiveness of “gay-vague”
ads. In 2005, Oakenfull and Greenlee found that LGBT members were more likely to
respond positively to these kinds of “gay-vague” ads compared to non-targeted ads.
Although Oakenfull and Greenlee (2005) explored this idea in an advertising context, it
was not clear if the same would hold true in health-oriented messages.
Nontarget market effects of distinctiveness. The previously mentioned marketing
studies examined the relatively favorable effects of being a minority member on his or
her response to targeted advertisements. Aaker, Brumbaugh, and Grier (2000) went
further, focusing on what they term nontarget market effects, which are “marked not by a
failure to achieve favorable target market effects, but rather a decreased preference for an
advertisement by people who believe they are not the target of the advertisement” (p.

12
128). Their findings indicate that in addition to favorable target market effects for
minority members, there are non-target market effects such that majority members really
dislike ads that do not target them. On the other hand, Oakenfull and Greenlee (2005), in
an advertising context, also examined the effects of gay-vague ads on non-LGBT
members’ responses: They found no significant difference in non-LGBT members’
responses to gay-vague ads versus non-targeted ads. This study, although not in a health
context, suggests that it might be possible to reach a gay and a non-gay audience with
“gay vague” messages.
Based on distinctiveness theory’s predictions and robustness, distinctiveness
theory has broad implications for marketing research in general, and I would argue, for
health marketing efforts in particular. That is, distinctiveness theory suggests that to
change risky behavior for people who are distinctive (e.g., LGBT individuals) health
messages need to include distinctive (e.g., gay) cues. At the same time, however, for
non-distinctive groups (e.g., non-LGBT individuals), those cues can’t be so obvious that
individuals in non-distinctive groups (e.g., non-LGBT individuals) believe that these
messages are targeting someone else. Before providing an overview of the current studies
proposed, we first consider research relevant to the effects that brief messages and
marketing ads may be expected to have on attitudes and behavior.
Messages: What Might They Change?
For many public health marketers, success is defined as changing the behavior of
people who are engaging in unhealthy behaviors. But public health media messages do
not function as magic bullets; there are few, if any, direct and immediate effects such

13
that, for example, those exposed to an anti-smoking message will quit smoking at once.
Rather, there is a “hierarchy of effects” model suggested by researchers such as McGuire
(1989, 1984) and others (e.g., see Singhal & Rogers, 1999). In brief, according to
McGuire, the basic stages in this model are as follows:
Exposure to a message
Paying attention to it
Liking and becoming interested in the message
Comprehending it
Attitude change
Behavioral change as advocated in the message
Maintaining behavioral change
The assumption is that getting to each step is dependent on having reached the
steps leading up to it. The model describes how people move from initial exposure to a
message to behavioral change and maintenance (McGuire, 1984 & 1989; Singhal &
Rogers, 1999). An important consideration for this media-effects model is that as one
moves from one step to the next, the probability of success for that next step is greatly
diminished. McGuire (1989) warns marketers not to fall prey to “the attenuated-effects
fallacy” (p. 49), which occurs when there is an unrealistic expectation for success.
McGuire suggests that if the probability of success at Step 1, exposure, is .50 (which he
says is a very high expectation), then the probability of success at the last step (n) is .50
to the nth power. In the abbreviated model above, the probability of Step 7 (behavioral
maintenance is .50 to the 7
th
power, or .008.

14
Another way to think of the low probability of reaching the last step of the model
on a macro-level is to apply the “rule of halves,” which is based on the idea that “for any
population, by any measure,” half of that population will end up on each side of the
median, or midpoint (Goff, 2001). The rule of halves illustrates why it is difficult to
address the health issues of a population; simply put, the numbers are against them. The
rule has often been used to describe what happens in populations of those with chronic
diseases such as hypertension: Only half the people with high blood pressure know they
have it, of the half that know it, only half get treatment for hypertension, and of those
who get treatment, only half are able to control their hypertension (Duggan, et al, 2001;
Hooker, Cowap, Newson & Freeman, 1999; Smith, Lee, Crombie, & Tunstall-Pedoe,
1990). In terms of public health campaigns, the Australian National Public Health
Partnership used the rule of halves to demonstrate how hard it is for health marketers to
achieve what most would consider success, which is to move a large percentage of their
target population toward behavioral change and maintenance. From this perspective,
only about half of the target audience (50%) will be exposed to a health message, and
only half of those who see it will attend to it, and so on down the line. Based on the
above, researchers should be clear about their change goals and design their studies
accordingly. For example, researchers interested in which messages participants might
like or view as intended for or relevant to them, might systematically differentially
expose participants to experimental and control messages – bypassing earlier steps in the
model and assess these more restricted outcomes. Similarly, it seems clear that it is
unrealistic to expect people to change their unhealthy behaviors (and maintain that

15
change) right after seeing a health message, as there are many steps between exposure
and long-term behavior change.
However, given the model above, participants could make initial judgments of
message effectiveness and initial assessments of attitude change following message
exposure. Furthermore, while it would not be typically possible to initially access
participants’ health behavior change, per se, immediately after message exposure, it
would be possible to access participants’ willingness to engage in behavior that might
move them to change that behavior over time. One such behavior, as suggested in the
hierarchical model of Singhal and Rogers (1999), involves willingness to share the
message with others after seeing that message. Encouraging the interpersonal sharing of
message with others – and getting people talking about the behavior in question -- is a
core idea or strategy in public health campaigns (Solomon, 1989). This is examined in
more detail below, especially as it related to anti-tobacco messages for a LGBT targeted
audience.
Social Support and Health Messages
While many anti-smoking studies focus on how to get smokers to quit, as
previously indicated, the road between the initial exposure to a health message and
behavior change/maintenance is a long and difficult one. Interpersonal communication,
however, can play a significant role in adding value to any mass media message.
Consider the two-step flow hypothesis, a staple of communication research for over fifty
years. The basic idea here is that that mass media messages “flow” through to “opinion
leaders,” and through them, to others in their circle (Katz, 1957, p. 61). An important

16
finding in this research area is that interpersonal influence is more persuasive than mass
media (see Katz, 1957, for a review). Katz also points out three things that make people
influential: who you are, what you know, and who you know. Also, people tend to listen
more to those who share their values and who they want to be like. Furthermore, social
influencers and the people they influence are similar in that they are often members of the
same family, personal friends, or come to know each other in the workplace. According
to Katz, “interpersonal relations are (1) channels of information, (2) sources of social
pressure, and (3) sources of social support” (p. 77). Similarly, according to Atkin (2001),
health marketers should try to reach the primary target audience’s “personal influencers”
(p. 53), that is, people in their inner circle (e.g., friends, family) in addition to the primary
audience members. Atkin suggests that personal influencers can often be more
persuasive than messages directed toward the primary audience itself. This is an
important point, especially for those trying to reach relatively small populations of
interest such as LGBT smokers. From this perspective, health marketers would do well
to try to open up the lines of communication between smokers and those around them in
the LGBT community.
Organizations that promote smoking cessation also stress the importance of social
support. According to the American Legacy Organization, social support “includes
informal support such as from family, friends, or coworkers, and formal support such as
from a support group, or the support of a health care provider.” The Clinical Practice
Guideline on Treating Tobacco Use and Dependence recommends that social support be
a part of smoking cessation programs (U.S. Department of Health and Human Services,

17
2000). The Surgeon General’s 2001 Report states that social support is especially
important to women and girls who smoke. Similarly, even for adolescents, a group that is
particularly vulnerable to become smokers due to peer pressure, Fagan et al (2003) found
at least a moderate relationship between having peer support and anti-smoking behavior.
In the current work, the focus is on participants’ perceptions and attitudes regarding
messages and their willingness to share them after exposure to them.

18
CHAPTER TWO

Overview of Studies and Specific Predictions

The current research focused on the effects of anti-smoking messages. The
messages varied in how explicitly they targeted LGBT individuals (explicit, less explicit,
and no targeting control messages). Study 1 examined the responses of LGBT members
(members of a distinctive group) to these messages, while Study 2 examined the
responses of non-LGBT members (members of a non-distinctive group) to the same
messages. Based on distinctiveness theory, the following hypotheses and questions were
derived.
H
1
: LGBT members would respond more favorably to messages that explicitly target
their group compared to messages that don’t target their group.
H
2
: LGBT members should respond better to “gay vague” messages than messages that
do not target LGBT individuals.
RQ
1
: Would LGBT members respond as positively to “gay vague” as explicit LGBT-
targeted messages?
H
3
: Non-LGBT individuals would respond less favorably to explicitly targeted LGBT
messages than to non-LGBT messages.
H
4
: Non-LGBT members should respond better to “gay vague” messages than messages
that explicitly target LGBT individuals.

19
RQ
2
: Would Non-LGBT respond as positively to “gay vague” messages as they do to
non- LGBT targeted messages?
H
1
, H
2
, and RQ
1
were examined in Study 1. H
3
, H
4
, and RQ
2
were examined in Study 2.

Study 1
In the first study, the population of interest was the LGBT community. I
predicted that these individuals would be more receptive to both gay-explicit and gay-
vague messages compared to messages that did not target them at all. Three primary
comparisons were made in terms of message effects: (a) gay-explicit versus control (H
1
),
(b) gay-vague versus control (H
2
), and (c) gay-explicit versus gay-vague (RQ
1
).
Method
Participants. In Study 1, a total of 61 LGBT individuals responded to the
recruitment announcement. Of these, all met the eligibility requirement, which was that
participants had to be at least 18 years old. Of those who responded, 17 were under 20
years old, 34 were between 20 and 30 years of age, 4 were between 31 and 40, and 6
were 41 and over. With regard to gender, there were 33 males, 24 females, and 4
transgendered individuals, and 17 LGBT participants were smokers versus 44
nonsmokers.
Procedure. The present research was conducted online. The study utilized a 1 x 3
within-subjects design to examine the responses of LGBT individuals to anti-smoking
messages that varied in terms of how explicitly they targeted members of their own

20
community. Specifically, there were three types of messages: gay-explicit, gay-vague,
and control messages, and all the LGBT participants saw all three types of messages.
The decision to use a within-subjects factor was based on several considerations.
Exposing participants to a series of messages rather than just one message is a strategy
that has been used by previous researchers interested in people’s responses to targeted
advertisements (Aaker, Brumbaugh, & Grier, 2000; Brumbaugh, 2002; Grier, &
Brumbaugh, 1999; Oakenfull & Greenlee, 2005). Moreover, it meant that fewer
participants would be needed, an important consideration because the population of
interest for Study 1 was the LGBT community, whose members can be difficult to locate
and recruit. Also, as all the participants saw the entire set of messages, they served as
their own control subjects, which addressed the issue of possible within-group
differences.
When participants are exposed to a series of messages, there is the potential
danger of carryover effects when the messages are always presented in the same order.
That is, the effect of a particular message might be partly due to what came before it,
making it impossible to isolate the effect of the message itself. Thus, the messages in this
study were counterbalanced, and a balanced Latin square design was deemed appropriate,
in which the order of messages presented to any given participant was random: That is,
across participants, an algorithm made it possible that each message appeared before and
after every other message (Wagenaar, 1969).
5
The number of participants who were

5
The actual number of different messages (“n”) shown to each participant was 8. Each
row represents a particular presentation order (“P”) of the 8 messages. In our example

21
assigned to the different presentation orders was monitored throughout the experiment to
ensure that the assignment was random. A Chi-square analysis indicated that for the 61

below, (M) stands for “message” and the number it’s paired with is the number of a
particular message (e.g., M1 is Message #1). Please see below for a detailed description
of each message. The formula for determining the presentation orders is as follows: Row
P1 shows the starting presentation order. After completing the first row, for the next
rows, just add 1 to the number just above it in the preceding row. But if adding 1 results
in a number greater than n (n=8), then add 1 to that number and subtract n. Example: if
the number in the cell just above the one you’re working on is 8: 8+1=9, and 9-8 =1.
1
st
2
nd
3
rd
4
th
5
th
6
th
7
th
8th
(P1) M1, M2, Mn, M3, Mn-1, M4, Mn-2 *
(P2) Add +1 to no. in preceding row for that column. If sum = >n, subtract n)

*In the first row (P1), the eighth message in the presentation order is determined by a
process of elimination after filling in the other seven spots in the row based on the
formula.

(http://www.city.ac.uk/psychology/dps/teaching/Kaminska/docconv_3277/Summary%20
of%20Lec%201/3350_4601.htm)

(Presentation
Order)

P1 M 1 M 2 M 8 M 3 M 7 M 4 M 6 M 5
P2 M 2 M 3 M 1 M 4 M 8 M 5 M 7 M 6
P3 M 3 M 4 M 2 M 5 M 1 M 6 M 8 M 7
P4 M 4 M 5 M 3 M 6 M 2 M 7 M 1 M 8
P5 M 5 M 6 M 4 M 7 M 3 M 8 M 2 M 1
P6 M 6 M 7 M 5 M 8 M 4 M 1 M 3 M 2
P7 M 7 M 8 M 6 M 1 M 5 M 2 M 4 M 3
P8 M 8 M 1 M 7 M 2 M 6 M 3 M 5 M 4

Description of each message used in the study
M 1: Black, gray, and white flag, no gay text (control)
M 2: Pink triangle, no gay text (gay-vague)
M 3: Blue triangle, no gay text (control)
M 4: Rainbow flag, no gay text (gay-vague)
M 5: Gay text only, no symbol (gay-explicit)
M 6: No gay text, no symbol (control)
M 7: Pink triangle with gay text (gay-explicit)
M 8: Rainbow flag with gay text (gay-explicit)

22
LGBT individuals, the frequency of individuals across cells did not significantly differ
from the expected value (Χ
2
(7)=7.85, p >.10). For each analysis, order (and the
interaction with condition) were included as factors.
As the present study was posted online, participants were recruited in two ways:
(a) A recruitment announcement was made in person in several communication classes at
a California university. Those interested in participating were given the web link for the
study and (b) The recruitment announcement was also emailed to various college-based
organizations (including LGBT organizations on different college campuses nationally).
The announcement stated the approximate time it would take to participate in the study
(10-15 minutes), and eligibility information (i.e., participants had to be at least 18 years
old). The announcement also stated that a randomly selected individual would receive a
$200 cash prize.
The actual recruitment script is as follows:
Hi! I’m a college student conducting a research study on what people think of
various health messages. Your participation is voluntary. You will be asked to
complete a questionnaire which should take about 10-15 minutes total. The only
requirement to take part in my study is that you have to be at least 18 years old.
Your responses will be completely confidential and won’t be shared with anyone
who isn’t affiliated with my study (for example, other researchers or people who
want to sell you stuff). One randomly selected person will receive a $200 cash
prize. Up to 200 participants will take part in the study, so the odds of winning the
cash prize will be approximately one in 200. You do not have to take part in the
survey in order to be eligible for the lottery. If there are any questions, you can
email me at: research.researche@sbcglobal.net. To access the study and more
information about the study, just click on the link below. Thanks!
http://www-scf.usc.edu/~lalexand/survey/

After interested participants went to the study’s website, they first saw the
Information Sheet page, which included the following information: the purpose of the

23
study; what they would be expected to do as participants; the expected time for
participation; instructions on what they could do to enter a lottery for a $200 cash prize;
and the names and contact information of the researchers. Participants were then
informed that “completion and return of the questionnaire will constitute consent to
participate in this research project.”
At the bottom of the Information Sheet, there was a “Participation” button for
interested individuals to continue on to the study. Once they clicked on this button,
participants were randomly assigned to one of eight conditions that varied the
presentation order of the anti-smoking messages.
Each participant viewed a series of eight messages (three gay-explicit, two gay-
vague, and three control messages). After each message, they were asked to respond to
the same set of questions or statements with regard to that particular message, i.e., “After
seeing this message, with whom would you share concerns about smoking?” “Overall, I
think the message is effective.” “I feel this message is intended for people like me.”
These questions, and the response scales associated with them, will be discussed in
greater detail later. After seeing and providing responses to the final message,
participants were asked for some demographic information: age, gender, sexual
orientation, and whether or not they were smokers. In addition, participants were shown
LGBT and non-LGBT symbols and asked to indicate whether these symbols had any
meaning for them. Participants were asked to type in their email address so that their
responses could be processed. Finally, they were given the option of entering the lottery
for the cash prize.

24
Message stimuli. There were a total of eight anti-smoking messages. The
messages varied in how explicitly they targeted the LGBT community. There were three
categories of messages: gay-explicit (three messages), gay-vague (two messages), and
those that did not target LGBT individuals at all (control condition, three messages total).
The messages differed in terms of textual and graphic content. All the messages were
presented to participants, as was the case in earlier studies (Aaker, Brumbaugh, & Grier,
2000; Brumbaugh, 2002; Grier & Brumbaugh, 1999; Oakenfull & Greenlee, 2005).
All the messages contained information about the harmful effects of smoking
(CDC, “Tobacco-Related Mortality”). The exact wording is as follows:
DID YOU KNOW THAT:
• According to the Centers for Disease Control (CDC), on average,
nonsmokers live 14 years longer than smokers.
• Compared to smokers, nonsmokers are more likely to be healthy because
they are not at risk from smoking-related diseases such as lung cancer,
heart disease, and chronic lung diseases such as emphysema.
• Each year, cigarette smoking is responsible for an estimated 440,000
deaths. This number includes approximately 35,000 deaths due to
secondhand smoke. So smoking really affects both smokers and
nonsmokers. Smoking affects us all.

For the three gay-explicit messages, the following statement was added to the
basic smoking information above: “Gays and lesbians smoke more than the general
population.” This finding has been reported by a number of researchers (California
Department of Health Services, 2005; Gruskin, et al., 2007; Ryan et al., 2001; Stall et al.,
1999). Apart from the added text, two of the gay-explicit messages also included LGBT

25
cultural symbols: one featured an upside-down pink triangle, another included the
rainbow flag, and a third did not use any symbols at all.
6

The single gay-vague ad used by Oakenfull and Greenlee (2005) for their study
was a beer ad with the tag line “Another one coming out.” The ad also contained two
LGBT cultural symbols, a rainbow flag and the less well-known upside-down pink
triangle. In the present study, two messages were created for the gay-vague condition.
As these were gay-vague messages, the smoking information specific to gays and
lesbians was excluded. Unlike the Oakenfull and Greenlee study in which several cultural
cues were present in a single ad, the present study attempted to explore the effect of each
symbol such that one gay-vague message featured an upside-down pink triangle and the
other featured a rainbow flag.
The LGBT symbols used in the current study were the upside-down pink triangle
and the rainbow flag. But might it be the case that it is actually the shape of the symbol
that has some sort of effect? For example, a triangle, regardless of color, can function as a
“pointer” to indicate important information. As a result, for the control condition, the
upside-down pink triangle was converted to a blue one, and the rainbow flag was
converted to a black, gray, and white color pattern. The “control” symbols were exactly

6
The decision to use the rainbow flag and pink triangle as LGBT cultural symbols
instead of others (e.g., the lambda symbol, λ) was based on their usage in previous
research (Aaker, Brumbaugh, & Grier, 2000; Oakenfull & Greenlee, 2005) and their
mention in an important work on marketing and the LGBT community (Penazola, 1996).
A 2006 survey conducted by the Advocate, a large-circulation LGBT magazine, also
revealed that these are significant cultural symbols for this community. Readers were
asked: “Which of the following symbols would you identify as representing the LGBT
experience?” and given a choice of six symbols. The top two vote-getters were the
rainbow flag (88%) and the pink triangle (63%) (Vary, 2006).

26
the same size as the originals. The third control message only presented the CDC
smoking information (no symbols; no gay/lesbian smoking information).
Dependent measures. Because participants had to respond to many messages, the
number of dependent measures per message had to be restricted given time constraints
for completing the study. There were three dependent variables in the current work. The
first dependent variable involved the extent to which participants felt the messages were
intended for them. Specifically, for each message they received, participants were asked
to indicate the extent to which they endorsed one item on a 5-point scale from strongly
disagree (1) to strongly agree (5) with a midpoint of neither agree or disagree. That item
read, “I feel this message is intended for people like me.” This item has been used in
marketing studies (e.g., Aaker & Brumbaugh, 2000). The second dependent variable
measured was participants’ perception of message effectiveness. The same Likert-type
scale was used, and the specific item was: “Overall, I think the message is effective.”
The third dependent variable was the likelihood of sharing anti-smoking concerns with
others after seeing each message.
To measure the likelihood that a health message might open the door to
interpersonal communication, research participants were asked the extent to which a
message was likely to stimulate conversations with others about smoking. Specifically,
participants were asked: “After seeing this message, with whom would you share
concerns about smoking?” Participants could select from a number of options (i.e., At
least one family member who smokes; At least one friend who smokes; Strangers who
you see smoking; and I would not share concerns about smoking with anyone).

27
Demographic and individual differences measures. Once participants completed
their responses to the entire set of anti-smoking messages, they were asked to provide
additional information about themselves. The population of interest for Study 1 was
LGBT individuals. Thus it was important to provide mutually exhaustive categories in
terms of sexual orientation (heterosexual/straight, bisexual, homosexual/gay/lesbian,
transgender, Other—asked to specify) and gender identification: male, female, or
transgender.
Participants were also asked to report their age, and whether or not they smoked.
Those who reported that they were smokers were asked if they smoked more than one
cigarette per week, and those who smoked more than one per week were classified as
smokers.
7

As one of the dependent measures was the likelihood of participants’ talking
about smoking with other people, the next question asked: “Among people that you are
close to, such as family members and/or friends, is at least one person among them a
smoker?”

7
Measures used to differentiate between smokers and nonsmokers do not define smokers
in the same way. In some studies, the definition of a smoker is someone who has smoked
within the last 30 days, in others, a smoker is someone who has smoked at least 100
cigarettes in one’s lifetime. There are also other definitions of smokers vs. nonsmokers
(see Smoking: Risk, Perception, and Policy, 2001, for a review). For the current study,
the focus was on smoking behavior from a weekly perspective, a period of time that is
easily recalled. In addition, although Wills and Cleary (1997) did not make smoking
behavior during one week’s time the sole criterion for categorizing participants as
smokers, they did ask their participants to respond to the following statement: “I usually
smoke a few cigarettes a week.” Similarly, in their initial screening for potential research
participants, Freeman et al. (2001) considered those who smoked less than one cigarette a
week to be nonsmokers.

28
Symbol recognition. This was the final item that was directly relevant to the
current study (Several symbols were used as filler items; they are not discussed here).
“What do the following symbols mean to you? What comes to mind when you see them?
Please write in your response next to each. If something has no meaning for you, that's
fine, just say so.”

Figure 2: Symbol Recognition

This was to see whether the rainbow flag and pink triangle were perceived as LGBT
symbols compared to their differently colored counterparts, the black and white flag and
blue triangle, both used as controls in the experiment.
Results and Discussion
The present research examined the effects of gay-explicit, gay-vague, and control
messages using paired-samples t-tests. There were three dependent variables in the
current work. LGBT participants indicated the extent to which they felt the messages
were intended for them, they evaluated message effectiveness, and they reported the
likelihood of sharing anti-smoking concerns with others after seeing each message.
Perception that the message was intended for the individual. A between-subjects
one-way ANOVA revealed significant order effects for two gay-vague messages, the
rainbow flag with no LGBT smoking statistics, F (7, 52) = 2.72, p = .018, and the pink
triangle, also with no LGBT smoking statistics, F (7, 53) = 2.72, p = 017, in terms of
participants’ perception that a message was intended for them . Subsequent post hoc

29
comparisons indicated that for the rainbow flag without the LGBT statistics, the mean
score when it appeared in Presentation Order 1 (M = 4.17, SD = .72) was significantly
different from the mean score in Presentation Order 4 (M = 2.40, SD = 1.52). For the
pink triangle without the LGBT statistics, its mean score in Presentation Order 3 (M =
2.50, SD = 1.20) was significantly different from its mean score in Presentation Order 6
(M = 3.91, SD = .54). All other comparisons did not yield significant differences across
presentation order. Therefore, we ran those orders that were not problematic as well as
all orders. Because the pattern was the same, we report the analyses with all presentation
orders.
8

H
1
predicted that LGBT members would be more likely to perceive that the
message was intended for them when presented with a gay-explicit message compared to
a control message in which there was no gay content. There were three gay-explicit
messages. The text-only version contained LGBT smoking statistics. The other two gay-

8
The results reported for H
2
were based on an analysis that considered all the
presentation orders. However, the analysis was also conducted using only the ones that
were not problematic. This excluded responses to the gay-vague pink triangle and
rainbow flag messages when they appeared in Presentation Orders 1, 3, 4 and 6. The
results were as follows: The gay-vague message with the pink triangle (M = 3.27, SD =
1.05) was significantly more likely to be perceived by LGBT individuals as intended for
them compared to a message with a blue triangle (control condition, M = 2.80, SD =
1.01), t(40) = 2.960, p = .005. The mean score for the gay-vague pink triangle (M = 3.26,
SD = 1.04) was significantly higher than the mean score for the control message with no
gay symbol or LGBT statistics (, M = 2.83, SD = .99), t(41) = 2.735, p = .009). In
addition, the gay-vague rainbow flag message was significantly more likely to be seen as
intended for them by the LGBT participants than the control message with the black and
white flag (M = 3.65, SD = 1.09 vs. M = 2.93, SD = 1.22), t(42) = 3.499, p = .001.
Furthermore, the mean score for the rainbow flag message (M = 3.65, SD = 1.09) was
significantly higher than the mean score for the message with no gay symbol or LGBT
statistics (control condition), (M = 2.65, SD = 1.17), t(42) = 4.637, p = .001.

30
explicit messages contained the same LGBT statistics but also included one of two LGBT
symbols, a rainbow flag or an upside-down pink triangle. There were three control
messages. Two featured symbols to offset the LGBT symbols; one used a black and
white flag and the other used a blue triangle. The third control message did not have any
symbols at all. None of the control messages contained LGBT smoking statistics.
H
1
was tested using a paired-samples t-test to compare a text-only gay-explicit
message to a text-only control message. The hypothesis was supported. As indicated in
Table 2, the message containing LGBT smoking statistics (gay-explicit condition) was
significantly more likely to be perceived by LGBT members as intended for them (M =
3.53, SD = 1.13) versus a control message that did not include LGBT statistics (M = 2.85,
SD = 1.10), t(59) = 4.759, p = .001. Similarly, the two other gay explicit messages with
symbols, the pink triangle message (M=3.64, SD = 1.11) and the rainbow flag message
(M=3.77, SD = 1.09) differed significantly from the control message without LGBT
statistics (t(60)= 4.708 for the triangle message; t(60)= 5.483 for the rainbow message,
p = .001. (see Table 2)

31
Table 2:
Results By Sample and Comparison With Control

LGBT

Straight

Dependent
Variable
Message Gay
Message
Control Gay
Message
Control
For me
Gay Explicit
Text Alone 3.53 2.85*

2.51 3.08*
Text With
Triangle
3.64 2.85* 2.39 3.08*
Text With
Rainbow
3.77

2.85* 2.34 3.09*
Gay Vague
Triangle 3.28 2.85* 2.95 3.08+
Rainbow 3.65 2.85* 2.54 3.08*

Effective
Gay Explicit
Text Alone 3.25 3.07+ 3.33 3.59*
Text With
Triangle
3.23 3.03 3.34 3.60*
Text With
Rainbow
3.33 3.03+ 3.35 3.60*
Gay Vague
Triangle 3.05 3.03 3.49 3.59+

Rainbow 3.22 3.07 3.35 3.59*

32
Table 2, Continued

Share
Gay Explicit
Text Alone .951 .852 .980 1.08*
Text With
Triangle
.879 .853 .919 1.08*
Text With
Rainbow
.950 .877 .909 1.08*
Gay Vague
Triangle .85 .80 1.13 1.08
Rainbow .85 .85 1.04 1.08

Note. Statistical analyses are performed for comparisons within a row and a sample.
*p <.05.
+ p <.10. Where there was more than one control (i.e., both a comparable symbol and a
no message control), the latter is reported in the table above)

H
2
predicted that LGBT members should respond better to a gay-vague message
than a message that does not target LGBT individuals. Here, the focus was on the use (or
not) of symbols rather than text. There were two gay-vague messages in this study; each
featured one of the two LGBT symbols mentioned earlier, an upside-down pink triangle
or a rainbow flag. The gay-vague messages did not include LGBT smoking statistics. As
indicated earlier, two control messages used in this set of comparisons featured either an
upside-down blue triangle (to control for the pink triangle) or a black and white flag (to

33
control for the rainbow flag), and the third control message did not include any gay
symbols or LGBT statistics.
As predicted, and as indicated in Table 2, paired-sample t-tests revealed
significant differences such that the gay-vague messages were more likely to be seen by
LGBT participants as intended for them than the control messages. In particular, the
message with the pink triangle (gay-vague condition, M = 3.28, SD = 1.06) was
significantly more likely to be perceived by LGBT individuals as intended for them
compared to a message with a blue triangle (control condition, M = 2.90, SD = 1.10),
t(59) = 3.098, p = .003. Similarly, the mean score for the pink triangle (M = 3.28, SD =
1.06) was significantly higher than the mean score for the message with no gay symbol or
LGBT statistics (control condition, M = 2.85, SD = 1.09), t(60) = 3.020, p = .004). Thus
it seems clear that LGBT members recognized the upside down pink triangle as an LGBT
symbol and viewed it as intended for them.
The other gay-vague message contained a rainbow flag, and it was compared to
two control messages, one with a black and white flag, and one with no gay symbol or
LGBT statistics. As in the case of the pink triangle message, the rainbow flag message
(gay-vague condition) was significantly more likely to be seen as intended for them by
the LGBT participants than the control message with the black and white flag (M = 3.65,
SD = 1.13 vs. M = 2.92, SD = 1.24), t(59) = 4.369, p = .001. And finally, the mean score
for the rainbow flag message (M = 3.65, SD = 1.13) was significantly higher than the
mean score for the message with no gay symbol or LGBT statistics (control condition),
(M = 2.85, SD = 1.10), t(59) = 4.427, p = .001. We can infer from these results that

34
LGBT individuals recognized the rainbow flag as a symbol of their community and were
more likely to view this message compared to the control as intended for them.
RQ
1
asked: Would LGBT members perceive that gay explicit and gay vague
messages were equally intended for them? A paired-samples t test revealed, as indicated
in Table 2.1, that a gay-explicit message containing a pink triangle in addition to LGBT
smoking statistics (M = 3.64, SD = 1.11) yielded a significantly higher mean score
compared to a gay-vague message that featured a pink triangle with no LGBT statistics
(M = 3.28, SD = 1.05), t(60) = 2.181, p = .033. While a message that combined a
rainbow flag with LGBT statistics (gay-explicit) also resulted in a higher mean score (M
= 3.78, SD = 1.09) than the gay-vague message, (a rainbow flag without the gay statistics,
M = 3.65, SD = 1.13), this difference was not significant in terms of LGBT participants’
perception that a message was intended for them, (t(59) = 1.112, p = .271.
9

9
As was the case in H
2
, the paired comparisons in RQ
1
also took into consideration the
gay-vague pink triangle and rainbow flag messages, so another analysis was run
excluding the problematic presentation orders, and the results were as in line with the all-
inclusive analysis. For LGBT individuals, a gay-explicit pink triangle message was
significantly more likely to be perceived as intended for them (M = 3.73, SD =.97)
compared to a gay-vague pink triangle message (M = 3.11, SD = 1.13), t(43) = 3.835, p =
.001). Similarly, a gay-explicit rainbow flag message garnered a higher mean score (M =
3.86, SD = .97) than the gay-vague rainbow flag message (M = 3.65, SD = 1.09), though
this difference was not significant in terms of LGBT participants’ perception that a
message was intended for them, (t(42) = 1.501, p = .141.

35

Table 2.1:

Comparisons Between Gay Vague and Gay Explicit Messages By Sample
LGBT Straight

Dependent
Variable

Message

Gay Explicit

Gay
Vague

Gay
Explicit

Gay
Vague
For me
Triangle 3.64 3.28* 2.39 2.94*
Rainbow 3.78 3.65 2.35 2.55*
Effective
Triangle 3.25 3.07 3.35 3.50*
Rainbow 3.37 3.23+ 3.36 3.36
Share
Triangle .87 .85 .92 1.13*
Rainbow .95 .87 .91 1.04*

Note. Statistical analyses are performed for comparisons within a row and a sample.

* p <.05.

+ p <.10.

Both gay explicit messages received higher mean scores compared to their gay
vague counterparts, though the effect was stronger, and only statistically significant, for
the pink triangle compared to the rainbow flag. Based on this finding, given a choice,
LGBT individuals prefer messages that target them in a straightforward manner, at least
in terms of their perception that a message is intended for them.

36
But why were the responses different with respect to the pink triangle and the
rainbow flag? After all, both are symbols of the LGBT community, and the majority of
the participants were able to identify them as such when they were presented with these
symbols. Ninety-five percent recognized the pink triangle and 90% recognized the
rainbow flag as gay symbols, and a chi-square test did not find this to be a significant
difference, X
2
(1, N = 61) = .344, p > .05. The results of the t-tests and chi-square
analyses seem to suggest that these symbols may not be perceived in quite the same way
by LGBT members. It could be the case that these days, even if the two symbols are
readily identified as LGBT symbols, the pink triangle is a lot more gay vague than the
rainbow flag. This would explain why LGBT participants were significantly more likely
to say that the pink triangle message with the LGBT statistics (gay explicit condition)
was intended for them compared to the gay-vague message that only featured the pink
triangle, and why we didn’t see this for the rainbow flag comparison.
The participants were mostly young adults, as this is the population of interest in
the current research. And for young adult LGBT individuals, the upside-down pink
triangle, a symbol tied to the oppression of homosexuals in Nazi Germany (Stonewall
Society), may be less likely to be thought of as a symbol that represents their experiences
compared to the now ubiquitous rainbow flag. As a result, a gay vague message
containing a pink triangle with no accompanying LGBT text may not be seen by LGBT
members as intended for them because at present, the pink triangle is not a powerful
enough LGBT symbol. On the other hand, it could be argued that the rainbow flag is
very well known, and merely incorporating a rainbow flag into a message may be explicit

37
enough even without the LGBT text. This might explain why the mean scores of a gay
vague message with only a rainbow flag and no LGBT text versus the gay explicit
rainbow flag message that included LGBT text were not significantly different.
Message effectiveness. A between-subjects one-way ANOVA was used to test
for possible order effects for message effectiveness. Results indicated that there were no
significant differences across the presentation orders.
H
1
predicted that LGBT members would view a gay-explicit message as more
effective compared to a control message in which there was no gay content. H
1
was
tested using a paired-samples t-test to compare a text-only gay-explicit message
(containing LGBT smoking statistics) to a text-only control message (no LGBT
statistics). As indicated in Table 2, although the gay explicit message yielded a higher
mean score for message effectiveness (M = 3.25, SD = .98) than the control message (M
= 3.07, SD = 1.07), the difference was marginally significant t(58) = 1.900, p = .062. For
the two other explicit messages that added a symbol, the difference between the rainbow
flag message and the control message was marginally significant (p = .063), while the
difference between the triangle message and the control message was not significant (see
Table 2).
H
2
predicted that LGBT members should view a gay-vague message as more
effective than a (control) message that does not target LGBT individuals. As was the
case for H
1
, paired-samples t tests revealed only one marginally significant difference
between them in terms of message effectiveness. The mean score for the pink triangle
message (gay vague condition) did not differ from than the mean score for the (control)

38
blue triangle message (M = 3.07, SD = 1.06 vs. M = 3.08, SD = 1.03), t(59) = .148, p =
.883). The pink triangle message was not more effective than the control message with
no gay symbol or LGBT statistics (M = 3.05, SD = 1.05 vs. M = 3.03, SD = 1.09, t(59) =
.123, p = .903). The rainbow flag message (gay vague condition), however, garnered a
higher mean score than the (control) message featuring a black and white flag, although
this was marginally significant (M = 3.23, SD = 1.05 vs. M = 2.93, SD = 1.15, t(59) =
1.941, p = .057). But, the rainbow flag message did not differ from the control message
with no gay symbol or LGBT statistics (M = 3.22, SD = 1.05 vs. M = 3.07, SD = 1.07,
t(58) = 1.026, p = .309.
RQ
1
asked: Would LGBT members view a gay explicit as equally effective as a
gay vague message? The results of a paired-samples t test revealed that the gay explicit
pink triangle message that included LGBT statistics was not viewed as more effective
than the gay vague pink triangle message without the LGBT statistics (M = 3.25, SD =
1.12 vs. M = 3.07, SD = 1.05), t(60) = 1.591, p = .117 (see also Table 2.1). The gay
explicit rainbow flag message with the LGBT statistics was, however, marginally more
effective than the gay vague rainbow flag message without the LGBT statistics (M =
3.37, SD = 1.01 vs. M = 3.23, SD = 1.05, t(59) = 1.734, p = .088).
Likelihood of sharing anti-smoking concerns with smokers. This measure was
originally a set of three nominal variables (i.e., sharing concerns with at least one family
member who smokes; at least one friend who smokes; and strangers who were seen
smoking). A new variable was created such that it represented the sum of the three
variables. For example, for a participant who indicated that he or she would not share

39
concerns with a family member, friend, or stranger, the assigned score was 0, while a
participant who reported that he or she would share concerns with family, friends, and
strangers would be assigned a score of 3. Thus the dependent variable was measured on
a three-point scale, and paired-samples t-tests were performed.
In terms of the likelihood of sharing smoking concerns, a between-subjects one-
way ANOVA revealed significant order effects for one gay–explicit message, the
message featuring a pink triangle with LGBT smoking statistics, F (7, 53) = 2.48, p =
.028. A post hoc test revealed that for this particular message, Presentation Orders 1, 2,
3, 4, 6, and 7 were problematic.
H
1
predicted that LGBT members would be more likely to share concerns in
response to a gay-explicit message compared to a message with no gay content. A
paired-samples t-test revealed that although the text-only gay-explicit message
(containing LGBT smoking statistics) obtained a slightly higher mean score for
likelihood of sharing anti-smoking concerns compared to the control message (no LGBT
statistics), the difference was not statistically significant (M = .951, SD = .90 vs. M =
.852, SD = .93, t(60) = 1.516, p = .135). Similar effects were found for the other
explicit messages (triangle, rainbow flag) with text (see Table 2).
H
2
predicted that LGBT members should be more likely to share concerns in
response to a gay-vague message than a (control) message that does not target LGBT
individuals. A paired-samples t tests found no significant differences between them for
likelihood of sharing smoking concerns. The mean score for the pink triangle message
(gay vague condition) did not significantly differ from the mean score for the (control)

40
blue triangle message (M = .85, SD = .87 vs. M = .80, SD = .85), t(60) = .830, p = .410).
The mean score for the pink triangle message was the same as the control message’s (a
message with no LGBT symbols or statistics) (M = .85, SD = .87 vs. M = .85, SD = .93,
t(60) = .000, p = 1.000).
The rainbow flag message (gay vague condition) did not significantly differ from
the mean score of the (control) message featuring a black and white flag (M = .85, SD =
.87 vs. M = .79, SD = .90; t(60) = 1.000, p = .321). The rainbow flag message obtained
the same mean score compared to the control message with no gay symbol or LGBT
statistics (M = .85, SD = .87 vs. M = .85, SD = .93, t(60) = .000, p = 1.000).
RQ
1
asked: Would LGBT members be more likely to share messages in response
to gay explicit and gay vague messages? The results of a paired-samples t test revealed
no significant differences between gay explicit and gay vague messages for likelihood of
sharing anti-smoking concerns, though the gay explicit messages obtained a slightly
higher mean score compared to their gay vague counterparts. Specifically, participants
didn’t respond differentially to the pink triangle message with the LGBT smoking
statistics (gay explicit condition) and the gay vague message that used a pink triangle
without the LGBT statistics (M = .87, SD = .85 vs. M = 85, SD = .87), t(60) = .207, p =
.837. Similarly, participants did not differentially respond to the rainbow flag with the
LGBT statistics (gay explicit condition) compared to their response to the gay vague

41
counterpart, a message containing a rainbow flag but no LGBT statistics (M = .95, SD =
.87 vs. M = .87, SD = .87), t(59) = 1.043, p = .301.
10

Interim summary. At the outset, there was an attempt to address the potential
problem of presentation order effects by utilizing a balanced Latin square design in which
each message was shown before and after every other message. As it turned out, order
effects were revealed for some of the messages as a result of between-subjects one-way
ANOVAs and post hoc tests. This meant that the analyses were conducted with and
without the problematic presentation orders, though the pattern of results ended up being
the same.
The purpose of Study 1 was to examine the responses of LGBT individuals to
gay-explicit, gay-vague, and control messages. There were three dependent measures
used in the study: the perception that a message was intended for the individual, message
effectiveness, and the likelihood of sharing anti-smoking concerns with others. The

10
The analysis was first conducted with all orders and then with only the nonproblematic
orders. The results were not significant in both instances. However, in the latter case, the
pattern was different such that the gay explicit pink triangle message obtained a lower
score than its gay vague counterpart, while the opposite was true for the rainbow flag
comparisons. The pink triangle message with the LGBT smoking statistics (gay explicit
condition) yielded a slightly lower mean score than the gay vague pink triangle message
(M = .93, SD = .83 vs. M = 1.00, SD = .79), t(13) = 1.00, p = .336. In contrast, the
rainbow flag with the LGBT statistics (gay explicit condition) obtained a slightly higher
mean score compared to its gay vague counterpart, a message containing a rainbow flag
but no LGBT statistics (M = 1.00, SD = .79 vs. M = .79, SD = .80), t(13) = 1.385, p =
.189. The initial analysis was reported because the results were not significant in either
case.

42
findings indicate that LGBT members respond more positively to both gay-explicit and
gay-vague messages compared to messages that do not target their community, and gay-
explicit messages seem to work better than gay-vague messages, at least in terms of the
perception that a message is intended for the individual exposed to the message. In
contrast, the results for the other two dependent variables seem to suggest that message
effectiveness may be too ambiguous a term, and it could be the case that even if an anti-
smoking message clearly targets LGBT members, it may not be enough to persuade them
to share smoking concerns with other people.
The LGBT symbols used in the present research, the upside down pink triangle
and the rainbow flag, are clearly identifiable as representing gay culture as shown by the
LGBT participants’ responses. Yet the two symbols did not always elicit the same
response, which was unexpected. As was suggested earlier, it might be that the symbols
differ with regard to how much they resonate with young adult members of the LGBT
community.
Study 2
Study 1 examined the effects on LGBT members of messages that varied in their
explicit targeting of the LGBT community. Study 2 explored the effects of the same
messages on non-LGBT individuals. Thus the only difference was the sexual orientation
of the participants.
The following hypotheses were examined in Study 2:
H
3
: Non-LGBT individuals would respond less favorably to explicitly targeted LGBT
messages than to non-LGBT messages.

43
H
4
: Non-LGBT members should respond better to “gay vague” messages than messages
that explicitly target LGBT individuals.
RQ
2
: Would non-LGBT members respond to “gay vague” messages as

they do to non-
LGBT targeted messages?
Method
As stated above, the method was the same as in Study 1, except for this sample
(of non-LGBT individuals), there was no attempt to recruit participants nationally. These
non-LGBT participants are described in more detail below.
Participants. A total of 209 non-LGBT individuals took part in the study, and
they all met the age requirement (at least 18 years old). Of those who responded, 53 were
under 20 years old, 145 were between 20 and 30 years of age, 5 were between 31 and 40,
and 6 were 41 and over. With regard to gender, there were 52 males and 157 females, and
21 participants were smokers vs. 187 nonsmokers.
Random assignment check. A Chi-square test was conducted to insure that the
frequency of individuals assigned to cells did not significantly differ. In fact, random
assignment was successful (χ
2
(7) =7.46, p>.10).
Results and Discussion
A between-subjects one-way ANOVA was used to test for differences among the
eight presentation orders. Results indicated that there were no order effects for
perception that a message was intended for the message viewer, message effectiveness,
and likelihood of sharing smoking concerns.

44
Perception that the message was intended for the individual. H
3
predicted that

non-LGBT individuals would respond less favorably to explicitly targeted LGBT
messages than to non-LGBT messages. A paired-samples t test was used to test this
hypothesis. The analysis revealed that as predicted, a control (i.e., non-LGBT) message
was significantly more likely to be seen by non-LGBT participants as intended for them
compared to a gay-explicit message. That is, a control message that did not include any
gay symbols or LGBT smoking statistics earned a significantly higher mean score (M =
3.08, SD = 1.10) than a gay-explicit text-only message that included LGBT smoking
statistics (M = 2.51, SD = 1.07), t(204), = 7.829, p = .001. The pattern was the same for
the two other gay explicit messages with symbols, the pink triangle message (M=2.39,
SD= 1.07; t(202)= 8.880) and the rainbow flag message (M=2.34, SD= 1.06; t(60)=
9.004, p = .001) differed significantly from the control message without LGBT statistics
(see Table 2).
H
4
predicted that non-LGBT members should respond better to gay vague
messages than gay explicit messages. Paired-sample t tests found support for the
hypothesis (see Table 2.1). Combining a pink triangle symbol with LGBT smoking
statistics (gay-explicit condition) resulted in a significantly lower mean score (M = 2.39,
SD = 1.07) than a message with only the pink triangle symbol (gay-vague condition) (M
= 2.94, SD = 1.09), t(203) = 7.937, p = .001. Similarly, a rainbow flag combined with
LGBT statistics (gay-explicit message) resulted in a significantly lower mean score (M =
2.35, SD = 1.06) than the message with only the rainbow flag (gay-vague message) (M =
2.55, SD = 1.07), t(206) = 2.989, p = .003.

45
RQ
2
asked whether non-LGBT members would respond to gay vague messages
as

they do to non- LGBT targeted messages (control messages). Paired-samples t tests
revealed several interesting findings. The mean score for a pink triangle message with no
LGBT statistics (gay vague condition) was not significantly different from the mean
score for a message with a blue triangle (a control condition), (M = 2.94, SD = 1.09 vs. M
= 2.88, SD = 1.07, t(205) = .915, p = .361). In contrast, in comparing the mean score for
the same gay-vague pink triangle message to a control message (without any gay symbols
or LGBT statistics), the control message was significantly more likely to be seen by non-
LGBT individuals as intended for them (M = 2.95, SD = 1.09 vs. M = 3.08, SD = 1.10,
t(204) = .915, p = .042).
In the case of the rainbow flag messages, the message containing a rainbow flag
with no LGBT statistics (gay-vague condition) was significantly less likely to be seen by
non-LGBT individuals as intended for them compared to the two control messages.
Specifically, the gay-vague rainbow flag message obtained a significantly lower mean
score compared to a control message with a black and white flag (M = 2.53, SD = 1.06
vs. M = 2.87, SD = 1.02, t(205) = 4.751, p = .001). Similarly, when the gay-vague
rainbow flag message was compared to a control message (no gay symbols or LGBT
statistics), the gay-vague rainbow flag message had a significantly lower mean score than
the control message (M = 2.54, SD = 1.07 vs. M = 3.08, SD = 1.10, t(204) = 7.450, p =
.001).
What could explain these results? As was proposed in Study 1, the rainbow flag
is widely recognized as a gay pride symbol among LGBT members today. It could also

46
be argued that many non-LGBT members also understand it to be a symbol of the LGBT
community. These days, rainbow flags are prominently displayed in many
neighborhoods and on college campuses, and whenever there is news coverage of LGBT
events, such as the always newsworthy Gay Pride Parade, there are many rainbow flags
to be seen. Thus it might be the case that the rainbow flag is no longer a gay-vague
symbol today even for non-LGBT individuals (“this message is clearly for someone other
than me”), while symbols such as the pink triangle are still relatively obscure in many
circles.
Message effectiveness. H
3
Predicted that non-LGBT individuals would respond
less favorably to explicitly targeted LGBT messages than to non-LGBT messages, and a
paired-samples t test found support for the hypothesis. Non-LGBT participants were
more likely to find a text-only control message (no gay symbol or LGBT smoking
statistics) significantly more effective compared to a gay-explicit one (this text-only
message included LGBT statistics but no symbol). The mean score for the control
message was significantly higher (M = 3.59, SD = .88) than the mean score for the gay
explicit message (M = 3.33, SD = .93), t(202) = 4.358, p = .001. Similarly, the two other
gay explicit messages with symbols, the pink triangle message (M=3.34, SD= .88) and
the rainbow flag message (M=3.35, SD= .89) differed significantly from the control
message without LGBT statistics (t(202)= 3.817 for the triangle message; t(204)= 3.583
for the rainbow message, p = .001.
H
4
predicted that non-LGBT members should respond better to “gay vague”
messages than messages that explicitly target LGBT individuals. Two gay symbols, the

47
pink triangle and the rainbow flag, were compared with and without the inclusion of
LGBT smoking statistics. As predicted, a paired-samples t test revealed that the message
with the pink triangle but no gay statistics (gay-vague condition) was significantly more
likely to be deemed effective versus the message with both the pink triangle and gay
statistics (gay-explicit condition), (M = 3.50, SD = .81 vs. M = 3.35, SD = .88, t(204) =
2.819, p = .005).
For the messages comparing the rainbow flag, however, there was no significant
difference in message effectiveness between the gay explicit rainbow flag message with
the LGBT statistics and the rainbow flag message without the LGBT statistics (gay-vague
condition), (M = 3.36, SD = .89 vs. M = 3.36, SD = .91, t(206) = 0, p = 1.00). As was
mentioned earlier, we might explain this finding by noting that due to the mainstream
population’s being frequently reminded that the rainbow flag is a gay symbol, it may no
longer be useful to think of it as gay vague.

RQ
2
asked whether non-LGBT members would respond to “gay vague”
messages as

they do to non- LGBT targeted messages (control condition). Paired-sample t
tests indicated that the pink triangle message with no LGBT statistics (gay vague
condition) was reported to be equally as effective as the blue triangle message (no LGBT
statistics, control condition), (M = 3.49, SD = .81 vs. M = 3.53, SD = .81, t(204) = .943, p
= .349). However, there was a marginally significant difference in message effectiveness
between the gay vague pink triangle message (no LGBT statistics) and the message with
no gay symbols and no LGBT statistics (another control condition), (M = 3.49, SD = .82
vs. M = 3.59, SD = .87, t(203) = 1.890, p = .06).

48
There was no significant difference in message effectiveness between the rainbow
flag message (no LGBT statistics, gay vague condition) and the black and white flag
message (another control condition), (M = 3.35, SD = .91 vs. M = 3.37, SD = .92, t(205)
= .310, p = .757). However, the gay vague message with the rainbow flag (no LGBT
statistics) obtained a significantly lower mean score than the control message (no gay
symbol and no gay statistics), (M = 3.34, SD = .90 vs. M = 3.59, SD = .87, t(203) = 3.922,
p = .001).
Likelihood of sharing anti-smoking concerns. H
3
predicted that

non-LGBT
individuals would respond less favorably to explicitly targeted LGBT messages than to
non-LGBT messages. A paired-samples t-test found support for the hypothesis. Non-
LGBT individuals were significantly more likely to share anti-smoking concerns with
others after viewing a message with no gay symbols or LGBT statistics (control
condition) versus a message with LGBT statistics (gay explicit condition), (M = 1.08, SD
= .83 vs. M = .980, SD = .76, t(207) = 2.102, p = .037). Similarly, the other gay explicit
messages with symbols, the pink triangle message (M=.919, SD= .77) and the rainbow
message (M=.909, SD= .79) differed significantly from the control message without
LGBT statistics (t(208)= 3.263 for the triangle message; t(208)= 3.580 for the rainbow
message, p = .001.
H
4
proposed that non-LGBT members should respond better to “gay vague”
messages than messages that explicitly target LGBT individuals, and support was found
for the hypothesis using a paired-samples t test. After seeing a message containing a pink
triangle but no LGBT statistics (gay vague condition), non-LGBT individuals were

49
significantly more likely to share anti-smoking concerns with other people compared to a
gay-explicit pink triangle message that contained LGBT statistics (M = 1.13, SD = .83 vs.
M = .92, SD = .77, t(208) = 4.918, p = .001). Similarly, a rainbow flag message that did
not include LGBT statistics (gay vague condition) obtained a significantly higher mean
score than a rainbow flag message that included LGBT statistics (M = 1.04, SD = .77 vs.
M = .91, SD = .79, t(208) = 3.492, p = .001).
RQ
2
asked whether non-LGBT members would respond to “gay vague” messages
as

they would to non- LGBT targeted messages (control condition). There were two gay
vague messages, a message that featured a pink triangle and one that included a rainbow
flag. Both gay vague messages did not contain LGBT statistics. There were three control
messages; none included LGBT statistics. The first featured a blue triangle (rather than a
pink one), the second message used a black and white flag (rather than a rainbow flag),
and the third had no LGBT symbols or statistics.
There wasn’t a significant difference in likelihood to speak to others about
smoking between the gay-vague pink triangle message and the blue triangle control
message (M = 1.14, SD = .83 vs. M = 1.15, SD = .82, t(207) = .333, p = .740), and no
difference between the same pink triangle message and the control message that had no
LGBT symbols or statistics (M = 1.13, SD = .83 vs. M = 1.08, SD = .83, t(208) = 1.398, p
= .164).
Similarly, there were no significant differences in comparing the gay vague
rainbow flag message to the other two control messages. Specifically, the mean score for
the gay vague rainbow flag message was not statistically different from that of the black

50
and white flag message (control condition) (M = 1.04, SD = .77 vs. M = 1.07, SD = .84,
t(208) = .716, p = .475). Moreover, the same result was revealed in comparing the same
gay vague rainbow flag message to the control message with no LGBT symbols or
statistics (M = 1.04, SD = .77 vs. M = 1.08, SD = .83, t(208) = .930, p = .354).
The second study’s population of interest was non-LGBT individuals. Analyses
revealed that in general, gay-vague messages elicit more positive responses than explicit
ones, and a message that doesn’t target the gay community was more positively received
than a message that explicitly targets that community. But there was a difference in how
non-LGBT individuals responded to gay-vague messages versus messages that don’t
target the gay community. The gay-vague message with the rainbow flag received more
negative responses compared to messages that did not target the gay community. This
might be an indication that even among non-LGBT individuals, the rainbow flag might
be too strongly identified with the gay community to be considered gay-vague, although
the findings seem to suggest that the pink triangle is still relatively obscure outside the
gay community.
In the present investigation, LGBT and non-LGBT individuals were considered as
separate populations of interest. Because the LGBT was a more difficult sample to
recruit, the recruitment process for these samples was not the same. LGBT were
recruited nationally and non-LGBT individuals were not. Although recruited nationally,
the LGBT sample was a third the size of the non-LGBT sample. This differentially
affected the power to detect differences between conditions in each sample.

51
In addition, there could be other reasons why the two groups might differ. For
example, were there proportionately more smokers in one of the samples than in the
other? After all, the percentage of LGBT individuals who smoke is greater than the
percentage of non -LGBT individuals who smoke (California Department of Health
Services; Gruskin, et al., 2007; Ryan, et al., 2001; Stall et al., 1999). If this was the case
here, perhaps non-LGBT individuals viewed the messages as more effective because they
were less likely to be smokers. As was the case in the previous studies discussed above,
a Chi-square test revealed that a significantly larger percentage of LGBT versus non-
LGBT participants were, in fact, smokers (27.9% vs. 10.1%, respectively), X
2
(1, N =
269) = 73.650, p < .05.
And second, the findings revealed in the current studies suggest a difference in
LGBT and non-LGBT individuals in their ability to recognize certain gay symbols.
Analyses were run to further explore this issue. With regard to symbol recognition,
participants were asked if they recognized the LGBT symbols used in the current
research after completing their responses to the messages in the study. Initially, the intent
was to determine whether or not LGBT and non-LGBT participants would know that the
rainbow flag and pink triangle were LGBT symbols. But the more critical issue,
especially in light of the differences in responses to the two symbols, was to examine the
difference between non-LGBT and LGBT participants in their recognition of the rainbow
flag versus the pink triangle as LGBT symbols. A Chi-square test revealed that a
significantly larger percentage of LGBT versus non-LGBT participants recognized the
pink triangle as an LGBT symbol (90.2% vs. 28.3%, respectively), X
2
(1, N = 266) =

52
73.650, p < .05. On the other hand, the difference between LGBT versus non-LGBT
participants in terms of recognizing the rainbow flag as an LGBT symbol was not
significant (95.1% vs. 88.8%, respectively), X
2
(1, N = 266) = 2.117, p = .218). These
results indicate that while the pink triangle may still be able to “fly under the radar” of
the non-LGBT population, this might not be the case for the rainbow flag. The
unexpected responses to the gay vague pink triangle versus the rainbow flag remind us
that 1) not all cultural symbols carry the same weight among members of a cultural
group, and 2) a cultural symbol is not only recognizable to members of that culture.

General Discussion

Distinctiveness theory is a robust theory that has been studied most recently in an
advertising context. Distinctiveness theory predicts that members of minority groups will
notice things in an ad that highlight their social identity, such as culturally significant
cues, and this leads to a positive response toward the ad (Aaker et al., 2000; Deshpande
& Stayman, 1994; Forehand, Reed (II), & Deshpande, 2002; Grier & Brumbaugh, 2004).
Interestingly, researchers have discovered that these cues do not have to be very explicit
for minority members to take note of them (Grier & Brumbaugh, 1999; Oakenfull &
Greenlee, 2005; Sender, 1999).
The goal of the present research was to take on the task of creating health
messages that a target audience will respond to positively but not result in a backlash
from members of the non-target audience. This research was based on the use of what has
been called “gay-vague” advertisements, a long-standing strategy used by marketers to

53
reach LGBT individuals. The use of subtle cues tells target audience members that they
are the focus of this type of ad, while those outside this group do not recognize these
cues. Researchers such as Grier and Brumbaugh (1999) found that compared to outgroup
members, those in the ingroup are more likely to recognize and understand these
culturally relevant cues. Moreover, Oakenfull and Greenlee (2005) discovered that LGBT
individuals, the population of interest in the current investigation, have positive responses
to gay-vague ads. These studies examined responses to commercial advertisements;
however, to date, the potential for gay-vague messages has not been explored in health
communication.
Study 1 and Study 2 tested the effects of gay-vague, gay-explicit, and control anti-
smoking messages on LGBT and non-LGBT members, and the hypotheses were largely
supported, especially for the first dependent variable. When comparing control messages
to the other two types of messages, LGBT participants reported more positive responses
to gay-explicit and gay-vague messages regarding whether the message was intended for
them, with more mixed results for perceived message effectiveness and with no
differences for sharing messages. In contrast, non-LGBT participants responded more
positively to gay-vague and control messages than gay-explicit ones generally across
dependent measures. Across samples, however, it’s important to bear in mind that
samples differed not only in LGBT status. These groups also differed in sample size
(affecting power to detect differences) as well as representation of smokers (with a
greater proportion of smokers in the LGBT group that might be more resistant to an anti-
smoking message).

54
The LGBT participants responded in a positive way to messages with gay content
even if the content happened to be gay vague. Non-LGBT participants, however, either
did not recognize gay-vague content when it was present, or if they did recognize the
content as gay, they did not respond positively.
The results reported in Table 2 suggest that while LGBT members will respond
more positively to gay-explicit messages compared to messages with no gay content in
terms of feeling that these messages are meant for them, the table also shows us that
LGBT individuals will respond more positively to gay vague messages compared to
messages that do not have any gay content. Moreover, the results summarized in Table
2.1 indicate that non-LGBT individuals clearly prefer gay-vague messages compared to
gay-explicit messages. Thus it may very well be that gay-vague messages are the best
compromise for an anti-smoking message that needs to be disseminated to a larger
audience.
But what exactly should be considered gay vague? One thing that became clear
early on is that these days, the rainbow flag might not be considered gay-vague by LGBT
or non-LGBT individuals, while a symbol such as the pink triangle is still not very likely
to be identified as a gay symbol by non-LGBT individuals. This might be considered a
limitation of the current work, yet one could argue that it highlights the importance of a
rigorous pretesting of stimulus materials in future examinations of the effects of cultural
cues. In addition, cultural cues are often not just a matter of symbols. Rather, certain
words or phrases resonate very strongly with different groups. Americans are stirred by
symbols that have meaning, such as the American flag, as well as words such as

55
“freedom.” In the LGBT community, words such as “diversity” and “coming out” have
special meaning. Thus, a future direction for researchers might be to build upon what has
been done here by examining the effects of culturally relevant words and phrases for
reaching different at-risk minority groups using health appeals.
The present research has significant implications for public health campaigns.
They are extremely difficult and expensive to initiate and run, especially if the goal is to
target a specific population. The current work suggests a cost-effective alternative to the
explicit targeting of a minority group. By incorporating subtle cultural cues that speak to
a target audience but are not easily decoded by those who are not members of this
audience, we can increase the reach and impact of health messages. This strategy could
increase the likelihood of success of a public health campaign. That is, of course, as long
as there is a clear understanding of what constitutes “vague” versus “explicit” cultural
cues for a particular minority population.

56

REFERENCES
Aaker, J. L., Brumbaugh, A.M., and Grier, S.A. (2000). Nontarget markets and viewer
distinctiveness: The impact of target marketing on advertising attitudes. Journal
of Consumer Psychology, 9, 127-140.

Alexander, L. (2005). A chronicle of tobacco’s very active recruitment of gays and
lesbians. Unpublished conference paper. International Communication
Association.

Alsop, R. (1999, June). Cracking the gay market code --- How marketers plant subtle
symbols in ads. Wall Street Journal. (Eastern edition). B1.

American Legacy Organization.
http://americanlegacy.org/PDF/Social_Support_and_Smoking_Cessation.pdf

Atkin, C.K. (2001). Theory and principles of media health campaigns. In R.E. Rice and
C.K. Atkin (Eds.), Public communication campaigns (pp. 49-68). Thousand
Oaks, CA: Sage Publications.

Bhat, S., Leigh, T.W., and Wardlow (1998). The effect of consumer prejudice on ad
processing: Heterosexual consumers’ responses to homosexual imagery in ads.
Journal of Advertising, 27, 9-28.

Brumbaugh, A.M. (2002). Source and nonsource cues in advertising and their effects on
The activation of cultural and subcultural knowledge on the route to persuasion.
Journal ofConsumer Research, 29, 258-270.

California Department of Health Services.
http://www.dhs.ca.gov/tobacco/documents/press/pressreleaseSept6_05.pdf.

CDC, “Tobacco-Related Mortality Factsheet (updated September 2006).
http://www.cdc.gov/tobacco/data_statistics/Factsheets/tobacco_related_mortality.
htm

Commercial Closet. http://www.commercialcloset.org

Costantino, J. Gay teens and smoking: At least the Camel doesn’t call them names or
kick them out of the house. http://www.gaysmokeout.net.

Deshpande, R. and Stayman, D.M. (1994). A tale of two cities: Distinctiveness theory
and advertising effectiveness. Journal of Marketing Research, 31, 57-64.

57

Duggan, S., Eccles, M.P., Steen, N., Jones, S. Ford, G.A. (2001). Management of older
patients with hypertension in primary care: Improvement on the rule of halves.
Age and Ageing, 30, Health Module, 73-76.

Fagan P, Eisenberg M, Frazier L, Stoddard A.M., Avruinin, J.S., Sorensen, G. (2003).
Employed adolescents and beliefs about self-efficacy to avoid smoking. Addictive
Behaviors, 28, 613-626.

Fassihi, F. (2007, August). Marketing Middle East to gay tourists: Creativity, discretion
are keys to promotions by Lebanese advertiser. Wall Street Journal, Eastern
edition, B5.

Forehand, M.R., Reed (II), A., and Deshpande, R. (2002). Identity salience and the
influence of differential activation of the social self-schema on advertising
response. Journal of Applied Psychology, 87, 1086-1099.

Freeman, M. A., Hennessy, E. V., & Marzullo, D. M. (2001). Defensive evaluation of
antismoking messages among college-age smokers: The role of possible selves.
Health Psychology, 20, 424-433).

Goff, M.K. (2001). The rule of halves.
http://www.decillion.us/dmax/theories/RuleOfHalves.html

Grier, S.A. and Brumbaugh, A.M. (1999). Noticing cultural meanings; Ad meanings
created by target and non-target markets. Journal of Advertising, 28, 79-93.

Grier, S.A. and Brumbaugh, A.M. (2004). Consumer distinctiveness and advertising
persuasion. Diversity in Advertising. Mahway, NJ. Lawrence Erlbaum
Associates.

Gruskin, E.P., Greenwood, G.L., Matevia, M., Pollack, L.M., and Bye, L.L. (2007).
Disparities in smoking between the lesbian, gay, and bisexual population and the
general population of California. American Journal of Public Health, 97, 1496-
1502.

Hansen, L. (2004). Massachusetts resists courting gay and lesbian travelers with
advertising. Weekend Edition, Sunday (transcript). Taped in Washington, D.C.
http://proquest.umi.com/pqdweb?did=652525071&sid=2&Fmt=3&clientId=4676
&RQT=309&VName=PQD

Heinz.com. http://www.heinz.com/Kosher.aspx

58
Hooker, R.C., Cowap, N., Newson, R. and Freeman, G.K. (1999). Better by half:
hypertension in the elderly and the ‘rule of halves’: A primary care audit of the
clinical computer record as a springboard to improving care. Family
Practice,16,123-128.

Horovitz, B. (1993, February 23). Alternative approach finding new ways to gays,
lesbians. Los Angeles Times.

Johnson, H. (1999). Pride and politics: Gay rights 30 years after Stonewall: Taking
care of business. Institute for Public Affairs in These Times.

Katz, E. (1957), The two-step flow of communication: An up-to-date report on an
hypothesis. Public Opinion Quarterly, 21, 61-78.

Last Drag LA. www.lastdrag.org.

Martin, B.A.S., Kwai-Choi Lee, C., and Yang, F. (2004). The influence of ad model
ethnicity and self-referencing on attitudes. Journal of Advertising, 33, 27-37.

McGuire, W.J. (1984). Public communication as a strategy for inducing health-
promoting behavioral change. Preventive Medicine, 13, 299-319.

McGuire, W.J. (1989). Theoretical foundations of campaigns. In R.E. Rice and C.K.
Atkin (Eds.), Public communication campaigns (pp. 43-65). Thousand Oaks, CA:
Sage Publications.

McGuire, W.J. (1999). Constructing Social Psychology: Creative and Critical
Processes. Cambridge University Press. New York.

McGuire, W.J. and Padawer-Singer, A. (1976). Trait salience in the spontaneous self-
concept. Journal of Personality and Social Psychology, 33, 743-754.

McGuire, W.J., McGuire, C.V., Child, P., and Fujioka, T. (1978). Salience of ethnicity in
the spontaneous self-concept as a function of one’s ethnic distinctiveness in the
social environment. Journal of Personality and Social Psychology, 36, 511-520.

National Public Health Partnership, Australia.
http://www.ausport.gov.au/fulltext/2002/nphp/chapter2.5.asp

Oakenfull and Greenlee (2005). Queer eye for a gay guy: Using market-specific symbols
in advertising to attract gay consumers without alienating the mainstream.
Psychology & Marketing, 22, 421-439.

59
Oakenfull, G.K. (2004) Targeting consumer segments based on sexual orientation: Can
advertisers swing both ways? In J.D. Williams, W. Lee, and C.P. Haugtvedt
(Eds.), Diversity in Advertising ( pp. 369-381). Mahway, NJ: Lawrence Erlbaum
Associates.

Penaloza, L. (1996). We’re here, we’re queer, and we’re going shopping! A critical
perspective on the accommodation of gays and lesbians in the U.S. marketplace.
Journal of Homosexuality, 31, 9-41.

Rothman, C. (1999, May 18). Big companies are openly courting gay consumers. Los
Angeles Times.

Ryan, H., Wortley, P.M., Easton, A., Pederson, L., and Greenwood, G. (2001). Smoking
among lesbians, gays, and bisexuals: A review of the literature. American
Journal of Preventive Medicine, 21, 142-149.

Sender, K. (1999). Selling sexual subjectivities: Audiences respond to gay window
advertising. Critical Studies in Mass Communication, 16, 172-196.

Singhal, A. and Rogers, E.M. (1999). Entertainment-Education: A communication
strategy for social change. Mahway, NJ: Lawrence Erlbaum Associates.

Smith, W.C. S., Lee, A.J., Crombie, I.K., and Tustall-Pedoe, H. (1990). Control of blood
pressure in Scotland: the rule of halves. British Medical Journal, 300, 981–983.

Smoking: Risk, perception, & Policy. (2001). P. Slovic (Ed.). Thousand Oaks, CA: Sage
Publications.

Solomon, D.S. (1989). A social marketing perspective on communication campaigns.
Public Communication Campaigns, second edition. Ronald E. Rice & Charles K.
Atkin, eds. Newbury Park, CA: Sage Publications. 87-104.

Stall, R.D., Greenwood, G.L., Acree, M., Paul, J., Coates, T.J. (1999). Cigarette smoking
among gay and bisexual men. Journal of Public Health, 89, 1875-1879.

Stonewall Society. http://www.stonewallsociety.com/gaysymb.htm

Surgeon General’s Report (2001).
http://www.cdc.gov/tobacco/data_statistics/sgr/sgr_2001/00_pdfs/execsumm.pdf

Tobacco Master Settlement Agreement Summary.
http://ag.ca.gov/tobacco/resources/msasumm.php

60
U.S. Department of Health and Human Services Clinical Practice Guideline (2000).
Treating Tobacco Use and Dependence.
http://www.surgeongeneral.gov/tobacco/treating_tobacco_use.pdf

Vary, A.B. (2006, 20 June). Is gay over? The Advocate, 98-102.

Wagennar, W.A. (1969). Not on the construction of diagram-balanced Latin squares.
Psychological Bulletin, 72, 384-386.

Wilke, M. (2001). http://www.commercialcloset.org.

Wills, T.A. and Cleary, S.D. (1997). The validity of self-reports of smoking: Analyses
by race/ethnicity in a school sample of urban adolescents. American Journal of
Public Health, 87, 56-62). Yin, S. (2003). Coming out in print. American
Demographics, 25, 18-21.

Abstract (if available)

Abstract Smoking kills over 440,000 individuals annually. Lesbian, gay, bisexual, and transgendered (LGBT) individuals, compared to non-LGBT individuals, are especially likely to smoke. Anti-smoking messages that are effective for LGBT individuals, but that also reach non-LGBT individuals would be desirable: But, to date, no research has specifically examined this. Based on theoretical foundations developed by psychologists and marketing researchers, two studies (one each for LGBT and non-LGBT) were conducted. In Study 1, using a within subjects design, 61 LGBT individuals viewed eight messages that differed in whether they specifically targeted LGBT individuals (in an explicit or vague way) or did not (control). The dependent measures were the perception that a message was intended for the individual, message effectiveness, and the likelihood of sharing anti-smoking concerns with others after exposure to a message. To avoid order effects, a balanced Latin square design was used such that each message appeared before and after every other message. Study 2 used the same design and stimulus materials as in Study 1: However, the 209 participants were heterosexuals. Analyses revealed that homosexuals and heterosexuals differ in their patterns of responses to these messages. An intriguing finding was that when both LGBT and non-LGBT audiences are exposed to the same health message, “gay-vague” messages that provide symbols but no explicit text might best facilitate LGBT receptivity and mitigate non-LGBT reactivity. Based on these findings, public health campaigns that target a minority population may be able to reach non-target as well as target populations by using less explicit messages.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The articulated thoughts of heterosexual male college students in reaction to anti-gay hate speech

PDF

Identification of the targets required for the anti-tumor effect of celecoxib and its non-COX-2 inhibitory derivative

PDF

The out field: professional sports and the mediation of gay sexualities

PDF

Upvoting the news: breaking news aggregation, crowd collaboration, and algorithm-driven attention on reddit.com

PDF

The footprint of pain: investigating persistence of altered trunk control in recurrent low back pain

PDF

Trunk control during dynamic balance: effects of cognitive dual-task interference and a history of recurrent low back pain

Asset Metadata

Creator Alexander, Linda (author)

Core Title “Under the radar”: the relative effectiveness of gay-vague, gay-explicit, and control anti-smoking messages for targeted and nontargeted audiences

School Annenberg School for Communication

Degree Doctor of Philosophy

Degree Program Communication

Publication Date 10/05/2009

Defense Date 09/09/2009

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

Tag anti-smoking messages,gays,health communication,LGBT,OAI-PMH Harvest

Place Name USA (countries)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Miller, Lynn Carol (committee chair), Cody, Michael J. (committee member), Read, Stephen J. (committee member)

Creator Email alexanl@wlac.edu,lalexand@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m2644

Unique identifier UC1496108

Identifier etd-ALEXANDER-3271 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-257888 (legacy record id),usctheses-m2644 (legacy record id)

Legacy Identifier etd-ALEXANDER-3271.pdf

Dmrecord 257888

Document Type Dissertation

Rights Alexander, Linda

Type texts

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

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu