University of Southern California Dissertations and Theses

Depression trajectories and risk typology among African Americans

(USC Thesis Other)

Depression trajectories and risk typology among African Americans

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

DEPRESSION TRAJECTORIES AND RISK TYPOLOGY AMONG AFRICAN AMERICANS

By

Krystal Hays, MSW, LCSW

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(SOCIAL WORK)
August 2017

i

Dedication
I would like to dedicate this dissertation to those individuals, both past and present, who
helped to make this work possible. Throughout my doctoral program I have been continually
motivated by the memory of my forbearers, many of whom were slaves and unable to read or
write. I am thankful for their sacrifices and I do not take for granted the awesome opportunity I
have to pursue my dreams. My grandmother, Shelia Ann “Nana”, instilled in me the value of
education and the attractiveness of being smart. She always said “You get your brains from me”
and she was right. My parents have always supported my goals and helped me develop such high
self confidence that I rarely doubted my ability to be successful. I also dedicate this dissertation
to my husband, life partner, and number one supporter. He has continually allowed me the space
to live out my full potential and encouraged me when I needed it. I am grateful for his sacrifices
over the last several years. This dissertation is also dedicated to the future generations.
Specifically, this work is meant to encourage my own children, Natalie and Nathaniel, to set high
goals for themselves. Although I am the first PhD in the family I will not be the last and I hope
that my achievement has helped them see unlimited possibilities for themselves. Lastly, this
dissertation is dedicated to those lives I hope to touch through this work.

ii

Acknowledgements
I would like to thank my committee chair Dr. Karen Lincoln for being a wonderful
mentor to me throughout the doctoral program. I knew after our first conversation, more than
five years ago, that she was an amazing person but I did not realize at the time how much she
would change my life. It was only because of her that I decided to apply for the PhD program
and she has been instrumental in my professional and scholarly development. She has helped me
muddle through various ideas, research questions, and conceptual models until I found my way.
She was always there when I needed someone to talk down my anxiety and remind me that “It
will all be fine” and she was always right. I would also like to thank my gracious committee
members, Dr. Robbyn Cox, Dr. Donald Lloyd, Dr. Lourdes Baezconde-Garbanati, and Dr. Cecil
Murray for all their guidance and time devoted to this process. I am so thankful to have had the
opportunity to work with such knowledgeable and supportive faculty who are all truly invested
in my success.
I am also grateful for the support given to me by other faculty members. Specifically, I
want to acknowledge Dr. Maria Aranda and Dr. Tamika Gilreath for serving as my co-authors on
my first publications. They went above and beyond to make sure that my first experiences with
publishing were successful. My success in the doctoral program has also been made possible by
mentoring received from Dr. Robert Taylor and the MCUAAAR team and Dr. Geraldine Meeks
and the CSWE Minority Fellowship staff. I am also deeply grateful for the financial support
given me by the Suzanne Dworak-Peck School of Social Work, the CSWE Minority Fellowship,
and the Roybal Institute. They believed in the value of my work and contributed immensely to
the successful completion of my dissertation. A special note of thanks is due to Malinda

iii

Sampson who has helped me stay on track and continued to remind me that “There’s a reason
everyone’s not called Dr.”
To my doctoral colleagues, I am thankful for the encouragement they have offered. We
have gotten to share our struggles and triumphs and I could not have asked for a better group of
ladies to go through this experience with. I would also like to thank my friends and family who
have encouraged me, listened to mock presentations, read over my papers, and provided support
in large and small ways. There is a saying, “it takes a village to raise a child” but in this case, it
took a village to finish a doctoral program. Lastly, none of this would be possible without the
unending love and favor from my savior Jesus Christ. There were times when my faith and trust
in His will for my life were the only things that kept me going. I believe that He will use me and
my work to touch lives and I am thankful that of all the people in the world He could have
chosen to do this work, He chose me.

iv

Table of Contents
Dedication ........................................................................................................................................ i
Acknowledgements ......................................................................................................................... ii
List of Tables ................................................................................................................................. vi
List of Figures ............................................................................................................................... vii
Abstract ........................................................................................................................................ viii
Chapter One: Introduction and Overview of the Three Studies ..................................................... 1
Purpose ........................................................................................................................................ 3
Rationale...................................................................................................................................... 4
Theoretical Background ............................................................................................................ 10
A Conceptual Model of Depression Risk Typologies and Trajectories .................................... 11
Organization of Dissertation ..................................................................................................... 11
Strengths and Implications ........................................................................................................ 13
References ................................................................................................................................. 17
Chapter Two (Study 1): A Risk Typology for Depression in African Americans ....................... 26
Background ............................................................................................................................... 28
Methods ..................................................................................................................................... 32
Results ....................................................................................................................................... 37
Discussion ................................................................................................................................. 39
References ................................................................................................................................. 52
Chapter Three (Study 2): Identifying Paths to Depression: Stability and Transitions in Risk for
Depression among African Americans ......................................................................................... 61
Background ............................................................................................................................... 61
Methods ..................................................................................................................................... 68
Results ....................................................................................................................................... 73
Discussion ................................................................................................................................. 76
References ................................................................................................................................. 93
Chapter Four (Study 3): Variable Depression Trajectories among African Americans ............. 101
Background ............................................................................................................................. 102
Methods ................................................................................................................................... 108
Results ..................................................................................................................................... 111
Discussion ............................................................................................................................... 114

v

References ............................................................................................................................... 129
Chapter Five: Conclusion ........................................................................................................... 138
Summary of Major Findings ................................................................................................... 138
Implications ............................................................................................................................. 144
Limitations .............................................................................................................................. 151
References ............................................................................................................................... 153

vi

List of Tables
Table 2.1. Overall Demographic and Risk Characteristics
Table 2.2. Fit Statistic Comparisons of LCA Models of Risk Types among African Americans
Table 2.3. Conditional Probabilities of Risk Types for Depression
Table 2.4. Multinomial Logistic Regression Results
Table 2.5 Mean CESD Score by Risk Type
Table 3.1. Overall Characteristics of African Americans at Wave 1
Table 3.2. Model Fit Indexes for Latent Transition Models with Two to Four Classes
Table 3.3. Item Response Probabilities for Latent Risk Types
Table 3.4. Transition Probabilities
Table 3.5. Multinomial Logistic Regression of Baseline Demographic Covariates on Risk
Transitions
Table 3.6. Differences in CES-D Score for Risk Transition Types
Table 4.1. Overall Demographic and Risk Type Distributions
Table 4.2. Fit Statistics for Growth Mixture Models with 1 to 7 Latent Classes
Table 4.3. Estimated Growth Factor Means
Table 4.4. Multinomial Logistic Regression of Demographic Covariates on Depression
Trajectories
Table 4.5. Risk Types and Demographic Characteristics of Depression Trajectories
Table 4.6. Multinomial Logistic Regression of Risk and Protective Types on Depression
Trajectories

vii

List of Figures
Figure 1.1. Conceptual Model of Depression Risk Typologies and Trajectories in African
American Adults Aged 25 or Older
Figure 2.1. Probabilities for Risk and Protective Factors for each Risk Type
Figure 3.1 Conceptual Model of Depression Risk Transitions in African American adults ages 25
and older
Figure 3.2. Probabilities for Risk and Protective Factors for each Risk Type
Figure 3.3. Proportion of Respondents in Each Transition Type
Figure 4.1 Conceptual Model of Depression Trajectories in African American adults ages 25 and
older
Figure 4.2. Growth Curves for 5 Trajectory Classes of Depressive Symptoms for African
Americans

viii

Abstract
Although rates of clinical depression for African Americans are low, research suggests
that disease burden is high as African Americans are likely to face chronic, debilitating, severe,
and persistent depressive symptoms. To alleviate the excess burden of depression in African
Americans we must better understand the various factors that contribute to depressive symptoms
in this population and how these factors vary within the population. The Cumulative
Advantage/Disadvantage (CAD) theory suggests that there may be specific constellations of risk
and protective factors that constitute paths toward better or worse depression outcomes. It also
suggests that early life exposure to certain life conditions, and the advantages and disadvantages
they engender, will impact depressive symptoms over the life course. The overall purpose of this
dissertation is to fill gaps in the current literature and uncover contributors to the excess burden
of depression in African Americans by exploring heterogeneity within the population and in the
experience of depressive symptoms. Specifically, this study examines distinct profiles of risk and
protective factors (types), transitions in risk and protective factors over time, and various
depression courses (trajectories) among African American adults aged 25 or older.
Using data from the Americans’ Changing Lives study, and the CAD theory as a
conceptual framework, this dissertation examines three primary research questions: 1) Are there
specific risk profiles (i.e. types) that can be characterized by constellations of psychosocial,
sociocultural factors, and individual demographic characteristics? 2) Do individual African
Americans transition or remain stable in their risk type over time? and 3) Are there multiple
courses of depression for African Americans? If so, are these courses (i.e. trajectories)
differentiated by psychosocial, sociocultural, and demographic factors?
The dissertation is presented in a multiple manuscript format, including three studies that
are distinct but related. Empirical study 1 (chapter 2) investigated unique constellations of

ix

multiple co-occurring risk and protective factors that offer greater explanatory potential
regarding depression burden for African Americans than those measures treated independently.
Study results revealed a risk typology with three distinct risk patterns for African Americans and
identified risks types that were associated with worse depressive symptoms. Study 2 (chapter 3)
identified distinct constellations of risk and protection for depression and investigated individual
stability, or transitions, between risk types over time using Latent Transition Analysis. Results
confirmed that there are distinct classes of risk and protection and further identified eight distinct
risk paths. Those who transitioned between risk types were found to have higher depressive
symptoms than those with stable risk. Empirical paper 3 (chapter 4) examined heterogeneity in
the course of depressive symptoms among African American adults and older adults to identify
variable depression trajectories and their demographic, psychosocial, and sociocultural
correlates. Growth mixture modeling was employed and results suggested that there are 5 distinct
depression trajectories for African Americans: Low Symptoms, High Symptoms, Increasers,
Slow Decliners, and Fast Decliners. Several factors were found to predict trajectory membership
including gender, age, education, marital status, and their psychosocial/sociocultural risk type.
Overall, this dissertation provides information needed to develop depression interventions
that are aimed at a moving target and tailored to the unique needs of subpopulations of African
Americans likely to experience poor depression outcomes. This study’s results suggest specific
combinations of protective factors that should be promoted among African Americans because
they buffer against depression. Also, a subpopulation of African Americans exists that is likely to
experience persistent or increasing depression. Information on the demographic profile of those
African Americans who are likely to suffer excessively should be consulted when designing
tailored interventions.
Hays
1

Chapter One: Introduction and Overview of the Three Studies
African Americans have lower prevalence rates of depression compared to non-Hispanic
Whites (10% vs. 17%, respectively; Breslau et al., 2006). However, African Americans have
more depressive symptoms (Bromberger, Harlow, Avis, Kravitz, & Cordal, 2004; Romero, Ortiz,
Finley, Wayne, & Lindeman, 2005) and when they are diagnosed with depression, it is often
more debilitating and persistent compared to Whites (Williams et al., 2007). Prior race
comparison studies have described the excess burden of depression in African American adults
which is evidenced by poorer depression prognoses (Williams et al., 2007), more chronic
depressive symptoms (U.S. Department of Health and HumanServices, 2011), more functional
disability (Noël et al., 2004), and more comorbid chronic health conditions (Pickett, Bazelais, &
Bruce, 2013) compared to Whites. The disproportionate burden of depression on African
Americans is a public health concern that negatively affects quality of life (Garbarski, 2015) and
creates a costly burden for the health care system (Donohue & Pincus, 2007).
Some studies have attempted to explain why African Americans have more depressive
symptoms and worse outcomes by exploring various psychosocial, sociocultural, and individual
risk and protective factors. Factors such as socioeconomic status (Salami & Walker, 2014),
social relationships (Santini, Koyanagi, Tyrovolas, Mason, & Haro, 2015), psychosocial stress
(Keyes, Barnes, & Bates, 2011), and religious involvement (Taylor, Chatters, & Abelson, 2012)
have been found to be related to depressive symptoms and outcomes in this group. Although this
body of research has significantly contributed to our understanding of the various factors
associated with depression in African Americans, it remains unclear why African Americans
experience an excess burden related to depression.
Hays
2

At least three underexplored issues might explain poor depression outcomes among
African Americans. First, studies of depression in African Americans typically treat this
population as a monolithic group and thus do not consider vast within-group variation. Race
comparative studies, although informative, potentially obscure important heterogeneity in the
African American population that might help us better understand which individuals are likely to
have poor depression outcomes and what factors contribute to this excessive depression burden.
Studies that account for heterogeneity in the African American population have the potential to
identify subgroups of individuals that are most likely to experience the poor depression outcomes
described by prior race comparison studies. This is important because it may be that African
Americans as a whole do not experience excess depression burden but that there is a
subpopulation of individuals who suffer in excess. Within group studies will help to differentiate
individuals with a risk profile linked to worse depression. Further, examining within-group
differences will help elucidate specific associations between individual characteristics and
psychosocial and sociocultural factors that contribute to the excess burden of depression in
African Americans.
Second, although studies have linked risk and protective factors to depression, few
studies have investigated the co-occurrence of these factors by identifying depression risk
profiles, or types, based on specific constellations of depression risk factors. Studies that
examine the co-occurrence of multiple sociocultural and psychosocial factors associated with
depression in African Americans might provide insight into the collection of life conditions that
contributes to excess depressive symptoms in this group. Further, life course research has
suggested that the experience of risk and protective factors may change throughout an
individual’s life to affect health outcomes. Thus, studies are needed that can empirically identify
Hays
3

the accumulation of risk and protection for African Americans over time. Third, the course of
depression is variable, with periods of remission, recurrence, and stability (Richards, 2011), yet
few studies have investigated this variability in African Americans. Identifying multiple
depression trajectories among African Americans might illuminate subpopulations of individuals
with a course of depression that is associated with more depressive symptoms and worse
outcomes. Again, this information would help to disaggregate what is known from prior research
that treats African Americans as a monolithic group and highlight the specific subgroup of
individuals who have a more chronic and persistent course of symptoms as these individuals are
likely to experience excess burden.
High depression burden for African Americans represent a significant public health
problem. The extant literature on depression in African Americans is limited in terms of
explaining why African Americans experience these negative outcomes. Studies that investigate
profiles of co-occurring depression risk factors (i.e., risk types), examine how depression risk
accumulates and changes over time (i.e., risk transitions), and identify multiple courses of
depression (i.e., depression trajectories) are needed to illuminate possible contributors to the
excess burden of depression in African Americans. Findings from this approach will be useful
for identifying and targeting profiles of African Americans most likely to experience poor
depression outcomes and tailoring interventions to these subpopulations of individuals based on
their particular risk profile and specific course of depression.
Purpose
The overall purpose of this dissertation is to uncover contributors to the excess burden of
depression by identifying distinct profiles of risk and protective factors (types), transitions in risk
and protective factors over time, and various depression courses (trajectories) among African
American adults aged 25 or older. Results will inform the development of depression
Hays
4

interventions tailored to meet the needs of individuals who match to specific risk profiles and
depression trajectories.
The dissertation contributes to the literature on depression in African Americans by
examining three primary research questions:
1. Can specific risk profiles (i.e., types) be identified based on constellations of
psychosocial, sociocultural factors, and individual demographic characteristics among
African Americans?
2. Do individual African Americans transition or remain stable in their risk type over
time?
3. Do African Americans experience multiple courses of depression and if so, are these
courses (i.e., trajectories) differentiated by psychosocial, sociocultural, and
demographic factors?
Rationale
African Americans and Depression
To address the excess burden of depression in African Americans adequately, the disease
must be understood within a psychosociohistorical context. African Americans have endured the
legacy of slavery, Jim Crow, institutional racism, and discrimination in the United States. Also,
African Americans are disproportionately exposed to psychosocial stressors (e.g., economic
deprivation, violent neighborhoods, racial discrimination) that negatively affect mental health
and well-being (Mizell, 1999). African Americans continue to face disparities in health and
social issues including higher rates of homicide victimization and incarceration (Boyd, 2007;
Rogers, Rosenblatt, Hummer, & Krueger, 2001), lower life expectancy, and higher infant
mortality (Xu, Murphy, Kochanek, & Bastian, 2016). Further, abuses of African Americans by
Hays
5

the medical community including atrocities like the Tuskegee syphilis experiments have
understandably made many African Americans distrustful of medical professionals.
In many ways, this sociohistorical context, which is marked by the normative experience
of negative life conditions, has shaped cultural norms regarding the expression and treatment of
mental disorders like depression. Studies have suggested that the qualitative nature of depression
may be unique in African American populations. For example, African Americans with
depression tend to exhibit somatic symptoms more often than Whites, especially older adults
(Barry et al., 2014). Also, African Americans are more likely to externalize depressive symptoms
and be diagnosed with conduct and psychotic disorders than Whites (Atdjian & Vega, 2005;
Woods, King, Hanna, & Murray, 2012). Evidence suggests that many more African Americans
experience depressive symptoms than are actually diagnosed with clinical depression because of
standards of clinical significance (Miller et al., 2004). Depressive symptoms, in the absence of
clinical depression, can be quite debilitating. Yet many studies that investigated depression
among African Americans only considered individuals who met criteria for clinical depression,
when general population samples might be more appropriate (Richards, 2011; Sutin et al., 2013).
Because depressive symptoms, in the absence of clinical depression, continue to represent a
public health concern that negatively affects the functioning of many African Americans, it is
worthy of more research focus.
Regarding the treatment of depression, research has suggested that African Americans
tend to rely on several informal sources of support, like family and clergy, along with primary
care providers rather than formal mental health services (Hays & Gilreath, 2016). Historically,
trusted institutions like churches were often the only culturally acceptable venue for African
Americans to voice their fears, anger, and sadness about living in a racially oppressive and
Hays
6

hostile society (Hays, 2015; C. E. Lincoln & Mamiya, 1990). Thus, religious coping became a
common strategy for many African Americans with depressive symptoms. Qualitative interviews
have indicated that many African Americans are taught to rely on God to manage mental and
emotional problems and are reluctant to seek mental health treatment because of discrimination,
insensitivity, and lack of understanding of their experiences by mental health providers
(Matthews, Corrigan, Smith, & Aranda, 2006). Help-seeking norms for African Americans
should not be understood as simple provider preference but rather are indicative of significant
barriers to accessing quality health care, including discrimination among treatment providers,
inadequate treatment, and cultural mistrust of medical providers experienced by African
Americans (Gary, 2005; Whitley & Lawson, 2010).
Although many African Americans share a common sociohistorical experience,
heterogeneity exists in this population, with great variation by region, socioeconomic status,
health profile, and many other factors (Williams & Jackson, 2000). Despite the fact that African
Americans are a heterogeneous population, many studies of depression have treated them as a
monolithic group to facilitate comparisons to other racial and ethnic groups. Race comparison
studies, although informative, may obscure important variation in the African American
population that can be used to distinguish subpopulations of individuals with different depression
experiences (e.g., African American woman with a high school education and low income).
Prior studies of depression that considered within-group differences have identified
subpopulations of African Americans with varying profiles based on depressive symptoms (K.
D. Lincoln, Chatters, Taylor, & Jackson, 2007), help-seeking behavior (Hays & Gilreath, 2016;
Woodward, Taylor, & Chatters, 2011), and level of distress (K. D. Lincoln, Taylor, Watkins, &
Chatters, 2011). Specifically, a study by K. D. Lincoln and associates (2007) identified two
Hays
7

distinct groups of Black Americans based on their depressive symptom profiles. They concluded
that heterogeneity exists in this population that can be characterized by either high or low
symptomatology and described by a specific sociodemographic profile (K. D. Lincoln et al.,
2007). Findings from this small body of research highlight the importance of considering within-
group heterogeneity among African Americans to explicate disparate depression outcomes.
Risk and Protective Factors for Depression
Previous cross-sectional studies have identified several social and behavioral correlates of
depression in African Americans, including demographic characteristics, psychosocial stress,
social relationships, and religion. Studies have linked demographic characteristics, like
socioeconomic status, age, gender, and marital status, to depression (Chatters, Taylor,
Woodward, & Nicklett, 2015; K. D. Lincoln, Abdou, & Lloyd, 2014; K. D. Lincoln et al., 2011;
Salami & Walker, 2014). For example, some results suggested that African Americans with less
than a high school education and those with low incomes are at increased risk of depression
(Hudson, Neighbors, Geronimus, & Jackson, 2012). However, these associations are
inconclusive because other studies have reported a positive association, with higher
socioeconomic status being linked to more depressive symptoms (Salami & Walker, 2014).
Psychosocial stress (e.g., financial stress) and negative life events (e.g., violence, trauma)
have also been linked to increased depressive symptoms in African Americans (Mitchell &
Ronzio, 2011). Some African Americans experience multiple adverse social conditions such as
community violence and poverty that contribute to the development of depressive symptoms. For
example, research by Cutrona et al. (2005) found that the experience of negative life events
significantly predicted the onset of major depression among African American women. Other
studies have also concluded that experiencing financial stress or events like a death in the family
Hays
8

or being assaulted is associated with increased depressive symptoms among African American
adults (Cutrona et al., 2005; Lantz, House, Mero, & Williams, 2005; Mitchell & Ronzio, 2011;
Turner & Avison, 2003).
Social relationships have been identified as both a risk and protective factor for
depression in African Americans. Specifically, low social support (Teo, Choi, & Valenstein,
2013), poor overall relationship quality, and negative social interactions (i.e., critical or
demanding social relationships; K. D. Lincoln & Chae, 2012) are associated with an increased
risk of depression. Conversely, positive social relationships and larger social networks are
protective and associated with a lower risk of depression (Chatters et al., 2015; K. D. Lincoln &
Chae, 2012).
Religious involvement has also been linked to depression. Religion is generally
considered to be a protective factor associated with improved mental health and well-being
(Moreira-Almeida, Lotufo Neto, & Koenig, 2006). For example, one study found that African
Americans who reported receiving a “great deal of guidance” from religion were about half as
likely to have major depression as those with lower levels of religiosity (Ellison & Flannelly,
2009). Studies also suggested that African Americans who attend religious services more
frequently are less likely to have depression than infrequent church attendees (Chatters et al.,
2008).
This body of research provided insight into the various risk and protective factors
associated with depression among African Americans. However, many of these cross-sectional
studies did not examine patterns of co-occurring risk factors associated with depression.
Evidence suggests that health risks should be studied simultaneously rather than focusing on the
effect of individual risk factors (Chen, Eaton, Gallo, & Nestadt, 2000; Laska, Pasch, Lust, Story,
Hays
9

& Ehlinger, 2009). Individuals face an array of suboptimal psychosocial and physical conditions
that interact in complex ways to influence the development of mental illness (Copeland,
Shanahan, Costello, & Angold, 2009). Studies investigating multiple risk factors in other
populations have concluded that risk is a configuration, or constellation, of experiences that can
help us to uncover etiological pathways to mental illness (Copeland et al., 2009). Thus
identifying profiles, or types, of co-occurring risk may explicate the collective role these factors
play in contributing to depression among African Americans.
Further, life course research has suggested that health risks vary and accumulate over
time. Some studies have found that risk factors, like socioeconomic status, stress, and social
support, are dynamic and change during the course of an individual’s life (Aneshensel &
Frerichs, 1982; Lorant et al., 2007). Cross-sectional studies that examine risk factors cannot
provide insight into how risk changes during the life course. Longitudinal studies are needed to
identify profiles of multiple risk factors and examine how those risk profiles remain stable or
transition over time for African Americans.
Course of Depression
The course of depression is highly variable, with periods of symptom persistence,
remission, and recurrence (Liang, Xu, Quiñones, Bennett, & Ye, 2011; K. D. Lincoln &
Takeuchi, 2010; Spence, Adkins, & Dupre, 2011) . Longitudinal studies have indicated that
depressive symptoms can improve, worsen, or remain stable at high or low levels over time (K.
D. Lincoln & Takeuchi, 2010; Spence et al., 2011)(K. D. Lincoln & Takeuchi, 2010; Spence et
al., 2011). However, variation in the course of depression for African Americans is rarely
documented. Thus the demographic, psychosocial, and sociocultural factors associated with the
course of depression in African Americans remain largely unknown.
Hays
10

Longitudinal studies are needed to increase our understanding of changes in depression
symptoms in African Americans over time and identify whether heterogeneity exists in
depression trajectories in this population. Understanding how the course of depression varies
among African Americans can provide insight into the excess burden of depression by
accounting for potentially divergent trajectories that might offset one another if considered in the
aggregate and confirm that African American have high levels of depressive symptomatology,
low prevalence rates, and higher disease burden, as observed in prior studies.
Theoretical Background
Theory and research have suggested that some health risks accumulate across the life
course in various ways to influence depression outcomes (Garbarski, 2015). This dissertation
employed the cumulative advantage/disadvantage (CAD) theory, focusing on the particular
accumulation of sociocultural and psychosocial risk factors for depression and the variable
course of depression to help explain how different risk types influence depressive symptoms and
trajectories among African Americans.
The CAD model suggests that early life advantages and disadvantages persist, or
accumulate, into later life and are often magnified during the life course (Crystal, Shea, & Reyes,
2016). Specifically, the path-dependent mechanism suggests that early life exposure to certain
risk factors (e.g., low socioeconomic status, chronic stress) leads to poor health outcomes in later
life (DiPrete & Eirich, 2006; Willson, Shuey, & Elder, 2007). For example, an African American
woman with low education, low income, and low social support as a young adult may have poor
mental health as an older adult because of the direct and indirect effects of these risks via the life
conditions they engender. However, whether these risk and protective factors remain stable or
change over time for African Americans remains unknown. Thus, existing research has not
conclusively suggested specific constellations of risk factors that constitute a path toward worse
Hays
11

depression outcomes for African Americans. Understanding the composition of individual risk
profiles, or types, and how these types describe paths toward depression will help explain why
African Americans have poor depression outcomes. As such, this dissertation sought to identify
specific profiles of risk and protective factors for African Americans, examine how risk changes
during the life course, and determine how this risk typology influences depression outcomes to
illuminate paths to depression suggested by CAD theory.
A Conceptual Model of Depression Risk Typologies and Trajectories
A conceptual model served as a guide for this dissertation (Figure 1). The model
incorporates conceptual, theoretical, and empirical elucidations from the extant literature on risk
and protective factors for depression among African Americans. Included are specific
demographic, psychosocial, and sociocultural factors that prior literature has shown to be
associated with depression outcomes in African Americans. In this model, specific constellations
of risk and protective factors are considered risk types and variable paths toward depression are
modeled as depression risk transitions. Last, depression courses are included in the model to
capture variations in depressive symptom persistence.
Organization of Dissertation
This dissertation conforms to a multiple-manuscript model and includes three empirical
studies of publishable quality. This introduction chapter highlights the extant literature on
depression in African Americans and the concluding chapter integrates findings and discusses
implications of all three papers.
The analytic framework was as follows: (a) identify classes of depression risk and
protective factors (depression risk types); (b) test for variation in associations between risk and
protective types and depressive symptoms by demographic characteristics (e.g., age, gender,
socioeconomic status); (c) identify transitions between risk types over time; (d) identify baseline
Hays
12

demographic features that predict risk transitions, (e) confirm the existence of multiple courses
of depression symptomatology as hypothesized; and (f) identify associations between depression
courses and psychosocial, sociocultural, and demographic factors.
Empirical Paper 1 (Chapter 2) addressed the following research question: Can a specific
depression risk typology be developed based on psychosocial, sociocultural factors, and
individual demographic characteristics among African Americans? This study sought to
empirically identify depression risk profiles, or types, among African American adults that are
characterized by unique constellations of psychosocial factors (i.e., chronic financial stress,
negative life events, social support, negative interactions) and a sociocultural factor (i.e.,
religious involvement). These factors were selected based prior literatures suggestion that they
are related to depressive symptoms in African Americans. Latent class analysis was used to
identify distinct risk types in the sample. Multiple logistic regression was employed to identify
associations between risk types and demographic characteristics (e.g., age, gender,
socioeconomic status). Differences in depressive symptoms by risk type were also examined.
Data from African American respondents in Wave 1 of the Americans’ Changing Lives survey
were used.
Empirical Paper 2 (Chapter 3) addressed the following research question: Do individual
African Americans remain stable or transition between risk types over time? This study
examined the extent to which African Americans transition from one risk type to another over
time. Three waves of Americans’ Changing Lives data and latent transition analysis were used.
Latent transition analysis is an extension of latent class analysis and assumes that unobserved
(latent) categorical variables explain the associations between observed measures. Two sets of
parameters were estimated: (a) probabilities describing the marginal distribution of latent classes
Hays
13

at baseline and transition probabilities between latent classes over time and (b) conditional
probabilities (i.e., probabilities that a symptom is present given the participant belongs to a
class). After identifying risk types and profiles using latent class analysis, latent transition
analysis was used to describe change over time (stability or transitions). Stepwise multinomial
logistic regression was used to assess whether baseline variables (e.g., gender, socioeconomic
status) can predict path memberships of interest. Additionally, path membership was used to
predict differences in depressive symptoms at Wave 5.
Empirical Paper 3 (Chapter 4) addressed the final research question: Does the course of
depression vary and can distinct courses of depression be identified among African American
adults? Further, if multiple trajectories of depression exist, can they be differentiated by
psychosocial, sociocultural, and demographic factors? This study sought to identify multiple
trajectories (i.e., courses) of depression among African Americans over time. Latent growth
mixture modeling was used to identify unique trajectories of depression characterized by distinct
patterns of change in depressive symptoms. The depression trajectories were then examined for
possible associations with participants’ psychosocial, sociocultural, and individual factors.
Multinomial logistic regression was used to identify baseline covariates (e.g., age, social support,
stress, risk type) associated with depression trajectories.
Strengths and Implications
The primary strength of this dissertation is that it employed a novel approach to uncover
contributors to depression in African Americans to better understand the excess burden of
depression described in the extant literature. The dissertation contributes to the literature by
addressing several conceptual, theoretical, and methodological limitations in the extant literature
on depression in African Americans. First, this study explored the degree to which risk types can
be described as a constellation of psychosocial, sociocultural, and individual-level
Hays
14

characteristics. The identification of risk profiles for depression is currently underexplored in the
literature and can provide information about a risk typology that can be used for primary
prevention and intervention efforts. For example, if this study empirically identified a risk profile
composed of a specific constellation of social support, religiosity, and stress coupled with a
demographic profile (e.g., men with low income who are unmarried and older than 50),
interventions can be developed that target risk types associated with worse depression outcomes
and are tailored to the demographic profile of individuals most at risk.
Second, this dissertation used longitudinal data, which is helpful for understanding the
accumulation of risk over time (Clarke, Marshall, House, & Lantz, 2011). This is important
because theory suggests that to address disparate depression outcomes, research must uncover
how risk factors influence depression outcomes during the life course. By investigating
transitions in risk types over time, we can better understand the dynamic nature of risk and
protective factors and design interventions aimed at a moving target.
Third, the dissertation sought to identify various courses of depression for African
Americans. This is a strength of the dissertation because existing research is limited in its ability
to distinguish subgroups of African Americans based on their experience of depressive
symptoms. Uncovering various courses of depression (e.g., persistently high, decreasing, or
increasing symptoms) will help disaggregate what is known from depression prevalence studies
among African Americans and highlight the courses marked by more persistent or chronic
symptoms. This has direct implications for the development of interventions to address the
excess burden of depression in African Americans by pinpointing individuals who are likely to
experience a worse course of depression. For example, if study results suggest that a
demographic profile (e.g., women younger than 29 with high incomes) of individuals experience
Hays
15

a particular course of depression (e.g., increasing from low to high), interventions can be tailored
to address contributors to this increasing course of depression among individuals who fit that
demographic profile.
In addition to increased knowledge and implications for depression interventions, results
from these studies can help to set future research and policy priorities. Namely, this approach
highlights the need for researchers to focus on uncovering contributors to poor depression
outcomes in African Americans as the key to reducing the burden of depression in this
population. Further, methodologies that allow researchers to investigate risk and protective
factors as a collection of events will more closely align with the real-life experiences of African
Americans. Health policies should focus on advancing research and developing interventions that
provide individualized treatments to reduce the burden of depression for subpopulations of
African Americans that are most likely to suffer disproportionately. This dissertation provides
evidence to support future research and development in this area.

Hays
16

Hays
17

References
Aneshensel, C. S., & Frerichs, R. R. (1982). Stress, support, and depression: A longitudinal
causal model. Journal of Community Psychology, 10(4), 363–376.
http://doi.org/10.1002/1520-6629(198210)10:4<363::AID-JCOP2290100407>3.0.CO;2-8
Atdjian, S., & Vega, W. (2005). Disparities in mental health treatment in U . S . racial and ethnic
minority groups : Implications for psychiatrists. Psychiatric Services, 56(12), 1600–1602.
http://doi.org/10.1176/appi.ps.56.12.1600
Barry, L. C., Thorpe, R. J., Penninx, B. W. J. H., Yaffe, K., Wakefield, D., Ayonayon, H. N., …
Simonsick, E. M. (2014). Race-related Differences in Depression onset and recovery in
older persons over time: The health, aging, and body composition study. American Journal
of Geriatric Psychiatry, 22(7), 682–691. http://doi.org/10.1016/j.jagp.2013.09.001
Blau, P. M., & Duncan, O. D. (1967). The American occupational structure. New York, NY:
Free Press.
Boyd, H. (2007). It’s hard out here for a Black man! The Black Scholar, 37(3), 2–9.
http://doi.org/10.1080/00064246.2007.11413403
Breslau, J., Aguilar-Gaxiola, S., Kendler, K. S., Su, M., Williams, D. R., & Kessler, R. C.
(2006). Specifying race-ethnic differences in risk for psychiatric disorder in a USA national
sample. Psychological Medicine, 36(1), 57–68. http://doi.org/10.1017/S0033291705006161
Bromberger, J. T., Harlow, S., Avis, N., Kravitz, H. M., & Cordal, A. (2004). Racial/ethnic
differences in the prevalence of depressive symptoms among middle-aged women: The
Study of Women’s Health Across the Nation (SWAN). American Journal of Public Health,
94(8), 1378–1385. http://doi.org/10.2105/AJPH.94.8.1378
Chatters, L. M., Bullard, K. M., Taylor, R. J., Woodward, A. T., Neighbors, H. W., & Jackson, J.
S. (2008). Religious participation and DSM IV disorders among older African Americans:
Hays
18

Findings from the National Survey of American Life. American Journal of Geriatric
Psychiatry, 16(12), 957–965. http://doi.org/10.1097/JGP.0b013e3181898081.
Chatters, L. M., Taylor, R. J., Woodward, A. T., & Nicklett, E. J. (2015). Social support from
church and family members and depressive symptoms among older African Americans.
American Journal of Geriatric Psychiatry, 23(6), 559–567.
http://doi.org/10.1016/j.jagp.2014.04.008
Chen, L. S., Eaton, W. W., Gallo, J. J., & Nestadt, G. (2000). Understanding the heterogeneity of
depression through the triad of symptoms, course and risk factors: A longitudinal,
population-based study. Journal of Affective Disorders, 59(1), 1–11.
http://doi.org/10.1016/S0165-0327(99)00132-9
Clarke, P., Marshall, V., House, J., & Lantz, P. (2011). The social structuring of mental health
over the adult life course: Advancing theory in the sociology of aging. Social Forces, 89(4),
1287–1313. http://doi.org/10.1093/sf/89.4.1287
Copeland, W., Shanahan, L., Costello, E. J., & Angold, A. (2009). Configurations of common
childhood psychosocial risk factors. Journal of Child Psychology and Psychiatry, 50(4),
451–459. http://doi.org/10.1111/j.1469-7610.2008.02005.x
Crystal, S., Shea, D. G., & Reyes, A. M. (2016). Cumulative advantage, cumulative
disadvantage, and evolving patterns of late-life inequality. Gerontologist, Advance on, 1–
11. http://doi.org/10.1093/geront/gnw056
Cutrona, C. E., Russell, D. W., Brown, P. A., Clark, L. A., Hessling, R. M., & Gardner, K. A.
(2005). Neighborhood context, personality, and stressful life events as predictors of
depression among African American women. Journal of Abnormal Psychology, 114(1), 3–
15. http://doi.org/10.1037/0021-843X.114.1.3
Hays
19

DiPrete, T. a., & Eirich, G. M. (2006). Cumulative advantage as a mechanism for inequality: A
review of theoretical and empirical developments. Annual Review of Sociology, 32(1), 271–
297. http://doi.org/10.1146/annurev.soc.32.061604.123127
Donohue, J., & Pincus, H. (2007). Reducing the societal burden of depression: A review of the
economic costs, quality of care and effects of treatment. Pharmacoeconomics, 25(1), 7–24.
http://doi.org/10.2165/00019053-200725010-00003
Ellison, C. G., & Flannelly, K. (2009). Religious involvement and risk of major depression in a
prospective nationwide study of African American adults. Journal of Nervous and Mental
Disease, 197(8), 568–573. http://doi.org/10.1097/NMD.0b013e3181b08f45
Ferraro, K. F., & Kelley-Moore, J. A. (2003). Cumulative disadvantage and health: long-term
consequences of obesity? American Sociological Review, 68(5), 707–729.
http://doi.org/10.2307/1519759
Garbarski, D. (2015). Racial/ethnic disparities in midlife depressive symptoms: The role of
cumulative disadvantage across the life course. Advances in Life Course Research, 23, 67–
85. http://doi.org/10.1016/j.alcr.2014.12.006
Gary, F. A. (2005). Stigma: Barrier to mental health care among ethnic minorities. Issues in
Mental Health Nursing, 26(10), 979–999. http://doi.org/10.1080/01612840500280638
Hays, K. (2015). Black churches’ capacity to respond to the mental health needs of African
Americans. Social Work & Christianity, 42(3), 296–312.
Hays, K., & Gilreath, T. (2016). Profiles of depression help seeking among Black Americans: A
latent class approach. Journal of Nervous and Mental Disease, Advanced online
publication. http://doi.org/10.1097/NMD.0000000000000575
Hudson, D. L., Neighbors, H. W., Geronimus, A. T., & Jackson, J. S. (2012). The relationship
Hays
20

between socioeconomic position and depression among a US nationally representative
sample of African Americans. Social Psychiatry and Psychiatric Epidemiology, 47, 373–
381. http://doi.org/10.1007/s00127-011-0348-x
Keyes, K. M., Barnes, D. M., & Bates, L. M. (2011). Stress, coping, and depression: Testing a
new hypothesis in a prospectively studied general population sample of US-born Whites
and Blacks. Social Science & Medicine, 72(5), 650–659.
http://doi.org/10.1016/j.socscimed.2010.12.005
Lantz, P. M., House, J. S., Mero, R. P., & Williams, D. R. (2005). Stress, life events, and
socioeconomic disparities in health: Results from the Americans’ Changing Lives study.
Journal of Health and Social Behavior, 46(3), 274–288.
http://doi.org/10.1177/002214650504600305
Laska, M. N., Pasch, K. E., Lust, K., Story, M., & Ehlinger, E. (2009). Latent class analysis of
lifestyle characteristics and health risk behaviors among college youth. Prevention Science,
10(4), 376–386. http://doi.org/10.1007/s11121-009-0140-2
Liang, J., Xu, X., Quiñones, A. R., Bennett, J. M., & Ye, W. (2011). Multiple trajectories of
depressive symptoms in middle and late life: Racial/ethnic variations. Psychology and
Aging, 26(4), 761–777. http://doi.org/10.1037/a0023945
Lincoln, C. E., & Mamiya, L. H. (1990). The Black church in the African American experience.
Durham, NC: Duke University Press.
Lincoln, K. D., Abdou, C. M., & Lloyd, D. A. (2014). Race and socioeconomic differences in
obesity and depression among Black and non-Hispanic white Americans. Journal of Health
Care for the Poor and Underserved, 25, 257–275.
Lincoln, K. D., & Chae, D. H. (2012). Emotional support, negative interaction and major
Hays
21

depressive disorder among African Americans and Caribbean Blacks: Findings from the
National Survey of American Life. Social Psychiatry and Psychiatric Epidemiology, 47(3),
361–372. http://doi.org/10.1007/s00127-011-0347-y
Lincoln, K. D., Chatters, L. M., Taylor, R. J., & Jackson, J. S. (2007). Profiles of depressive
symptoms among African Americans and Caribbean Blacks. Social Science & Medicine,
65(2), 200–213. http://doi.org/10.1016/j.socscimed.2007.02.038
Lincoln, K. D., & Takeuchi, D. T. (2010). Variation in the trajectories of depressive symptoms:
results from the Americans’ Changing Lives study. Biodemography and Social Biology,
56(1), 24–41. http://doi.org/10.1080/19485561003709180
Lincoln, K. D., Taylor, R. J., Watkins, D. C., & Chatters, L. M. (2011). Correlates of
psychological distress and major depressive disorder among African American men.
Research on Social Work Practice, 21(3), 278–288.
http://doi.org/10.1177/1049731510386122
Lorant, V., Croux, C., Weich, S., Deliège, D., Mackenbach, J., & Ansseau, M. (2007).
Depression and socio-economic risk factors : 7-year longitudinal population study. British
Journal of Psychiatry, 190, 293–298. http://doi.org/10.1192/bjp.bp.105.020040
Matthews, K. A., Corrigan, P. W., Smith, B., & Aranda, F. (2006). A qualitative exploration of
African Americans’ attitudes toward mental illness and mental illness treatment seeking.
Rebahilitation Education, 20(4), 253–268. http://doi.org/10.1891/088970106805065331
Miller, D. K., Malmstrom, T. K., Joshi, S., Andresen, E. M., Morley, J. E., & Wolinsky, F. D.
(2004). Clinically relevant levels of depressive symptoms in community-dwelling middle-
aged African Americans. Journal of the American Geriatrics Society, 52(5), 741–748.
http://doi.org/10.1111/j.1532-5415.2004.52211.x
Hays
22

Mitchell, S. J., & Ronzio, C. R. (2011). Violence and other stressful life events as triggers of
depression and anxiety: What psychosocial resources protect African American mothers?
Maternal and Child Health Journal, 15(8), 1272–1281. http://doi.org/10.1007/s10995-010-
0668-6
Mizell, C. A. (1999). Life course influences on African American men’s depression: Adolescent
parental composition, self-concept, and adult earnings. Journal of Black Studies, 29, 467–
490. http://doi.org/10.1177/002193479902900401
Moreira-Almeida, A., Lotufo Neto, F., & Koenig, H. G. (2006). Religiousness and mental health:
A review. Revista Brasileira de Psiquiatria, 28(919), 242–250.
http://doi.org/10.1590/S1516-44462006000300018
Noël, P. H., Williams, J.W, J., Unutzer, J., Worchel, J., Lee, S., Cornell, J., … Hunkeler, E.
(2004). Depression and comorbid illness in elderley primary care patients: Impact on
multiple domains of health status and well-being. Annals of Family Medicine, 2, 555–562.
http://doi.org/10.1370/afm.143.
Pickett, Y. R., Bazelais, K. N., & Bruce, M. L. (2013). Late-life depression in older African
Americans: A comprehensive review of epidemiological and clinical data. International
Journal of Geriatric Psychiatry, 28(6), 903–913. http://doi.org/10.1002/gps.3908
Richards, D. (2011). Prevalence and clinical course of depression: A review. Clinical Psychology
Review. http://doi.org/10.1016/j.cpr.2011.07.004
Rogers, R. G., Rosenblatt, R., Hummer, R. A., & Krueger, P. M. (2001). Black-White
differentials in adult homicide mortality in the United States. Social Science Quarterly, 82,
435–452. http://doi.org/10.1111/0038-4941.00034
Romero, L. J., Ortiz, I. E., Finley, M., Wayne, S., & Lindeman, R. D. (2005). Prevalence of
Hays
23

depressive symptoms in New Mexico Hispanic and non-Hispanic white elderly. Ethnicity &
Disease, 15(4), 691–697.
Salami, T. K., & Walker, R. L. (2014). Socioeconomic status and symptoms of depression and
anxiety in African American college students: The mediating role of hopelessness. Journal
of Black Psychology, 40(3), 275–290. http://doi.org/10.1177/0095798413486158
Santini, Z. I., Koyanagi, A., Tyrovolas, S., Mason, C., & Haro, J. M. (2015). The association
between social relationships and depression: A systematic review. Journal of Affective
Disorders, 175, 53–65. http://doi.org/10.1016/j.jad.2014.12.049
Spence, N. J., Adkins, D. E., & Dupre, M. E. (2011). Racial differences in depression trajectories
among older women: Socioeconomic, family, and health influences. Journal of Health and
Social Behavior, 52(4), 444–459. http://doi.org/10.1177/0022146511410432
Sutin, A. R., Terracciano, A., Milaneschi, Y., An, Y., Ferrucci, L., & Zonderman, A. B. (2013).
The trajectory of depressive symptoms across the adult life span. JAMA Psychiatry, 70(8),
803–811. http://doi.org/10.1001/jamapsychiatry.2013.193
Taylor, R. J., Chatters, L. M., & Abelson, J. M. (2012). Religious involvement and DSM-IV 12-
month and lifetime major depressive disorder among African Americans. Journal of
Nervous and Mental Disease, 200, 856–862.
http://doi.org/10.1097/NMD.0b013e31826b6d65
Teo, A. R., Choi, H. J., & Valenstein, M. (2013). Social relationships and depression: Ten-year
follow-up from a nationally representative study. PLOS One, 8(4), e62396.
http://doi.org/10.1371/journal.pone.0062396
Turner, R. J., & Avison, W. R. (2003). Status variations in stress exposure: Implications for the
interpretation of research on race, socioeconomic status, and gender. Journal of Health and
Hays
24

Social Behavior, 44(4), 488–505. http://doi.org/10.2307/1519795
U.S. Department of Health and HumanServices. (2011). 2010 National healthcare disparities
report. AHRQ Publication No.11-0005. Rockville, MD.
Whitley, R., & Lawson, W. (2010). The psychiatric rehabilitation of African Americans with
severe mental illness. Psychiatric Services, 61(5), 508–11.
http://doi.org/10.1176/appi.ps.61.5.508
Williams, D. R., González, H. M., Neighbors, H., Nesse, R., Abelson, J. M., Sweetman, J., &
Jackson, J. S. (2007). Prevalence and distribution of major depressive disorder in African
Americans, Caribbean blacks, and non-Hispanic whites: Results from the National Survey
of American Life. Archives of General Psychiatry, 64(3), 305–315.
http://doi.org/10.1001/archpsyc.64.3.305
Williams, D. R., & Jackson, J. S. (2000). Race/ethnicity and the 2000 census: Recommendations
for African American and other Black populations in the United States. American Journal
of Public Health, 90(11), 1728–1730. http://doi.org/10.2105/AJPH.90.11.1728
Willson, A. E., Shuey, K. M., & Elder, G. H. (2007). Cumulative advantage processes as
mechanisms of inequality in life course health. American Journal of Sociology, 112(6),
1886–1924. http://doi.org/10.1086/512712
Woods, V. D., King, N. J., Hanna, S. M., & Murray, C. (2012). We ain’t crazy! Just coping with
a crazy system”: Pathways into the Black population for eliminating mental health
disparities. African American Health Institute of San Bernardino County. San Bernardino,
CA.
Woodward, A. T., Taylor, R. J., & Chatters, L. M. (2011). Use of professional and informal
support by black men with mental disorders. Research on Social Work Practice, 21(3), 328–
Hays
25

336. http://doi.org/10.1177/1049731510388668
Xu, J., Murphy, S. L., Kochanek, K. D., & Bastian, B. A. (2016). Deaths: Final Data for 2013.
National Vital Statistics Reports, 64(2), 1–119.

Hays
26

Chapter Two (Study 1): A Risk Typology for Depression in African Americans
African Americans are likely to experience severe, chronic, and disabling depressive
symptoms leading to poor depression outcomes according to prior race comparison studies. The
excess burden of depression in African Americans is evidenced by the fact that they have more
depressive symptoms than Whites (Bromberger, Harlow, Avis, Kravitz, & Cordal, 2004;
Romero, Ortiz, Finley, Wayne, & Lindeman, 2005) and when they are diagnosed with
depression, it is often more debilitating and persistent compared to Whites (Williams et al.,
2007). Not only do African Americans suffer disproportionately from depression they are also
likely to die prematurely due to comorbid health conditions related to depression (Druss, Zhao,
Von Esenwein, Morrato, & Marcus, 2011; Parks, Svendsen, Singer, Foti, & Mauer, 2006).
However, very few studies have attempted to understand contributors to depression within the
African American population. Further, few interventions address mental illness among African
American adults in a way that is tailored to their cultural and religious experiences (Hankerson &
Weissman, 2012; Hays & Aranda, 2016).
Reducing the burden of depression and improving outcomes for African Americans must
be a public health priority. One potential strategy to address this problem is to target specific
contributors to poor depression outcomes using targeted interventions. Disease prevention
models suggest that efforts to improve health require a focus on reducing risk factors and
increasing protective factors that contribute to health outcomes (Ezzati et al., 2002). Thus,
understanding the collection of life conditions that constitute risk and protective factors for
depression in African Americans may inform interventions seeking to reduce the depression
burden in this population. Further, intervention efforts must be tailored to particular subgroups of
African Americans that are likely to suffer disproportionately. Examining heterogeneity among
Hays
27

African Americans and identifying demographic characteristics that are linked to worse
depression will inform tailored depression interventions.
The extant literature has identified several risk and protective factors associated with
depression in African Americans. Psychosocial and sociocultural factors such as socioeconomic
status (Salami & Walker, 2014), social relationships (Santini, Koyanagi, Tyrovolas, Mason, &
Haro, 2015), psychosocial stress (Keyes, Barnes, & Bates, 2011), and religious involvement
(Taylor, Chatters, & Abelson, 2012) have all been related to depressive symptoms in this group.
Although this body of research has significantly contributed to our understanding of the various
factors associated with depression in African Americans, there are at least two underexplored
issues that can increase our ability to explain and address poor depression outcomes. First,
existing studies rarely investigated how individual risk factors operate collectively to affect
depression outcomes. The human experience is complex and various risk and protective factors
(e.g., social support and negative interaction) often co-occur to jointly influence health outcomes
(K. D. Lincoln & Chae, 2012). Thus, the identification of unique constellations of multiple co-
occurring risk and protective factors may offer greater explanatory potential into depression
burden for African Americans than when those measures are examined independently. The
identification of unique constellations of risk and protective factors associated with depression
will be useful in the development of interventions that target the collection of life conditions
associated with worse depression in African Americans.
Second, existing studies typically treated African Americans as a monolithic group and
thus did not consider vast within-group variation when examining depression. This is
problematic because researchers have suggested that African Americans are a heterogeneous
population with great variation in help-seeking behavior (Hays & Gilreath, 2016; Woodward,
Hays
28

Taylor, & Chatters, 2011), level of distress (K. D. Lincoln, Taylor, Watkins, & Chatters, 2011),
and other social and demographic factors (Williams & Jackson, 2000). Examining within-group
differences, without racial and ethnic comparisons, will help elucidate subpopulations of African
Americans, which will be useful in tailoring intervention efforts to specific subgroups based on
characteristics like age, gender, and education level.
The current study sought to address the excess depression burden in African Americans
by adding to the literature regarding risk and protective factors. This exploratory analysis
focused on three research questions: (1) Are there specific risk profiles (i.e., types) characterized
by constellations of psychosocial and sociocultural factors? (2) Are these risk types associated
with depressive symptoms? (3) Can these risk types be predicted by a demographic profile?
Results of this analysis provide information that can advance the development of interventions
targeting the collection of risk and protective factors (i.e., risk types) associated with increased
depressive symptoms and tailored to the specific demographic characteristics of those most
likely to experience excess depression burden.
Background
Risk and Protective Factors for Depression
For the purpose of this study, risk and protective factors are considered to be life
conditions that increase or decrease the likelihood of mental health or illness (Substance Abuse
and Mental Health Services Administration, 2015). Risk and protective factors for depression
span a wide variety of issues, from physical health status (Beekman et al., 1997) to racial
discrimination (Cheng, Cohen, & Goodman, 2015) and stigmatized attitudes toward depression
(Menke & Flynn, 2009). However, this study focused exclusively on the role that particular
sociocultural and psychosocial factors play in influencing depression outcomes in African
Americans because of their potential for modification through intervention and importance as
Hays
29

documented in prior studies. Previous studies have identified several psychosocial and
sociocultural risk and protective factors that are associated with depression in African
Americans. Specifically, psychosocial stress, social relationships, and religious involvement
have been connected to depression in this population. A brief review of this literature is
necessary because it highlights current knowledge gaps that should be investigated to inform
future prevention and intervention efforts.
Social stress (e.g., financial stress) and negative life events (e.g., violence, trauma) have
been linked to increased depressive symptoms in African Americans (Mitchell & Ronzio, 2011).
African Americans often experience multiple adverse social conditions such as community
violence and poverty that contribute to the development of depressive symptoms. For example,
research by Cutrona et al. (2005) found that experiencing negative life events significantly
predicts the onset of major depression among African American women. Other studies have also
concluded that experiencing financial stress or events like a death in the family or being
assaulted is associated with increased depressive symptoms among African American adults
(Cutrona et al., 2005; Lantz, House, Mero, & Williams, 2005; Mitchell & Ronzio, 2011; Turner
& Avison, 2003).
Social relationships have been identified as both a risk and protective factor for
depression in African Americans. Specifically, low social support (Teo, Choi, & Valenstein,
2013), poor overall relationship quality (Shim et al., 2012), and negative social interactions (i.e.,
critical or demanding social relationships; Lincoln & Chae, 2012) are associated with increased
depression. Conversely, positive social relationships and larger social networks are protective
and associated with lower depressive symptoms (Chatters, Taylor, Woodward, & Nicklett, 2015;
K. D. Lincoln & Chae, 2012).
Hays
30

Religious involvement has also been linked to depression. Religion is generally
considered to be a protective factor associated with improved mental health and well-being
(Moreira-Almeida, Lotufo Neto, & Koenig, 2006). For example, one study found that African
Americans who reported receiving a “great deal of guidance” from religion were about half as
likely to have major depression as those with lower levels of religiosity (Ellison & Flannelly,
2009). Studies also suggested that African Americans who attend religious services more
frequently are less likely to have depression than infrequent church attendees (Chatters et al.,
2008).
Studies have also linked demographic characteristics, like socioeconomic status, age,
gender, and marital status, to depression (Chatters et al., 2015; K. D. Lincoln, Abdou, & Lloyd,
2014; K. D. Lincoln et al., 2011; Salami & Walker, 2014). For example, some results suggested
that African Americans with less than a high school education and those with low incomes are at
increased risk of depression (Hudson, Neighbors, Geronimus, & Jackson, 2012). However these
associations are inconclusive because other studies have reported a positive association, with
higher socioeconomic status being linked to more depressive symptoms (Salami & Walker,
2014).
This body of research provided insight into the various psychosocial and sociocultural
risk and protective factors associated with depression among African Americans. However, these
studies did not examine patterns of co-occurring risk factors that are associated with depression.
This is a significant limitation in light of what is known about the co-occurrence of risk and
protective factors. For example, research has suggested that social support involves both
supportive relationships and negative social interactions that each represent specific kinds of
social experiences that can co-occur and collectively affect health and well-being (K. D. Lincoln,
Hays
31

Taylor, & Chatters, 2013). Currently, no studies have demonstrated how various psychosocial
and sociocultural factors combine to form a typology of co-occurring risk factors associated with
depression in African Americans.
Risk Types
Researchers have suggested that the co-occurrence of psychosocial risk factors is
ubiquitous (Evans, 2004; Kessler, Davis, & Kendler, 1997; Rutter, 2000) and health risks should
be studied simultaneously rather than focusing on the effect of individual risk factors (Chen,
Eaton, Gallo, & Nestadt, 2000; Laska, Pasch, Lust, Story, & Ehlinger, 2009). However, few
empirical investigations have focused on the co-occurrence of psychosocial and sociocultural
risk factors for depression or other mental problems. Researchers have made the case that some
combinations of risk factors are more predictive of psychopathology than others and that prior
research that focused on the additive contributions of individual risk factors, or interactions
among two or three risk factors, did not adequately capture the complexity of what individuals
face in the real world (Copeland, Shanahan, Costello, & Angold, 2009).
For example, in a study of children aged 9–13, Copeland and associates (2009) identified
specific configurations of psychosocial risk (i.e., socioeconomic disadvantage, nonnuclear family
structure, parental risk characteristics, family dysfunction, and stressful life events) that were
associated with emotional and disruptive disorders. They concluded that investigating co-
occurring risk factors helps pinpoint the collection of life conditions that is predictive of mental
health outcomes (Copeland et al., 2009). Other studies focusing on children have found similar
results, suggesting identifiable profiles of risk factors are associated with mental and emotional
problems (Mendelson, Turner, & Tandon, 2010; Walrath et al., 2004). However, no existing
studies attempted to identify a typology of risk and protection for African American adults
Hays
32

regarding depression. This is a critical gap in the literature given the problem of excess
depression burden and the role psychosocial and sociocultural factors play in contributing to
depression among African Americans. This study addressed this gap by explicating the various
risk types that might be connected to worse depression for African Americans.
This work contributes to the literature on depression in African Americans in very
specific ways. Examining the existence of unique combinations of risk and protective factors can
inform the targeting and tailoring of interventions. For example, if results reveal a risk type
characterized by high stress, low support, and low religious involvement that is associated with
more depressive symptoms, individuals who fit this risk type should be the target of
interventions seeking to improve depression in African Americans. Further, this study
investigated heterogeneity among African Americans by suggesting subgroups based on
demographic characteristics. This is necessary for tailoring interventions to individuals most
likely to suffer. For example, results might suggest that single men with low education are likely
to be in the risk type related to poor depression. The results of this study could lead to tailored
interventions that meet the needs of particular subpopulations.
Methods
Data
The Americans’ Changing Lives study (ACL) is an ongoing nationally representative
study focused on differences between Black and White Americans in middle and late life. The
study features a range of sociological, psychological, mental, and physical health items. Wave 1
of the study began in 1986 with face-to-face interviews of 3,617 adults aged 25 or older. African
Americans and people older than 60 were oversampled. The analytic sample for this study
features 1,174 respondents who identified as African American in Wave 1. Mplus software
Hays
33

(version 7) uses full-information maximum likelihood estimation based on the assumption that
data are missing at random (Arbuckle, 1996; Little & Schenker, 1995).
Variables in the Latent Classes
Psychosocial stress. Psychosocial stress includes the experience of chronic financial
stress and negative life events. Chronic financial stress was measured using three items from the
Americans’ Changing Lives study that assessed each respondent’s financial status. The items
asked respondents to indicate the following: (a) “How satisfied are you with your/your family’s
present financial situation?” Responses ranged from 1 (completely satisfied) to 5 (not at all
satisfied), (b) “How difficult is it for you/your family to meet the monthly payments on your
(family’s) bills?” Response options ranged from 1 (extremely difficult) to 5 (not difficult at all),
(c) “In general, how do your (family’s) finances usually work out at the end of the month—do
you find that you usually end up with …”; responses categories were 1 (some money left over), 2
(just enough money to make ends meet), and 3 (not enough money to make ends meet).
Responses to these items were reverse coded so that lower values indicated less stress. These
three items were indexed into a single variable representing chronic financial stress by obtaining
the mean score across the three items then recoding the mean score into a single variable. For the
indexed variable response options included 1 (low financial stress) which represented original
responses of “not difficult”, “slightly difficult”, and “some money left over”; 2 (moderate
financial stress) representing “some difficulty” and “just enough money to make ends meet”; and
3 (high financial stress) representing “very difficult”, “extremely difficult”, and “not enough
money to makes ends meet”.
Negative life events were assessed by asking respondents to indicate whether or not they
had experienced any trauma or loss including the death of a spouse, child, parent or stepparent;
Hays
34

divorce; job loss; assault; robbery; or “anything else bad” during the 3 years preceding the Wave
1 interview. Because this analysis focused on the overall experience of negative life events and
not any one event, a count variable was created with the number of events experienced ranging
from 0 (lowest) to 6 (highest). Next, to increase result interpretability, a categorical variable was
created so that 0 or 1 events = low, 2 or 3 events = moderate, and 4 or more events = high.
Religious involvement. Religious involvement was assessed using three dimensions.
Organizational religious involvement was measured by asking respondents, “How often do you
usually attend religious services?” Response categories ranged from 1 (never) to 6 (more than
once a week). Response items were reverse coded and recoded so that 1 (low involvement)
represented “never” and “less than once a month”; 2 (moderate involvement) represented “once a
month” and “2-3 times a month”; and 3 (high involvement) represented attending church “once a
week” or “more than once a week”. Non-organizational religious involvement was measured
with items that asked respondents to indicate the frequency with which they read religious books
and watch or listen to religious programs on the television or radio. The original response
categories ranged from 1 (never) to 6 (more than once a week). Response to this item were also
recoded as 1 (low involvement), 2 (moderate involvement), and 3 (high involvement). Subjective
religiosity was measured by asking respondents, “In general, how important are religious or
spiritual beliefs in your day to day life?” The original response categories ranged from 1 (not at
all important) to 4 (very important). For ease of interpretation for the analysis responses were
reverse coded and recoded so that 1 (low religiosity) represented “not at all important” and “not
too important”; 2 (moderate religiosity) represented “fairly important”; and 3 (high religiosity)
represented “very important”.
Hays
35

Social support. Social support was measured by asking respondents to evaluate their
relationships with friends and relatives, spouse, children, mother, and father. Social support from
each source was assessed with two questions: (a) “On the whole, how much do your […] make
you feel loved and cared for?” and (b) “How much are these […] willing to listen when you need
to talk about your worries or problems?” Response categories ranged from 1 (a great deal) to 5
(not at all). These 10 items were reverse coded and indexed into one social support variable by
calculating the mean score across the 10 items. Next the distribution of the mean scores was
consulted to identify the cut off points that most closely reflected the distribution of responses
for the original variables. Mean scores of 2.33 and 2.83 were used to create the response
categories for the indexed variable which included 1 = low, 2 = moderate, and 3 = high.
Negative interaction. Negative interaction with friends and relatives, spouse, children
mother, and father were each measured by two items: (a) “How much do you feel your […]
make too many demands on you?” and (b) “How much are they critical of you or what you do?”
Response categories ranged from 1 (a great deal) to 5 (not at all). These 10 items were reverse
coded and indexed into one variable representing negative interaction using the same process as
the social support variable. Response cut off points of 3 and 3.8 were used to create response
categories of 1 = low, 2 = moderate, and 3 = high negative interaction.
Covariates
Depression. The outcome of interest in this study is depression. Depressive symptoms
were measured with 11 items from the Center for Epidemiological Studies Depression Scale
(Radloff, 1977). Items measured the extent to which respondents felt happy, lonely, sad, that
everything was an effort, that their sleep was restless, that people were unfriendly, that they did
not feel like eating, that people dislike them, that they could not get going, and that they enjoyed
Hays
36

life. This abbreviated scale has been found to have acceptable reliability and a similar factor
structure compared to the original 20-item version (Foley, Reed, Mutran, & DeVellis, 2002;
Radloff, 1977). Responses were on a Likert-type scale ranging from 1 (hardly ever) to 3 (most of
the time). Positively worded items (i.e., felt happy and enjoyed life) were reverse coded. Scores
from the 11 items were averaged so that higher mean scores indicated more depressive
symptoms.
Demographic characteristics. Demographic variables included in this analysis were
age, gender, marital status, and education level. Age was recoded from a continuous to a
categorical variable (25–39 years old, 40–59 years old, and 60 years old or older). Marital status
was categorized as married, separated, divorced, widowed, and never married. Education level
was recoded as a single variable with categories of below high school, high school, some college,
and college degree. Age, marital status, and education level were all dummy coded for logistic
regression analysis.
Analysis
The goal of this study was to empirically identify a risk typology for African American
adults that describes life conditions, or risk types, characterized by unique constellations of
psychosocial factors (i.e., chronic financial stress, negative life events, social support, and
negative interactions) and a sociocultural factor (i.e., religious involvement). Latent class
analysis (LCA) was conducted using Mplus 7 to identify distinct risk types in the sample
(McCutcheon, 1987; Muthén & Muthén, 2011). LCA was selected because it is designed to
distinguish homogeneous subgroups within a heterogeneous population (Lanza, Collins,
Lemmon, & Schafer, 2007; Magidson & Vermunt, 2002), thus making it an appropriate analytic
method for identifying classes of individuals based on risk profiles.
Hays
37

A series of models was estimated to determine the appropriate number of classes (or
profiles), which represent distinct risk types. The three-step method for conducting LCA
simultaneously with multiple logistic regression was employed (Nylund-Gibson, Grimm, Quirk,
& Furlong, 2014). First, a one-class (no covariates) model was estimated, followed by a series of
models with demographic covariates specifying an increased number of classes (e.g., two
classes, three classes). The optimal model was selected based on goodness-of-fit indexes,
including the Bayesian information criterion (BIC), entropy, and the parametric bootstrapped
likelihood ratio test (BLRT; Nylund, Asparouhov, & Muthén, 2007). Once the ideal number of
latent classes was identified based on the fit indexes, they were analyzed for substantive
meaning. Each class received an appropriate descriptive title. Finally, the results of the multiple
logistic regression were analyzed to identify significant associations between risk types and
demographic characteristics (e.g., age and gender).
To assess for differences in depression for each risk type distal outcome LCA was
conducted (Asparouhov & Muthén, 2013). In the distal outcome model, mean estimates of
CESD were obtained for each risk type to determine whether CESD scores differed between
each of the risk types. In this step of the analysis, age, gender, education, and marital status were
controlled for and were mean centered. Statistical analyses in the initial LCA and distal outcome
estimation steps accounted for the complex multistage clustered design of the ACL sample by
including sample weights and stratum.
Results
The characteristics of the overall sample (N = 1,174) are summarized in Table 1. Females
comprised the majority of the sample (66%). Respondents aged 60 or older represented a large
proportion of the sample (43%). Overall, the respondents had low levels of educational
attainment, with more than half (51%) having less than a high school education. Concerning
Hays
38

marital status, many of the respondents were either married (39%) or widowed (21%). Most
respondents had high levels of religiosity with nearly half of respondents having high
organizational and non-organizational religion (46% and 49%) and 80% of respondents
endorsing high subjective religiosity. Further, most respondents had high social support (74%).
On the other hand, there were very few respondents who endorsed having high levels of
traditional risk factors including negative life events (0.09%), financial stress (21%), and
negative social interaction (11%).
For the LCA, a three-class model was shown to have the best overall fit to the data (see
Table 2). This is exemplified by the significant BLRT p-value, increase in entropy, and decrease
in adjusted BIC compared to the two-class model. In this sample, 41% was most likely to be in
the High Protective/Low Risk Type, which is characterized by high religious involvement,
positive social support, and low stress. Specifically, these respondents had high levels of
religiosity across all three domains (i.e., organizational, non-organizational, and subjective), low
experience of negative life events, low financial stress, high social support, and low negative
social interaction. Individuals in the second class, titled Moderate Protective/Low Risk, had
moderate levels of religiosity, low experience of negative life events, low financial stress,
moderate social support, and low negative social interaction. Those in the Moderate
Protective/Low Risk type comprised 44% of the total sample. The third class, Low
Protective/Low Risk, was the smallest class (15%).The Low Protective/Low Risk type was
characterized by low organizational and non-organizational religiosity, fairly low financial stress,
and low social support (see Table 3 and Figure 1).
Table 4 presents the multinomial logistic regression results, which indicate that gender,
age, and education level were statistically significant predictors of class membership. Compared
Hays
39

to women, men were more likely to be in the Moderate Protective/Low Risk (OR = 2.08, CI =
1.15, 3.74) and the Low Protective/Low Risk (OR = 2.94, CI = 1.58, 5.49) types versus the
referent (High Protective/Low Risk). Also, respondents aged 25-39 had more than twice the odds
of being in the Moderate Protective/Low Risk class versus the High Protective/Low Risk type
(OR = 2.23, CI = 1.13, 4.39) compared to respondents aged 60 or older. Concerning education,
having less than a high school education (OR = 9.61, CI = 1.40, 66.00) or a high school diploma
(OR = 10.28, CI = 1.43, 73.69) was associated with nearly 10 times the odds of being in the
Moderate Protective/Low Risk type versus the High Protective/Low Risk type compared to those
with a college degree. There were also significant differences in depression score for the risk
types. Specifically, those in the High Protective/Low Risk class had significantly lower mean
depression scores that those in the Moderate Protective/Low Risk class (1.45 vs. 1.55; p=.005).
Also, those in the High Protective/Low Risk class had significantly lower mean depression
scores that those in the Low Protective/Low Risk class (1.45 vs. 1.55; p=.05). There was no
statistically significant difference in depressive symptoms between the Moderate Protective and
Low Protective groups.
Discussion
This study used data from the American’s Changing Lives survey to identify a risk
typology for depression for African American adults and older adults. First, this study
investigated the existence of unique constellations of psychosocial and sociocultural risk and
protective factors. Results suggest that three distinct risk profiles (High Protective/Low Risk,
Moderate Protective/Low Risk, and Low Protective/Low Risk) characterize unique combinations
of sociocultural and psychosocial risk. These emergent classes are consistent with other studies
that suggested that specific combinations of co-occurring risk factors can be identified using
person-centered methods like LCA. Findings indicate that two dimensions of religiosity (i.e.,
Hays
40

organizational and non-organizational), social support, and stress play key roles in distinguishing
individuals based on their risk type.
Second, this study investigated the association between a risk typology and depressive
symptoms. There were significant differences in depressive symptoms for the risk types
identified in this population of African Americans. Last, associations between several
demographic variables were analyzed to describe the individual characteristics of participants in
each risk type. The multinomial logistic regression results confirmed that gender, age, and
education status are predictors of risk type. These findings are consistent with prior literature that
suggested that African Americans are a heterogeneous population with great variation in
characteristics related to depression.
High Protective/Low Risk
Those in the High Protective/Low Risk type were characterized as having high religious
involvement, high social support, low stress, and low negative interaction. More than 40% of the
sample was classified in this type and individuals in this group were more likely to have low
depressive symptoms than those in the other risk types. This study’s findings suggest an additive
effect of certain risk and protective factors for African Americans. To understand why high
religiosity, high social support, and low stress might function together to create a High
Protective/Low Risk profile, existing studies can offer clues. For example, research has
suggested that individuals who attend church more often benefit from larger social networks and
social support from fellow congregants (Chatters et al., 2015; Ellison & George, 1994; Holt &
McClure, 2006). Further, organizational religious involvement might increase one’s ability to
receive instrumental support. Many churches, especially historically Black churches, are known
to be sources of instrumental support for African American communities. It may be the case that
Hays
41

individuals in the High Protective/Low Risk Type have more social connections through church
that provide them with financial assistance, employment resources, and other instrumental
supports that lower their psychosocial stress.
It is also important to note the demographic profile of the individuals in the High
Protective/Low Risk type. Women, those older than 60, and those with a college education were
most likely to be in the High Protective/Low Risk type, described as highly religious, with strong
social connections, and low stress. These results are consistent with previous studies that suggest
that women tend to have stronger social networks and be more religious than men (Barker,
Morrows, & Mitteness, 1998; Levin, Taylor, & Chatters, 1994; Taylor, Mattis, & Chatters,
1999). Although research has suggested that older African Americans are more religious than
their younger counterparts (Taylor, Chatters, & Joe, 2011) some studies indicated that older age
is associated with smaller social networks and lower social support (Schwarzbach, Luppa,
Forstmeier, König, & Riedel-Heller, 2014). However, the results of the current study go beyond
prior research to suggest that older African Americans are more likely to experience a
combination of high religiosity and high social support than respondents aged 25–39. Further, a
college education increased the likelihood of African Americans being in the High
Protective/Low Risk type. This suggests that educational attainment is related to depressive
symptoms because of its associated risk and protective factors. Higher education may increase an
individual’s connection supportive networks through school and work relationships while
providing increased resources to manage stressors.
Moderate Protective/Low Risk
A large proportion of the sample (44%) belonged to the Moderate Protective/Low Risk
type. This group included individuals with moderate organizational and non-organizational
Hays
42

religiosity, low experience of negative life events, low financial stress, moderate social support,
and low negative social interaction. Individuals in this class were significantly more likely to
have higher depression scores than those in the High Protective/Low Risk type. What
differentiates the Moderate Protective/Low Risk type from the High Protective/Low Risk type
primarily centers on religious involvement and social support. Although subjective religiosity
was high across all classes, organizational and non-organizational religiosity were noticeably
lower among those in the Moderate Protective/Low Risk compared to the High Protective/Low
Risk type. Similarly, the experience of positive social support was lower among those in this
class. Again, this exemplifies the interconnected nature of religiosity and social support, because
lower religiosity corresponded to lower positive social support in this group.
These results are consistent with the literature, which suggests that social support and
religiosity protect against depression. However, a unique finding of this study is that those in the
Moderate Protective/Low Risk type also experienced slightly higher levels of negative social
interaction than those in the High Protective/Low Risk type. This suggests that protective factors
(i.e., social support and religiosity) are related to increased depression in conjunction with risk
factors (i.e., negative interaction). Identifying this constellation of risk and protection indicates
that efforts to improve depression in African Americans must support increased protective
factors but also reduce risk factors linked to worse depression.
Individuals in the Moderate Protective/Low Risk type also fit a unique demographic
profile. Results show that men, those aged 25–39, and individuals with a high school education
or less were significantly more likely to be in the Moderate Protective/Low Risk type than the
High Protective/Low Risk type. Those in the class also had higher mean depressive symptom
scores than those in the High Protective/Low Risk type. This may be because men, younger
Hays
43

adults, and those with lower educational attainment may have fewer supports and social
connections to help them manage daily life.
Low Protective/Low Risk
Although the Low Protective/Low Risk type only constituted 15% of the total sample,
individuals in this group had the highest proportion of low religiosity and social support. The
specific constellation of risk and protective factors that were prevalent in this class are consistent
with prior literature. Research has shown that risk factors like stress and negative social
interaction are individually related to increased depression (K. Lincoln & Chae, 2011; Mitchell
& Ronzio, 2011). However, this study suggests that the combination of higher risk factors and
lower protective factors like religion and social support is especially problematic, because the
individuals in the Low Protective/Low Risk type have significantly higher depressive symptoms
than those in the High Protective/Low Risk type. However, there was no significant difference in
depression score between those in this class and the Moderate Protective/Low Risk class.
Regarding demographics, only gender was predictive of being in the Low Protective/Low
Risk type; men were nearly 3 times more likely than women to be in this class compared to the
High Protective/Low Risk type. This finding aligns with prior research that suggested that
African American men have fewer social connections and are less religious than women (Barker
et al., 1998; Levin et al., 1994; Taylor et al., 1999). More importantly however, these results
provide evidence of the need for increased focus on depression in African American men.
Although many prior studies suggested that women have more depressive symptoms than men
(Bracke, 2000; Essau, Lewinsohn, Seeley, & Sasagawa, 2010), these results suggests that
African American men are at increased risk of more depressive symptoms because of their most
likely risk profile. Other researchers have suggested that African American men are often
Hays
44

underdiagnosed and face depression treatment disparities (Mizell, 1999; Watkins, 2012). Based
on results of this study, more focus on depression, and its risks, in African American men is
warranted.
Another noteworthy finding is that risk factors (negative social interaction, financial
stress, and negative life events) were low overall across all risk types. This is especially
important considering that African Americans are often cited as facing adverse social conditions
that increase their exposure to stress and negative events like neighborhood violence and crime.
However, this study revealed that the experience of these events was low among African
Americans overall. This might be reflective of the nature of sample selection in that respondents
represent a selective of group of African Americans that excludes individuals who are
incarcerated or otherwise institutionalized or may not have the mental or emotional capacity to
participate in the interview. Another explanation may be that within group analysis is able to
capture the experience of negative life events of African Americans without comparing them to
other race groups who may have an even lower experience of these events. Race comparisons
might conflate the prevalence of high negative life events cited in prior research. It may be the
case that for many African Americans, the experience of protective factors like religiosity and
social support, or lack thereof, is more indicative of risk of depression than risk factors like
negative social interaction and stress. In addition, it should be noted that subjective religiosity
was very high across all classes. This is not surprising given research that showed African
Americans are the most religious racial group in the nation (Pew Research Center, 2015). This
does suggest, however, that when exploring the relationship between religiosity and depression,
assessing organizational and non-organizational dimensions might be more robust for African
American samples.
Hays
45

Table 2.1 Overall Demographic and Risk Characteristics
% n
Sex
Male 33.73 396
Female 66.27 778
Age
25–39 31.52 370
40–59 25.81 303
60+ 42.50 499
Marital status
Married 39.18 460
Separated 9.97 117
Divorced 13.03 153
Widowed 21.38 251
Never married 16.44 193
Education
Below high school 51.36 603
High school 24.11 283
Some college 16.87 198
College degree 7.67 90
Non-organizational religiosity
Low 14.35 168
Moderate 36.12 423
High 49.53 580
Organizational religiosity
Low 25.17 295
Moderate 28.07 329
High 46.76 548
Subjective Religiosity
Hays
46

Low 4.17 49
Moderate 15.76 185
High 80.07 940
Negative life events
Low (0 or 1 events) 90.03 1057
Moderate (2 or 3 events) 9.88 116
High (4 or more events) .09 1
Chronic financial stress
Low 41.31 485
Moderate 37.22 437
High 21.12 248
Social support
Low 11.84 139
Moderate 14.14 166
High 74.02 869
Negative social interaction
Low 71.64 841
Moderate 17.12 201
High 11.24 132

Hays
47

Table 2.2. Fit Statistic Comparisons of LCA Models of Risk Types among African Americans

Adjusted BIC BLRT p Entropy
One class (no covariates) 12,181.318 < .001
Two classes 11,898.467 < .001 .702
Three classes 11,889.297 < .001 .715
Four classes 11,941.256 < .001 .649
Five classes 12,000.245 < .001 .611
Hays
48

Table 2.3. Conditional Probabilities of Risk Types for Depression
High Protective/Low
Risk
Moderate
Protective/Low Risk
Low Protective/Low
Risk
(n = 481, 41%) (n = 516, 44%) (n = 176, 15%)
Non-organizational religiosity
Low .007 .099 .650
Moderate < .001 .744 .227
High .993 .158 .124
Organizational religiosity
Low .078 .179 .940
Moderate .168 .461 .060
High .755 .360 < .001
Subjective religiosity
Low < .001 < .001 < .001
Moderate < .001 .010 .249
High 1.000 .990 .751
Negative life events
Low .897 .908 .887
Moderate .101 .092 .113
High .002 < .001 < .001
Chronic financial stress
Low .397 .425 .431
Moderate .398 .389 .262
High .205 .186 .307
Social support
Low .093 .078 .306
Moderate .104 .178 .139
High .804 .744 .555
Negative social interaction
Low .772 .688 .648
Moderate .135 .200 .186
High .094 .111 .166
Hays
49

Figure 2.1. Probabilities for Risk and Protective Factors for each Risk Type
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Probability of Endorsement
High Protective/Low Risk
Moderate Protective/Low Risk
Low Protective/Low Risk
Hays
50

Table 2.4. Multinomial Logistic Regression Results
Moderate Protective/Low Risk
vs. High Protective/Low Risk
Low Protective/Low Risk vs.
High Protective/ Low Risk
OR 95% CI OR 95% CI
Male
a
2.08* 1.15, 3.74 2.94* 1.58, 5.49
25–39 years old
b
2.23* 1.13, 4.39 1.98 0.88, 4.43
40–59 years old
b
0.96 0.53, 1.73 1.75 0.86, 3.59
Below high school
c
9.61* 1.40, 66.00 0.86 0.36, 2.08
High school
c
10.28* 1.43, 73.69 0.81 0.33, 1.99
Some college
c
5.42 0.72, 40.57 1.10 0.43, 2.81
Married
d
0.70 0.28, 1.79 1.29 0.56, 2.97
Separated
d
1.09 0.38, 3.10 0.95 0.36, 2.54
Widowed
d
1.38 0.53, 3.59 1.10 0.34, 3.55
a
Reference is female.
b
Reference is 60 years old or older.
c
Reference is college degree.
d
Reference is never married.
e
Mean score on Center for Epidemiological Studies Depression Scale.

Hays
51

Table 2.5. Mean CESD Score by Risk Type
High Protection/Low
Risk vs. Moderate
Protection Low Risk
High Protection/Low
Risk vs. Low
Protection/Low Risk
Moderate
Protection/Low Risk
vs. Low
Protection/Low Risk
CESD Mean 1.45 vs. 1.55** 1.45 vs. 1.55* 1.55 vs. 1.55
Note: ** p< .005
* p< .05

Hays
52

References
Asparouhov, T., & Muthen, B. (2013). Auxiliary Variables in Mixture Modeling : 3-Step
Approaches Using Mplus. Mplus Web Notes: No. 15, (15), 1–48.
Barker, J. C., Morrows, J., & Mitteness, L. S. (1998). Gender, informal social support networks,
and elderly urban African Americans. Journal of Aging Studies, 12, 199–222.
http://doi.org/http://dx.doi.org/10.1016/S0890-4065(98)90015-9
Beekman, A. T. F., Penninx, B. W. J. H., Deeg, D. J. H., Ormel, J., Braam, A. W., & van
Tilburg, W. (1997). Depression and physical health in later life: results from the
Longitudinal Aging Study Amsterdam (LASA). Journal of Affective Disorders, 46, 219–
231. http://doi.org/10.1016/S0165-0327(97)00145-6
Bracke, P. (2000). The three-year persistence of depressive symptoms in men and women. Social
Science & Medicine, 51–64. http://doi.org/10.1016/S0277-9536(99)00432-3
Bromberger, J. T., Harlow, S., Avis, N., Kravitz, H. M., & Cordal, A. (2004). Racial/ethnic
differences in the prevalence of depressive symptoms among middle-aged women: The
Study of Women’s Health Across the Nation (SWAN). American Journal of Public Health,
94(8), 1378–1385. http://doi.org/10.2105/AJPH.94.8.1378
Chatters, L. M., Bullard, K. M., Taylor, R. J., Woodward, A. T., Neighbors, H. W., & Jackson, J.
S. (2008). Religious participation and DSM IV disorders among older African Americans:
Findings from the National Survey of American Life. American Journal of Geriatric
Psychiatry, 16(12), 957–965. http://doi.org/10.1097/JGP.0b013e3181898081.
Chatters, L. M., Taylor, R. J., Woodward, A. T., & Nicklett, E. J. (2015). Social support from
church and family members and depressive symptoms among older African Americans.
American Journal of Geriatric Psychiatry, 23(6), 559–567.
Hays
53

http://doi.org/10.1016/j.jagp.2014.04.008
Chen, L. S., Eaton, W. W., Gallo, J. J., & Nestadt, G. (2000). Understanding the heterogeneity of
depression through the triad of symptoms, course and risk factors: A longitudinal,
population-based study. Journal of Affective Disorders, 59(1), 1–11.
http://doi.org/10.1016/S0165-0327(99)00132-9
Cheng, E., Cohen, A., & Goodman, E. (2015). The role of perceived discrimination during
childhood and adolescence in understanding racial and socioeconomic influences on
depression in young adulthood. The Journal of Pediatrics, 166(2), 370–377.e1.
http://doi.org/10.1016/j.jpeds.2014.10.010.
Copeland, W., Shanahan, L., Costello, J. E., & Angold, A. (2009). Configurations of Common
Childhood Psychosocial Risk Factors. Journal of Child Psychology and Psychiatry, 50(4),
451–459. http://doi.org/10.1111/j.1469-7610.2008.02005.x
Cutrona, C. E., Russell, D. W., Brown, P. A., Clark, L. A., Hessling, R. M., & Gardner, K. A.
(2005). Neighborhood context, personality, and stressful life events as predictors of
depression among African American women. Journal of Abnormal Psychology, 114(1), 3–
15. http://doi.org/10.1037/0021-843X.114.1.3
Druss, B. G., Zhao, L., Von Esenwein, S., Morrato, E. H., & Marcus, S. C. (2011).
Understanding excess mortality in persons with mental illness: 17-year follow up of a
nationally representative US survey. Medical Care, 49(6), 599–604.
http://doi.org/10.1097/MLR.0b013e31820bf86e
Ellison, C. G., & Flannelly, K. (2009). Religious involvement and risk of major depression in a
prospective nationwide study of African American adults. Journal of Nervous and Mental
Disease, 197(8), 568–573. http://doi.org/10.1097/NMD.0b013e3181b08f45
Hays
54

Ellison, C. G., & George, L. K. (1994). Religious involvement, social ties, and social support in a
Southeastern community. Journal for the Scientific Study of Religion, 33, 46–61.
Essau, C. A., Lewinsohn, P. M., Seeley, J. R., & Sasagawa, S. (2010). Gender differences in the
developmental course of depression. Journal of Affective Disorders, 127, 185–190.
http://doi.org/10.1016/j.jad.2010.05.016
Evans, G. W. (2004). The environment of childhood poverty. American Psychologist, 59, 77–92.
Ezzati, M., Lopez, A. D., Rodgers, A., Vander Hoorn, S., Murray, C. J. L., & The Comparative
Risk Assessment CollaboratingGroup. (2002). Selected major risk factors and global and
regional burden of disease. Lancet, 360, 1347–1360. http://doi.org/10.1016/S0140-
6736(02)11403-6
Foley, K. L., Reed, P., Mutran, E., & DeVellis, R. (2002). Measurement adequacy of the CES-D
among a sample of older African–Americans. Psychiatry Research, 109(1), 61–69.
Retrieved from http://www.sciencedirect.com/science/article/pii/S0165178101003602
Hankerson, S. H., & Weissman, M. M. (2012). Church-based health programs for mental
disorders among African Americans: A review. Psychiatric Services, 63(3), 243–249.
http://doi.org/10.1176/appi.ps.201100216
Hays, K., & Aranda, M. P. (2016). Faith-based mental health interventions with African
Americans: A review. Research on Social Work Practice, 26, 777–789.
http://doi.org/10.1177/1049731515569356
Hays, K., & Gilreath, T. (2016). Profiles of depression help seeking among Black Americans: A
latent class approach. Journal of Nervous and Mental Disease, Advance on.
http://doi.org/10.1097/NMD.0000000000000575
Holt, C. L., & McClure, S. M. (2006). Perceptions of the religion-health connection among
Hays
55

African American church members. Qualitative Health Research, 16, 268–281.
http://doi.org/10.1177/1049732305275634
Hudson, D. L., Neighbors, H. W., Geronimus, A. T., & Jackson, J. S. (2012). The relationship
between socioeconomic position and depression among a US nationally representative
sample of African Americans. Social Psychiatry and Psychiatric Epidemiology, 47, 373–
381. http://doi.org/10.1007/s00127-011-0348-x
Kessler, R. C., Davis, C. G., & Kendler, K. S. (1997). Childhood adversity and adult psychiatric
disorder in the US National Comorbidity Survey. Psychological Medicine, 27, 1101–1119.
http://doi.org/10.1017/S0033291797005588
Keyes, K. M. K., Barnes, D. ., & Bates, L. M. (2011). Stress, coping, and depression: Testing a
new hypothesis in a prospectively studied general population sample of US-born Whites
and Blacks. Social Science & Medicine, 72(5), 650–659.
http://doi.org/10.1016/j.socscimed.2010.12.005
Lantz, P. M., House, J. S., Mero, R. P., & Williams, D. R. (2005). Stress, life events, and
socioeconomic disparities in health: Results from the Americans’ Changing Lives study.
Journal of Health and Social Behavior, 46(3), 274–288.
http://doi.org/10.1177/002214650504600305
Lanza, S., Collins, L., Lemmon, D. R., & Schafer, J. (2007). PROC LCA: A SAS procedure for
latent class analysis. Structural Equation Modeling, 14, 671–694.
http://doi.org/10.1080/10705510701575602
Laska, M. N., Pasch, K. E., Lust, K., Story, M., & Ehlinger, E. (2009). Latent class analysis of
lifestyle characteristics and health risk behaviors among college youth. Prevention Science,
10(4), 376–386. http://doi.org/10.1007/s11121-009-0140-2
Hays
56

Levin, J. S., Taylor, R. J., & Chatters, L. M. (1994). Race and gender differences in religiosity
among older adults: Findings from four national surveys. Journal of Gerontology, 49(3),
S137–S145. http://doi.org/10.1093/geronj/49.3.S137
Lincoln, K., & Chae, D. H. (2011). Social support, negative interaction, and lifetime major
depressive disorder among African Americans and Caribbean Blacks. Social Psychiatry and
Psychiatric Epidemiology, 47(3), 361–372.
Lincoln, K. D., Abdou, C. M., & Lloyd, D. (2014). Race and socioeconomic differences in
obesity and depression among Black and non-Hispanic White Americans. Journal of Health
Care for the Poor and Underserved, 25(1), 257–275. http://doi.org/10.1353/hpu.2014.0038
Lincoln, K. D., & Chae, D. H. (2012). Emotional support, negative interaction and major
depressive disorder among African Americans and Caribbean Blacks: Findings from the
National Survey of American Life. Social Psychiatry and Psychiatric Epidemiology, 47(3),
361–372. http://doi.org/10.1007/s00127-011-0347-y
Lincoln, K. D., Taylor, R. J., & Chatters, L. M. (2013). Correlates of Emotional Support and
Negative Interaction Among African Americans and Caribbean Blacks. Journal of Family
Issues, 34(9), 1262–1290. http://doi.org/10.1177/0192513X12454655
Lincoln, K. D., Taylor, R. J., Watkins, D. C., & Chatters, L. M. (2011). Correlates of
psychological distress and major depressive disorder among African American men.
Research on Social Work Practice, 21(3), 278–288.
http://doi.org/10.1177/1049731510386122
Magidson, J., & Vermunt, J. K. (2002). Latent class models for clustering: A comparison with K-
means. Canadian Journal of Marketing Research, 20(1), 37–44.
McCutcheon, A. L. (1987). Latent class analysis. Newbery Park, CA: Sage.
Hays
57

Mendelson, T., Turner, A. K., & Tandon, S. D. (2010). Violence exposure and depressive
symptoms among adolescents and young adults disconnected from school and work.
Journal of Community Psychology, 38(5), 607–621. http://doi.org/10.1002/jcop.20384
Menke, R., & Flynn, H. (2009). Relationships between stigma, depression, and treatment in
white and African American primary care patients. Journal of Nervous and Mental Disease,
197(6), 407–411. http://doi.org/10.1097/NMD.0b013e3181a6162e
Mitchell, S. J., & Ronzio, C. R. (2011). Violence and other stressful life events as triggers of
depression and anxiety: What psychosocial resources protect African American mothers?
Maternal and Child Health Journal, 15(8), 1272–1281. http://doi.org/10.1007/s10995-010-
0668-6
Mizell, C. (1999). Life course influences on African American men’s depression: Adolescent
parental composition, self-concept, and adult earnings. Journal of Black Studies, 29, 467–
490. http://doi.org/10.1177/002193479902900401
Moreira-Almeida, A., Lotufo Neto, F., & Koenig, H. G. (2006). Religiousness and mental health:
A review. Revista Brasileira de Psiquiatria, 28(919), 242–250.
http://doi.org/10.1590/S1516-44462006000300018
Muthén, L. K., & Muthén, B. O. (2011). Mplus user’s guide (6th Edition). Los Angeles, CA:
Muthén & Muthén.
Nylund-Gibson, K., Grimm, R., Quirk, M., & Furlong, M. (2014). A latent transition mixture
model using the three-step specification. Structural Equation Modeling, 21(July), 439–454.
http://doi.org/10.1080/10705511.2014.915375
Nylund, K., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in
latent class analysis and growth mixture modeling: A Monte Carlo simulation study.
Hays
58

Structural Equation Modeling, 14(4), 535–569.
Parks, J., Svendsen, D., Singer, P., Foti, M. E., & Mauer, B. (2006). Morbidity and mortality in
people with serious mental illness. Alexandria, VA, USA. Retrieved from
http://nasmdarchive.aphsa.org/NASMHPD Morbidity and Mortality Report.pdf
Pew Research Center. (2015). America’s changing religious landscape. Retrieved from
http://www.pewforum.org/2015/05/12/americas-changing-religious-landscape/
Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general
population. Applied Psychological Measurement, 1(3), 385–401. Retrieved from
http://apm.sagepub.com/content/1/3/385.short
Romero, L. J., Ortiz, I. E., Finley, M., Wayne, S., & Lindeman, R. D. (2005). Prevalence of
depressive symptoms in New Mexico Hispanic and non-Hispanic white elderly. Ethnicity &
Disease, 15(4), 691–697.
Rutter, M. (2000). Psychosocial influences: Critiques, findings, and research needs. Development
and Psychopathology, 12, 375–405. http://doi.org/10.1017/S0954579400003072
Salami, T. K., & Walker, R. L. (2014). Socioeconomic status and symptoms of depression and
anxiety in African American college students: The mediating role of hopelessness. Journal
of Black Psychology, 40(3), 275–290. http://doi.org/10.1177/0095798413486158
Santini, Z. I., Koyanagi, A., Tyrovolas, S., Mason, C., & Haro, J. M. (2015). The association
between social relationships and depression: A systematic review. Journal of Affective
Disorders, 175, 53–65. http://doi.org/10.1016/j.jad.2014.12.049
Schwarzbach, M., Luppa, M., Forstmeier, S., König, H. H., & Riedel-Heller, S. G. (2014). Social
relations and depression in late life: A systematic review. International Journal of Geriatric
Psychiatry, 29(1), 1–21. http://doi.org/10.1002/gps.3971
Hays
59

Shim, R., Ye, J., Baltrus, P., Fry-Johnson, Y., Daniels, E., & Rust, G. (2012). Racial/ethnic
disparities, social support, and depression: Examining a social determinant of mental health.
Ethnicity & Disease, 22(1), 15–20. http://doi.org/10.1016/j.biotechadv.2011.08.021
Substance Abuse and Mental Health Services Administration. (2015). Risk and protective
factors. Retrieved from https://www.samhsa.gov/capt/practicing-effective-
prevention/prevention-behavioral-health/risk-protective-factors
Taylor, R. J., Chatters, L. M., & Abelson, J. M. (2012). Religious involvement and DSM-IV 12-
month and lifetime major depressive disorder among African Americans. Journal of
Nervous and Mental Disease, 200, 856–862.
http://doi.org/10.1097/NMD.0b013e31826b6d65
Taylor, R. J., Chatters, L. M., & Joe, S. (2011). Non-organizational religious participation,
subjective religiosity, and spirituality among older African Americans and Black
Caribbeans. Journal of Religion and Health, 50(3), 623–645. http://doi.org/10.1007/s10943-
009-9292-4
Taylor, R. J., Mattis, J. S., & Chatters, L. M. (1999). Subjective religiosity among African
Americans: A synthesis of findings from five national samples. Journal of Black
Psychology, 25(4), 524–543. http://doi.org/10.1177/0095798499025004004
Teo, A. R., Choi, H. J., & Valenstein, M. (2013). Social relationships and depression: Ten-year
follow-up from a nationally representative study. PLOS One, 8(4), 1–8.
http://doi.org/10.1371/journal.pone.0062396
Turner, R. J., & Avison, W. R. (2003). Status variations in stress exposure: Implications for the
interpretation of research on race, socioeconomic status, and gender. Journal of Health and
Social Behavior, 44(4), 488–505. http://doi.org/10.2307/1519795
Hays
60

Walrath, C. M., Petras, H., Mandell, D. S., Stephens, R. L., Holden, E. W., & Leaf, P. J. (2004).
Gender differences in patterns of risk factors among children receiving mental health
services: Latent class analyses. Journal of Behavioral Health Services & Research, 31,
297–311. http://doi.org/10.1007/BF02287292
Watkins, D. (2012). Depression over the adult life course for African American men: Toward a
framework for research and practice. American Journal of Mens Health, 6(3), 194–210.
http://doi.org/10.1177/1557988311424072
Williams, D. R., González, H. M., Neighbors, H. W., Nesse, R. M., Abelson, J. M., Sweetman,
J., & Jackson, J. S. (2007). Prevalence and distribution of major depressive disorder in
African Americans, Caribbean blacks, and non-Hispanic whites: Results from the National
Survey of American Life. Archives of General Psychiatry, 64(3), 305. Archives of General
Psychiatry, 64(3), 305–315. http://doi.org/10.1001/archpsyc.64.3.305
Williams, D. R., & Jackson, J. S. (2000). Race/ethnicity and the 2000 census: Recommendations
for African American and other Black populations in the United States. American Journal
of Public Health, 90(11), 1728–1730. http://doi.org/10.2105/AJPH.90.11.1728
Woodward, A. T., Taylor, R. J., & Chatters, L. M. (2011). Use of professional and informal
support by black men with mental disorders. Research on Social Work Practice, 21(3), 328–
336. http://doi.org/10.1177/1049731510388668

Hays
61

Chapter Three (Study 2): Identifying Paths to Depression: Stability and Transitions in
Risk for Depression among African Americans
Background
African Americans have relatively low prevalence rates of clinical depression
(approximately 5% in 2015; Substance Abuse and Mental Health Services Administration, 2015).
However, when African Americans experience depression, it is often debilitating and persistent
(Williams et al., 2007). African American adults with depression are likely to have poor
prognoses (Williams et al., 2007), chronic symptoms (U.S. Department of Health and Human
Services, 2011), functional disability (Noël et al., 2004), and comorbid chronic health conditions
(Pickett, Bazelais, & Bruce, 2013). This suggests that although the prevalence of depression
among African Americans is fairly low, the disease burden is high. However, more needs to be
known about contributors to depression in African Americans, the dynamic life conditions
associated with depression, or whether or not these life conditions change or remain stable over
time to influence depression outcomes. Not only would this information help to illuminate
contributors to the burden of depression in African Americans, but it could also inform
interventions targeting these causes to improve depression outcomes.
Some studies have attempted to explain why African Americans might have poor
depression outcomes by exploring various psychosocial, sociocultural, and individual risk and
protective factors. Factors such as socioeconomic status (Salami & Walker, 2014), social
relationships (Santini, Koyanagi, Tyrovolas, Mason, & Haro, 2015), psychosocial stress (Keyes,
Barnes, & Bates, 2011), and religious involvement (Taylor, Chatters, & Abelson, 2012) have all
been found to be related to depressive symptoms and outcomes in this group. However, many of
these studies were cross-sectional or focused on race comparisons to identify factors related to
poor depression outcomes for African Americans. Although this body of research has
Hays
62

significantly contributed to our understanding of the various factors associated with depression in
African Americans, the specific mechanisms at play that cause African Americans to experience
high burden related to depression still remain unclear.
At least three underexplored issues might explain poor depression outcomes among
African Americans. First, studies of depression in African Americans typically treat this
population as a monolithic group and thus do not consider vast within-group variation. Race
comparative studies, although informative, potentially obscure important heterogeneity in the
African American population (Whitfield, Allaire, Belue, & Edwards, 2008). This is a limitation
of existing research because it may be the case that not all African Americans experience high
depression burden but that there are subpopulations of African Americans who suffer in excess
which are not identifiable in studies that aggregate African Americans for race comparisons.
Studies that account for within-group heterogeneity have the potential to identify subgroups of
individuals who are most likely to experience high depressive symptoms and poor outcomes.
Second, although studies have linked individual risk and protective factors to depression,
research has not identified a depression risk typology based on specific constellations of risk and
protective factors for depression. Studies that examine the co-occurrence of multiple
sociocultural and psychosocial factors associated with depression in African Americans might
provide insight into contributors to excess depressive symptoms in a more comprehensive way.
Third, cross-sectional studies are limited in their ability to highlight changes or stability in the
experience of risk and protective factors over the life course. The human experience is dynamic
and cumulative advantage and disadvantage (CAD) theory suggests that early life experiences,
advantages, and disadvantages accumulate over time (Dannefer, 2003). Studies are needed to
identify how risk and protective factors accumulate over the life course, creating paths toward
Hays
63

depressive symptoms in African Americans. The identification of risk stability, or transitions, is
needed to detect variable paths, or transitions, in risk types that might lead to worse depression
outcomes.
Excessive depressive symptoms and poor depression outcomes, for African Americans
represent a significant public health problem. The extant literature on depression in African
Americans is limited in terms of explaining why African Americans experience more depressive
symptoms and poor depression outcomes. Studies that investigate profiles of co-occurring
depression risk factors (i.e., risk types) and examine how depression risk accumulates and
changes over time (i.e., risk paths) can potentially illuminate contributors to the excess burden of
depression in African Americans. Findings from this approach will be useful for identifying
African Americans with the highest risk of poor depression outcomes and tailoring interventions
to these subpopulations based on their particular risk path.
Risk and Protective Factors for Depression
Previous cross-sectional studies have identified several social and behavioral correlates of
depression in African Americans, including demographic characteristics, psychosocial stress,
social relationships, and religion. Characteristics like socioeconomic status, age, gender, and
marital status have been linked to depression (Chatters, Taylor, Woodward, & Nicklett, 2015;
Lincoln, Abdou, & Lloyd, 2014; Lincoln, Taylor, Watkins, & Chatters, 2011; Salami & Walker,
2014). For example, some results suggested that African Americans with less than a high school
education and those with low incomes are at increased risk of depression (Hudson, Neighbors,
Geronimus, & Jackson, 2012). However, these associations are inconclusive because other
studies have reported a positive association, with higher socioeconomic status being linked to
more depressive symptoms (Salami & Walker, 2014).
Hays
64

Several psychosocial factors have also been linked to depression outcomes in African
Americans. Social stress (e.g., financial stress) and negative life events (e.g., violence, trauma)
have been found to contribute to depressive symptoms. Studies have concluded that experiencing
financial stress or events like a death in the family or being assaulted is associated with increased
depressive symptoms among African American adults (Cutrona et al., 2005; Lantz, House, Mero,
& Williams, 2005; Mitchell & Ronzio, 2011; Turner & Avison, 2003). Further, social
relationships have been identified as both a risk and protective factor for depression in African
Americans. Specifically, negative social interactions (i.e., critical or demanding social
relationships; Lincoln & Chae, 2012) are associated with an increased risk of depression,
whereas positive social relationships and larger social networks are protective and associated
with a lower risk of depression (Chatters et al., 2015; Lincoln & Chae, 2012).
Religious involvement is a sociocultural factor that has also been linked to depression.
Studies have found that African Americans who reported receiving a “great deal of guidance”
from religion were about half as likely to have major depression as those with lower levels of
religiosity (Ellison & Flannelly, 2009). Studies also suggested that African Americans who
attend religious services more frequently are less likely to have depression than those who are
infrequent church attendees (Chatters et al., 2008).
This body of research provides insight into the individual risk and protective factors
associated with depression for African Americans. However, many of these cross-sectional
studies did not examine patterns of co-occurring risk factors associated with depression. This is a
significant gap in the literature because evidence suggests that health risks should be studied
simultaneously rather than focusing on the effect of individual risk factors (Chen, Eaton, Gallo,
& Nestadt, 2000; Laska, Pasch, Lust, Story, & Ehlinger, 2009). Identifying profiles, or types, of
Hays
65

risk may explicate the collective role these factors play in contributing to depression among
African Americans.
Further, life course research has suggested that certain life conditions considered to be
risk or protective factors vary and accumulate over time. Specifically, socioeconomic status,
stress, and social support are known to be dynamic and change over the course of an individual’s
life (Aneshensel & Frerichs, 1982; Lorant et al., 2007). Cross-sectional studies that examine risk
factors cannot provide insight into how risk changes over the life course. Longitudinal studies
are needed to identify changes in risk and depression outcomes over time. Longitudinal analysis
is also suited for uncovering the distinct risk paths that lead to excess depression burden in this
population.
No existing studies have examined stability or change in risk types for depression in
African Americans. However, a few examples from the substance abuse literature demonstrate
the benefits of identifying patterns and transitions in health risks. A study by Bray, Smith, Piper,
Roberts, and Baker (2016) investigated the relationship between smoking quit attempts and
transitions in social networks. The authors identified five classes of social networks and
concluded that smoking abstinence at Years 1 and 2 was associated with shifts in participants’
social networks to less contact with smokers and larger networks in Years 2 and 3. The authors
concluded that information about social network transitions can inform behavioral interventions
that increase nonsmoker social networks as a way of increasing successful smoking cessation
(Bray, Smith, Piper, Roberts, & Baker, 2016).
Another study by Huh and Leventhal (2016) investigated transitions in the use of
multiple tobacco products among adolescents. Study results revealed three patterns of tobacco
use and found that adolescents typically transition to a tobacco use pattern that involves the use
Hays
66

of more tobacco products (i.e., from hookah to e-cigarettes and cigarettes). Their results suggest
the need for increased surveillance of adolescents who are likely to progress into using more
tobacco products and limiting access to common entry points (e.g., hookah; Huh & Leventhal,
2016). Results from both of these studies indicate that patterns and transitions in health risks can
be empirically identified and linked to potential health outcomes. Given the significant public
health problem of poor depression outcomes for African Americans, research that seeks to
uncover possible contributors, like risk paths, is needed.
Theoretical Foundation
In addition to research, theory has suggested that some health risks accumulate across the
life course in various ways to influence depression outcomes (Garbarski, 2015). The current
study employed one mechanism from CAD theory that focuses on the accumulation of
experiences, like sociocultural and psychosocial risk factors, to help explain how different
pathways of risk influence depressive symptoms for African Americans.
The CAD model suggests that early life advantages and disadvantages persist, or
accumulate, in later life and are often magnified over the life course (Crystal, Shea, & Reyes,
2016; Dannefer, 2003). The path-dependent mechanism of the theory is most applicable to this
investigation and suggests that early life exposure to certain risk factors (e.g., low socioeconomic
status, chronic stress) persist over time and lead to poor health outcomes in later life (DiPrete &
Eirich, 2006; Willson, Shuey, & Elder, 2007). For example, an African American woman with
low educational attainment, low income, and low social support as a young adult is likely to
continue to experience the disadvantages associated with these life conditions throughout her
life. The accumulation of these disadvantages constitutes a path toward poor mental health as an
older adult because of the direct and indirect effects of these risks and the life conditions they
Hays
67

engender. However, existing research has not conclusively suggested specific constellations of
risk factors that constitute a path toward worse depression outcomes for African Americans.
Moreover, how these paths toward depression remain stable or change over time for African
Americans remains unknown. Understanding the composition of individual risk profiles and how
these profiles transition or remain stable will help identify which African Americans are likely to
have high depressive symptoms, how these risk profiles contribute to worse depression, and
consequently, which type of treatment options might be more effective.
A Conceptual Model of Risk Types and Paths to Depression
This study addressed three research questions: (a) Are there distinct constellations of risk
and protective factors (i.e., types) that describe unique constellations of psychosocial and
sociocultural factors for African Americans? (b) Do individuals remain stable in their risk type or
transition over time (i.e., paths)? (c) Can these paths (i.e., stable or transitioning risk) be
predicted by certain demographic factors and predict depressive symptoms? A conceptual model
served as a guide for this investigation (Figure 1). The model incorporates conceptual,
theoretical, and empirical elucidations from the extant literature on risk and protective factors for
depression among African Americans. Included are specific demographic, psychosocial, and
sociocultural factors that prior literature has shown to be associated with depression outcomes in
African Americans. In this model, specific constellations of risk and protective factors are
considered risk types and variable paths toward depression are modeled as risk transitions. It is
posited that the experience of particular paths of risk and protection, characterized by transitions
or stability in risk types, will lead to particular depression outcomes. By investigating transitions
in risk types over time, we can better understand the dynamic nature of risk and protective
factors and design depression interventions aimed at a moving target.
Hays
68

Methods
Data
Three waves of data from the Americans’ Changing Lives study were used to examine
stability or change in risk types for depression in African Americans. This ongoing nationally
representative study focuses on differences between Black and White Americans in middle and
late life. The study features a range of sociological, psychological, mental, and physical health
items. Wave 1 of the study began in 1986 with face-to-face interviews of 3,617 adults aged 25 or
older. African Americans and people older than 60 were oversampled. Data from Wave 2 (N =
2,867) was collected in 1989 via follow-up face-to-face and telephone interviews with
individuals who participated in Wave 1. During Wave 3 in 1994 (N = 2,562), Wave 4 in 2001
and 2002 (N = 1,787), and Wave 5 in 2011 (N = 1,427), participants were re-interviewed by
telephone, or when available, face to face. The average response rate for these five waves was
79%. The analytic sample for this analysis includes respondents who identified as African
American and participated in the study at Waves 1, 2, and 5 (N = 405). The data used in this
study were restricted to these waves because they contained all of the measures needed for this
analysis.
Dependent Variable
Depressive Symptoms. The outcome of interest in this study is depressive symptoms,
which were measured with 11 items from the Center for Epidemiological Studies Depression
Scale (CES-D; Radloff, 1977). Items measured the extent to which respondents felt happy,
lonely, sad, that everything was an effort, that their sleep was restless, that people were
unfriendly, that they did not feel like eating, that people dislike them, that they could not get
going, and that they enjoyed life. This abbreviated scale has been found to have acceptable
reliability and a similar factor structure compared to the original 20-item version (Foley, Reed,
Hays
69

Mutran, & DeVellis, 2002; Radloff, 1977). Responses were on a Likert-type scale with 1 =
hardly ever, 2 = some of the time, and 3 = most of the time. Positively worded items (i.e., felt
happy and enjoyed life) were reverse coded. Scores from the 11 items were averaged so that
higher mean scores indicated more depressive symptoms. Mean depression scores from Wave 5
of the survey were used.
Independent Variables
Psychosocial stress. Psychosocial stress includes the experience of chronic financial
stress and negative life events. Chronic financial stress was measured using three items from the
Americans’ Changing Lives study that assessed each respondent’s financial status. The items
asked respondents to indicate the following: (a) “How satisfied are you with your/your family’s
present financial situation?” Responses ranged from 1 (completely satisfied) to 5 (not at all
satisfied), (b) “How difficult is it for you/your family to meet the monthly payments on your
(family’s) bills?” Response options ranged from 1 (extremely difficult) to 5 (not difficult at all),
(c) “In general, how do your (family’s) finances usually work out at the end of the month—do
you find that you usually end up with …”; responses categories were 1 (some money left over), 2
(just enough money to make ends meet), and 3 (not enough money to make ends meet).
Responses to these items were reverse coded so that lower values indicated less stress. These
three items were indexed into a single variable representing chronic financial stress by obtaining
the mean score across the three items then recoding the mean score into a single variable. Unlike
Chapter 2, this study’s analysis required binary variables for improved model fit and ease of
interpretation in the Latent Transition Analysis. Thus, the indexed variable response options
included 1 (low financial stress) which represented original responses of “not difficult”, “slightly
difficult”, “some difficulty” and “just enough money to make ends meet” and ; 2 (high financial
Hays
70

stress) representing “very difficult”, “extremely difficult”, and “not enough money to makes
ends meet”.
Negative life events were assessed by asking respondents to indicate whether or not they
had experienced any trauma or loss including the death of a spouse, child, parent or stepparent;
divorce; job loss; assault; robbery; or “anything else bad” during the 3 years preceding the Wave
1 interview. Because this analysis focused on the overall experience of negative life events and
not any one event, a count variable was created with the number of events experienced ranging
from 0 (lowest) to 6 (highest). Next, to increase result interpretability and model fit, a
dichotomous variable was created so that 0 or 1 events = low, and 2 or more events = high.
Religious involvement. Religious involvement was assessed using two dimensions.
Organizational religious involvement was measured by asking respondents, “How often do you
usually attend religious services?” Response categories ranged from 1 (never) to 6 (more than
once a week). Response items were reverse coded and recoded so that 1 (low involvement)
represented attending church “never”, “less than once a month”, and “once a month”; and 2 (high
involvement) represented attending church “2-3 times a month” “once a week” or “more than
once a week”. Subjective religiosity was measured by asking respondents, “In general, how
important are religious or spiritual beliefs in your day to day life?” The original response
categories ranged from 1 (not at all important) to 4 (very important). For ease of interpretation
for the analysis responses were reverse coded and recoded so that 1 (low religiosity) represented
“not at all important”, “not too important”, and “fairly important”; and 2 (high religiosity)
represented “very important”. Although Chapter 2 included a measure for non-organizational
religious involvement this item was omitted in the current analysis as respondents were not asked
about non-organizational religious involvement at Waves 2 and 5.
Hays
71

Social support. Social support was measured by asking respondents to evaluate their
relationships with friends and relatives, spouse, children, mother, and father. Social support from
each source was assessed with two questions: (a) “On the whole, how much do your […] make
you feel loved and cared for?” and (b) “How much are these […] willing to listen when you need
to talk about your worries or problems?” Response categories ranged from 1 (a great deal) to 5
(not at all). These 10 items were reverse coded and indexed into one social support variable by
calculating the mean score across the 10 items. Next the distribution of the mean scores was
consulted to identify the cut off points that most closely reflected the distribution of responses
for the original variables. A mean score of 3.9 was used to create the response categories for the
indexed variable which included 1 = low, 2 = high.
Negative interaction. Negative interaction with friends and relatives, spouse, children
mother, and father were each measured by two items: (a) “How much do you feel your […]
make too many demands on you?” and (b) “How much are they critical of you or what you do?”
Response categories ranged from 1 (a great deal) to 5 (not at all). These 10 items were reverse
coded and indexed into one variable representing negative interaction. The distribution of the
mean scores was consulted to identify appropriate cut off points that most closely reflected the
distribution of responses for the original variables. A mean score of 2.44 was used to create
response categories of 1 = low, 2 = high negative interaction.
Demographic characteristics. Demographic variables included in this analysis were
age, gender, marital status, and education level. Age was recoded as a categorical variable (25–
39 years old, 40–59 years old, and 60 years old or older). Marital status was categorized as
married, separated, divorced, widowed, and never married. Education level was measured with a
categorical variable that included, below high school, high school, some college, and college
Hays
72

degree. Age, marital status, and education level were all dummy coded for logistic regression
analysis.
Missing Data
Mplus software (version 7) uses full-information maximum likelihood estimation based
on the assumption that data are missing at random (Arbuckle, 1996; Little & Schenker, 1995).
Analysis
This study examined the extent to which African Americans transition from one risk type
to another over time. Latent transition analysis (LTA) was conducted using Mplus 7 to identify
distinct risk transitions in the sample (Collins & Wugalter, 1992). LTA was selected as the
statistical method because it is designed to model change in categorical data (Roberts & Ward,
2011), thus making it an appropriate analytic method for identifying latent classes of individuals
based on their risk profile at multiple time points and probability of transitioning between latent
classes over time.
Statistical analysis proceeded in five steps as modeled by Roberts and Ward (2011). First,
latent class analysis (LCA) was used to identify and describe specific constellations of risk and
protective factors across Waves 1, 2, and 5. Using LCA, models with two to four classes were
preliminarily considered at each wave to provide guidance during model selection with LTA.
Model selection was based on the Akaike information criterion (AIC), Bayesian information
criterion (BIC), entropy, and model interpretability (Akaike, 1974; Schwarz, 1978). Item-
response probabilities represent the probabilities of having a particular response to the six items
that measure risk and protection in the model. These probabilities were examined and provided
the foundation for interpreting and naming the latent classes.
Hays
73

Second, a series of LTA models was run using the selected LCA models as a guide.
Models with two to four classes were considered to determine whether subgroups of participants
with unique types of risk and protection could be identified across multiple waves. Transition
probabilities were identified, which represent the probabilities of transitioning from one class to
another at each wave. Third, the selected LTA model was used to examine change over time in
risk and protection from Wave 1 to Wave 2 (1986 to 1989) and Wave 2 to Wave 5 (1989 to
2011). Fourth, a variable representing the most likely class assignment information was exported
back into the original dataset (Jung & Wickrama, 2008). To improve interpretability, and
because of small cell sizes of some of the transitioning classes, one variable representing the 8
paths was created which included three categories: Stable Low Protection/Low Risk, Stable High
Protection/Low Risk, and Transitioning Risk. One-way analysis of variance for continuous
variables was conducted to identify differences between the identified risk paths and final level
of depressive symptoms (CES-D score at Wave 5). Finally, multinomial logistic regression was
conducted to test associations between baseline demographic characteristics (i.e., gender, age,
education, and marital status) and risk transition type.
Results
Table 1 summarizes the demographic characteristics of the overall sample at baseline (N
= 405). Women constituted the majority of the sample (more than 68%). Younger respondents
(25–39 years old) were the largest age group (58.66%), followed by those who were 40–59 years
old (33.17%). There were few older adult respondents in the sample, with those 60 years old or
older constituting 8.17% of the sample. Individuals who were not married comprised more than
half (57.28%) of the sample. Educational attainment varied among respondents, with relatively
equivalent proportions of respondents distributed across three education categories; less than a
high school education (31.11%), a high school diploma (30.62%), and some college education
Hays
74

(27.41%). Fewer respondents in the sample had a college degree (10.86%). Concerning the risk
and protective factors, most respondents reported high levels of protective factors like
organizational (63.46%) and subjective religiosity (76.79%), and social support (65.96%). Very
few respondents reported having 2 or more negative life events (16.06%), high financial stress
(5.21%), or negative social interaction (27.10%).
Preliminary LCAs at each of the three waves suggested that a two-class solution was the
best overall fit for the data. This guided model selection for the LTA. A two-model solution was
also found to have the best overall fit to the data for the LTA (see Table 2). Although there was a
slight improvement in model for the 3-class versus the 2-class model the 3-class had poor
interpretability and overall fit to the data as there were no respondents in one of the classes as
Waves 2 and 3. Further, the two-class LTA model was consistent with the significant 2-class
LCA results at all three waves.
At baseline, 60% of respondents were described as belonging to a class named High
Protection/Low Risk. Respondents in this class tended to have low endorsement of risk factors
(e.g., stress) and high endorsement of protective factors (e.g., social support). Specifically, most
individuals in this class had high organizational religiosity (93%), high subjective religiosity
(95%), low experiences of negative life events (64%), low financial stress (93%), high social
support (71%), and low negative social interaction (71%). The other 40% of respondents
belonged to the Low Protection/Low Risk type. Respondents in this class were likely to
experience few protective factors (e.g., religiosity) and few risk factors. Specifically, most
respondents in this class had low organizational religiosity (83%), relatively low subjective
religiosity (60% = high vs. 95% = high in the High Protection/Low Risk class), low experiences
of negative life events (68%), low endorsement of financial stress (96%), relatively low social
Hays
75

support (57% = high vs. 71% = high in the High Protection/Low Risk class), and low negative
social interaction (62%) (see Table 3 and Figure 2). For the LTA, item response probabilities
were constrained to be equal across waves.
Transition probabilities are described in Table 4, which indicates that the majority of
respondents remained stable in their risk type from Wave 1 (1986) to Wave 2 (1989). Only 7%
of those in the High Protection/Low Risk type at Wave 1 transitioned to the Low Protection/Low
Risk type at Wave 2. Similarly, only 9% of those classified as Low Protection/Low Risk at Wave
1 transitioned to the High Protection/Low Risk type at Wave 2. From Wave 2 to Wave 5, most
respondents in the High Protection/Low Risk type remained stable in their class, with only 9%
transitioning into the Low Protection/Low Risk type. However, the proportion of respondents
who transitioned between classes from Wave 2 to Wave 5 increased among the Low
Protection/Low Risk respondents. Of those in the Low Protection/Low Risk type at Wave 2, 38%
transitioned to the High Protection/Low Risk type at Wave 5.
Eight possible risk transition patterns existed across the three waves of data. Figure 3
illustrates the proportion of respondents in the various transition types. Most respondents
belonged to the transition pattern that involved stable membership in the High Protection/Low
Risk type over time (54.32%). Those who remained stable in the Low Protection/Low Risk Type
across all three waves made up 23.95% of the sample. However, 21.73% of respondents were in
one of the six transition patterns characterized by a move between the two risk types at one or
more points in time during the 25-year study period.
Multinomial logistic regression was conducted to identify significant associations
between demographic variables and transition types (see Table 5). Because of the small cell sizes
of the six transitioning risk patterns, one summary variable representing risk transition types was
Hays
76

created and included three response categories: (a) Stable Low Protection/Low Risk across all
three waves, (b) Stable High Protection/Low Risk across all three waves, and (c) Transitioning
Risk (any change between risk types). Regression results indicated significant relationships
between risk transition type and gender and marital status. Female respondents were less likely
than males to have Transitioning Risk (OR = 0.55; CI = 0.30, 0.99) and Stable Low
Protection/Low Risk (OR = 0.37; CI = 0.21, 0.65) patterns versus the Stable High
Protection/Low Risk pattern. Married respondents were less likely than those who were never
married to have Stable Low Protection/Low Risk versus Stable High Protection/Low Risk (OR =
0.52; CI = 0.27, 0.99). Mean depression score at Wave 5 was also significantly related to risk
transitions. Analysis of variance results indicated that those in the Transitioning Risk category
had significantly higher CES-D scores than those in the Stable High Protection/Low Risk group
(difference in means = 0.15; CI = 0.03, 0.27; Table 6).
Discussion
This study examined stability and change in risk types for depression among African
Americans. There are three major findings from this study which highlight subpopulations of
African Americans most likely to experience poor depression and improve our understanding of
the contributors to depressive symptoms in this population. First, results revealed two risk types
(High Protection/Low Risk and Low Protection/Low Risk) that represent distinct constellations
of risk and protective factors for African Americans. Second, these risk types were found to
remain stable for some individuals and change over time for others. Both gender and marital
status were predictive of respondent’s risk transition type. Lastly, results suggested that
individuals whose risk type changed over time were likely to have higher depressive symptoms
than those with stable risk. These results are generally consistent with other studies that have
posited that risk and protective factors co-occur to affect health (Lincoln, 2000; Lincoln & Chae,
Hays
77

2012) and theory that suggests that risk and protective factors may accumulate over the life
course to constitute distinct paths towards depression (Dannefer, 2003).
Risk Types
Chapter 2 of this dissertation has more extensively examined the various risk types
associated with depression in African Americans. However, it was important to revisit this risk
typology for the current analysis which focuses on understanding stability and change in risk
types over time. This current study identified two distinct risk types which represent
constellations of life conditions that prior research has determined are individual risk or
protective factors for depression. Individuals in the High Protection/Low Risk type had low risk
factors (i.e., negative life events, financial stress, and negative social interaction) and high
protective factors (i.e., organizational and subjective religiosity, social support). Those in the
Low Protection/Low Risk type also tended to have relatively low experience of risk factors, like
financial stress and negative life events; however, they were less likely to endorse experiencing
high levels of protective factors like religiosity and social support. What appears to distinguish
the two risk types the most is the experience of protective factors, rather than risk factors. That
is, both classes tended to have similarly low endorsement of risk factors like psychosocial stress,
however, the two risk types differed greatly in the extent to which respondents experienced high
social support and religiosity, which are suggested protective factors. These findings suggest the
importance of emphasizing the collective role of protective factors in influencing health
outcomes.
Stable and Transitioning Risk
In addition to identifying distinct risk types for African Americans, this study also
identified stability and change in risk over time. The two stable transition types, High
Hays
78

Protection/Low Risk and Low Protection/Low Risk across all three waves, were the most
commonly endorsed, with nearly 54% and 24%, respectively. In other words, most respondents
remained stable in their risk type across 25 years. Those African Americans who had strong
social relationships, high levels of religiosity, and low stressors in 1986 were likely to report
experiencing those same life conditions 3 years and 25 years later. Likewise, most of those who
were disadvantaged by poor social support and low religious involvement earlier in life were
likely to remain disadvantaged over time.
Those in the Stable High Protection/Low Risk type can be described by certain
demographic characteristics. Women were more likely to be in the Stable High Protection/Low
Risk type compared to the Stable Low Protection/Low Risk or Transitioning Risk groups. This is
not surprising given the number of studies that have suggested that women tend to have stronger
social networks and are more religious than men (Barker, Morrows, & Mitteness, 1998; Levin,
Taylor, & Chatters, 1994; Taylor, Mattis, & Chatters, 1999). Many studies also suggested that
women are more likely to experience depression than men (Bracke, 2000; Essau, Lewinsohn,
Seeley, & Sasagawa, 2010). However, results of this study highlight a constellation of mediating
factors that more fully describe the relationship between gender and depression. In this study,
men were more likely than women to belong to the risk transition type (i.e. Transitioning Risk)
associated with significantly more depressive symptoms. This suggests that although men may
be less likely to have depression, they experience risk and protective factors during the adult life
course that are linked to higher depressive symptoms. Other studies have suggested that
depression in African American men is often overlooked, leading to underdiagnoses and
treatment disparities (Mizell, 1999; Watkins, 2012). More research should focus on the multiple
factors that contribute to depressive symptoms in African American men, including the
Hays
79

psychosocial and sociocultural factors examined in this study, but also factors such as violence
and incarceration, which are more likely to affect African American men than women (Perkins,
2014; Xanthos, Treadwell, & Holden, 2010).
Respondents who were married at baseline were also more likely to follow a Stable High
Protection/Low Risk path than a Stable Low Protection/Low Risk path. Married individuals may
benefit from larger social and family networks and religious involvement (which are defining
features of the High Protection/Low Risk type) to a greater extent than those who have never
been married. However, this study’s results suggest that it is not marriage alone but rather the
experience of specific sociocultural and psychosocial factors over time that protects against
depression. This provides support for interventions that seek to increase family cohesion and
social support for nonmarried individuals, who may not have as many naturally occurring
protective factors to buffer them from depression as their married counterparts.
Nearly 20% of all respondents in the sample transitioned between High Protection/Low
Risk and Low Protection/Low Risk at some point during the 25 years they participated in the
survey. Most respondents who transitioned between risk types did so between Wave 2 (1989)
and Wave 5 (2011) with most of those transitioning from Low Protection/Low Risk to High
Protection/Low Risk. This might be attributed to the larger amount of time elapsed between
Waves 2 and 5 versus Waves 1 and 2. During this 22-year period, respondents would have
experienced many developmental and life changes that are common to progression through the
life course which may have resulted in a change in their risk type.
Nearly 40% of those in the Low Protection/Low Risk type at Wave 2 transitioned to High
Protection/Low Risk at Wave 5. This trend in risk transition might be explained by what is
known about life course development. Life course researchers have suggested that as people age,
Hays
80

they move through various stages of family life, education and career trajectories, retirement, and
health statuses (Clarke, Marshall, House, & Lantz, 2011). Young adulthood is typified by gains
in status and roles (i.e., early career and marriage) that have mental health benefits. Midlife is
characterized by stability of marital, employment, and social statuses, during which mental
illness is at its lowest point (Clarke et al., 2011). Late life is marked by role exits and physical
declines linked to worse mental health (Clarke et al., 2011; Mirowsky & Ross, 1992). In this
study, the move from Low Protection/Low Risk (marked by low social support and religiosity) to
High Protection/Low Risk (marked by high support and religiosity) might reflect changes in
roles and statuses related to individuals’ moving from early and midlife to mid- and late life. It
may be that African Americans who experienced Low Protection/Low Risk earlier in life attain
more stability and support through marriage, employment, and social status as they advance into
mid- and late life.
It is also very likely that many African Americans become more religiously involved as
they age because studies have uniformly concluded that age is positively associated with
religiosity among African Americans (Taylor, Chatters, & Levin, 2003). This trend towards
increasing religiosity with age might also help explain transitions into the High Protection/Low
Risk group (marked by high levels of religiosity). Further, socioemotional selectivity theory
suggests that as individuals move through adulthood they become more focused on emotional
regulation and thus become highly selective in their choice of social partners (Carstensen, 1995).
Thus, some African Americans who experienced low protective factors earlier in life may
become more selective of their social networks as they advance through the life course and
choose to interact with others who can provide them with more social support and emotional
regulation.
Hays
81

Risk Transitions and Depression
The path-dependent mechanism of CAD theory posits that early life exposures to certain
risk factors (e.g., low socioeconomic status, chronic stress) accumulate and lead to poor health
outcomes in later life (DiPrete & Eirich, 2006; Willson et al., 2007). Moreover, the theory posits
that the particular accumulation of life advantages and disadvantages constitutes paths towards
health outcomes (depression in this case). Based on theory, it could be assumed that individuals
in the Stable Low Protection/Low Risk group would be most likely to have high depressive
symptoms because these individuals experience a path characterized by persistent, or
accumulating, disadvantage evidenced by few protective factors over time. However, results
indicate that although African Americans in the Stable Low Protection/Low Risk group have
higher mean depression scores than their High Protection/Low Risk counterparts this difference
is not statistically significant. Essentially, individuals with low protective factors across time are
no more likely to have high depressive symptoms that those with high protective factors over
time.
The subgroup of African Americans with the highest depressive symptoms were those in
the Transitioning Risk group characterized by movement between Low Protection/Low Risk and
High Protection/Low Risk. Transitioning risk was associated with significantly higher mean
depression (m=1.57) compared to Stable High Protection/Low Risk (m=1.42) or Stable Low
Protection/Low Risk (m=1.53). In other words, change in risk type is related to increased
depression. Because of the small number of respondents in some of the 6 transitioning paths it
was not feasible to analyze individual paths or separate those paths that moved towards more
protection from those that moved towards less. Nevertheless, results suggest that identifying risk
types, or constellations of risk and protective factors, alone may not explain the contributors to
Hays
82

depression in African Americans. Understanding the experience of risk and protection over the
life-course and how individuals change from one type to another will illuminate the dynamic life
conditions that lead to worse depression. These results should not be surprising given research
indicating that significant life events and changes, both positive and negative, are related to some
level of psychological distress (George, 1993). For example, becoming a parent and entering
marriage are common experiences in the life-course however for some these events may lead to
significant emotional distress. Thus, for African Americans, the gain or loss of protective factors,
like religiosity and social support, may be indicative of instability in the life course or the
experience of non-normative life changes. It is this instability which is predictive of worse
depressive symptoms and should be the focus of attention.
Additionally, there is partial support for CAD theory in that there were identified paths
(i.e. Transitioning Risk) that were predictive of higher depression. However, as an addition to
CAD theory we now know that some flexibility might exist in the path toward health and that it
may be possible for some African Americans who are described as Low Protection/Low Risk
earlier in life to gain protective factors and move to a High Protection/Low Risk status. Because
individuals with transitioning risk were likely to have higher depressive symptoms the goal for
interventions should be to help African Americans gain and maintain protective factors early in
life to reduce the likelihood of experiencing the changes in risk types that lead to poor depression
outcomes.
Hays
83

Figure 3.1. Conceptual Model of Depression Risk Transitions in African American Adults, Ages
25 and Older

Hays
84

Table 3.1. Overall Characteristics of African Americans at Wave 1 (n=405)
Variable % (Mean) n (high, low)
Sex
Men 32.85 129
Women 68.15 276
Age
25-39 58.66 237
40-59 33.17 134
60+ 8.17 33
Marital status
Married 42.72 173
Separated 10.86 44
Divorced 15.31 62
Widowed 6.17 25
Never Married 24.94 101
Education
Below High School 31.11 126
High School 30.62 124
Some College 27.41 111
College Degree 10.86 44
Organizational religiosity
Low 36.54 148
High 63.46 257
Subjective religiosity
Low 23.21 94
High 76.79 311
Negative life events
Low 84.94 344
High 16.06 61
Chronic financial stress
Low 94.79 382
High 5.21 21
Social support
Low 34.07 138
High 65.93 267
Negative social interaction
Hays
85

Low 72.10 292
High 27.10 113
CESD 1.48 1.00, 2.73
Hays
86

Table 3.2. Model Fit Indexes for Latent Transition Models with Two to Four Classes
BIC AIC Entropy
Two classes 14,950.788 14,864.629 .644
Three classes 14,650.207 14,518.434 .649
Four classes 14,649.923 14,472.537 .616
Note. Preliminary LCA of each individual wave supported a two-class solution. Increased entropy and lower BIC
and AIC suggest improved model fit. These indexes support the use of a two-class model because there is little
statistical benefit to the three- or four-class models

Hays
87

Table 3.3. Item Response Probabilities for Latent Risk Types (N = 405)
High
Protection/Low Risk
Low Protection/Low
Risk
(60%) (40%)
Organizational religiosity
Low .07 .83
High .93 .17
Subjective religiosity
Low .05 .40
High .95 .60
Negative life events
Low .64 .68
High .36 .32
Chronic financial stress
Low .93 .96
High .07 .04
Social support
Low .29 .43
High .71 .57
Negative social interaction
Low .71 .62
High .29 .38
Note. Percentage of the sample in each class is at Wave 1. Items constrained to be equal across waves
for transition analysis.
Hays
88

Figure 3.2 Probabilities for Risk and Protective Factors for each Risk Type
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
High Organized
Religion
High
Subjective
Religion
High Social
Support
High Negative
Events
High Financial
Stress
High Negative
Interaction
Probability of Endorsement
High Protection and
Low Risk
Low Protection and
Low Risk
Hays
89

Table 3.4. Transition Probabilities
Wave 2 (1989)
High Protection/Low Risk Low Protection/Low Risk
Wave 1 (1986)
High Protection/Low Risk .93 .07
Low Protection/Low Risk .09 .91
Wave 5 (2011)
Wave 2 (1989)
High Protection/Low Risk .91 .09
Low Protection/Low Risk .38 .62
Note. Transition probabilities interpreted as the probabilities of membership in classes at time t + 1
(columns) conditional on membership in classes at time t (rows); transition probabilities sum to 1.0 (within
rounding error) within a row.
Hays
90

Figure 3.3. Proportion of Respondents in Each Transition Type
Note: Hp = High Protection/Low Risk
Lp = Low Protection/Low Risk

54.3
3
0.7
1.5
2
0
14.6
24
0
10
20
30
40
50
60
HpHpHp HpHpLp HpLpHp HpLpLp LpHpHp LpHpLp LpHpHp LpLpLp
Percent of Sample
Transition Types
Risk Paths
Hays
91

Table 3.5. Multinomial Logistic Regression of Baseline Demographic Covariates
on Risk Transitions
Transitioning Risk
vs.
Stable High Protection/
Low Risk
Stable Low
Protection/Low Risk
vs.
Stable High Protection/
Low Risk
OR (95% CI) OR (95% CI)
Female
a
0.55 (0.30, 0.99) 0.37 (0.21, 0.65)
Age
b

25–39 2.77 (0.77, 9.93) 3.17 (0.78, 12.83)
40–59 1.74 (0.49, 6.22) 2.23 (0.55, 9.10)
Marital status
c

Married 0.83 (0.47, 1.64) 0.52 (0.27, 0.99)
Separated 1.69 (0.68, 4.16) 0.55 (0.19, 1.53)
Divorced 0.84 (0.35, 2.04) 0.60 (0.26, 1.40)
Widowed 1.98 (0.54, 7.25) 0.69 (0.15, 3.09)
Education
d

Less than high school 1.83 (0.63, 5.30) 1.28 (0.49, 3.35)
High school 1.72 (0.61, 4.86) 1.62 (0.65, 4.01)
Some college 1.32 (0.47, 3.67) 0.57 (0.23, 1.46)
a
Reference is male.
b
Reference is 60 or older.
c
Reference is never married.
d
Reference is college degree

Hays
92

Table 3.6. Differences in CES-D Score for Risk Transition Types
Stable High
Protection/Low Risk
Stable Low
Protection/Low Risk
Transitioning
Risk

M (SD) M (SD) M (SD) F p
CES-D 1.42 (0.36)* 1.53 (0.39) 1.57 (0.45)* 5.50 .004
* Statistically significant difference between means.

Hays
93

References
Akaike, H. (1974). A new look at the statistical model identification. IEEE Transactions on
Automatic Control, 19(6), 716–723. http://doi.org/10.1109/TAC.1974.1100705
Aneshensel, C. S., & Frerichs, R. R. (1982). Stress, support, and depression: A longitudinal
causal model. Journal of Community Psychology, 10(4), 363–376.
http://doi.org/10.1002/1520-6629(198210)10:4<363::AID-JCOP2290100407>3.0.CO;2-8
Arbuckle, J. L. (1996). Full information estimation in the presence of incomplete data. In G. A.
Marcoulides & R. E. Schumacker (Eds.), Advanced structural equation modeling: Issues
and techniques (pp. 243–277). Mahwah, NJ: Lawrence Erlbaum.
Barker, J. C., Morrows, J., & Mitteness, L. S. (1998). Gender, informal social support networks,
and elderly urban African Americans. Journal of Aging Studies, 12, 199–222.
http://doi.org/http://dx.doi.org/10.1016/S0890-4065(98)90015-9
Bracke, P. (2000). The three-year persistence of depressive symptoms in men and women. Social
Science & Medicine, 51–64. http://doi.org/10.1016/S0277-9536(99)00432-3
Bray, B. C., Smith, R. A., Piper, M. E., Roberts, L. J., & Baker, T. B. (2016). Transitions in
smokers’ social networks after quit attempts: A latent transition analysis. Nicotine &
Tobacco Research, 18, 2243–2251. http://doi.org/10.1093/ntr/ntw173
Carstensen, L. L. (1995). Evidence for a life-span theory of socioemotional selectivity. Current
Directions in Psychological Science, 4(5), 151–156.
Chatters, L. M., Bullard, K. M., Taylor, R. J., Woodward, A. T., Neighbors, H. W., & Jackson, J.
S. (2008). Religious participation and DSM IV disorders among older African Americans:
Findings from the National Survey of American Life. American Journal of Geriatric
Psychiatry, 16(12), 957–965. http://doi.org/10.1097/JGP.0b013e3181898081.
Chatters, L. M., Taylor, R. J., Woodward, A. T., & Nicklett, E. J. (2015). Social support from
Hays
94

church and family members and depressive symptoms among older African Americans.
American Journal of Geriatric Psychiatry, 23(6), 559–567.
http://doi.org/10.1016/j.jagp.2014.04.008
Chen, L. S., Eaton, W. W., Gallo, J. J., & Nestadt, G. (2000). Understanding the heterogeneity of
depression through the triad of symptoms, course and risk factors: A longitudinal,
population-based study. Journal of Affective Disorders, 59(1), 1–11.
http://doi.org/10.1016/S0165-0327(99)00132-9
Clarke, P., Marshall, V., House, J., & Lantz, P. (2011). The social structuring of mental health
over the adult life course: Advancing theory in the sociology of aging. Social Forces, 89(4),
1287–1313. http://doi.org/10.1093/sf/89.4.1287
Collins, L. M., & Wugalter, S. E. (1992). Latent class models for stage-sequential dynamic latent
variables. Multivariate Behavior Resesearch, 27(1), 131–157.
http://doi.org/10.1207/s15327906mbr2701_8
Crystal, S., Shea, D. G., & Reyes, A. M. (2016). Cumulative advantage, cumulative
disadvantage, and evolving patterns of late-life inequality. Gerontologist, Advance on, 1–
11. http://doi.org/10.1093/geront/gnw056
Cutrona, C. E., Russell, D. W., Brown, P. A., Clark, L. A., Hessling, R. M., & Gardner, K. A.
(2005). Neighborhood context, personality, and stressful life events as predictors of
depression among African American women. Journal of Abnormal Psychology, 114(1), 3–
15. http://doi.org/10.1037/0021-843X.114.1.3
Dannefer, D. (2003). Cumulative advantage/disadvantage and the life course: cross-fertilizing
age and social science theory. Journals of Gerontology, Series B: Psychological Science
and Social Sciences, 58(6), S327–S337. http://doi.org/10.1093/geronb/58.6.S327
Hays
95

DiPrete, T. a., & Eirich, G. M. (2006). Cumulative advantage as a mechanism for inequality: A
review of theoretical and empirical developments. Annual Review of Sociology, 32(1), 271–
297. http://doi.org/10.1146/annurev.soc.32.061604.123127
Ellison, C. G., & Flannelly, K. (2009). Religious involvement and risk of major depression in a
prospective nationwide study of African American adults. Journal of Nervous and Mental
Disease, 197(8), 568–573. http://doi.org/10.1097/NMD.0b013e3181b08f45
Essau, C. A., Lewinsohn, P. M., Seeley, J. R., & Sasagawa, S. (2010). Gender differences in the
developmental course of depression. Journal of Affective Disorders, 127, 185–190.
http://doi.org/10.1016/j.jad.2010.05.016
Foley, K. . L., Reed, P. S., Mutran, E. J., & DeVellis, R. F. (2002). Measurement adequacy of the
CES-D among a sample of older African–Americans. Psychiatry Research, 109(1), 61–69.
http://doi.org/10.1016/S0165-1781(01)00360-2
Garbarski, D. (2015). Racial/ethnic disparities in midlife depressive symptoms: The role of
cumulative disadvantage across the life course. Advances in Life Course Research, 23, 67–
85. http://doi.org/10.1016/j.alcr.2014.12.006
George, L. K. (1993). Sociological perspectives on life transitions. Annual Review of Sociology,
19, 353–373. http://doi.org/10.1146/annurev.so.19.080193.002033
Hudson, D. L., Neighbors, H. W., Geronimus, A. T., & Jackson, J. S. (2012). The relationship
between socioeconomic position and depression among a US nationally representative
sample of African Americans. Social Psychiatry and Psychiatric Epidemiology, 47, 373–
381. http://doi.org/10.1007/s00127-011-0348-x
Huh, J., & Leventhal, A. M. (2016). Progression of poly-tobacco product use patterns in
adolescents. American Journal of Preventive Medicine, 51(4), 513–517.
Hays
96

http://doi.org/10.1016/j.amepre.2016.04.004
Jung, T., & Wickrama, A. (2008). An introduction to latent class growth analysis and growth
mixture modeling. Social and Personality Psychology Compass, 2(1), 302–317.
http://doi.org/10.1111/j.1751-9004.2007.00054.x
Keyes, K. M., Barnes, D. M., & Bates, L. M. (2011). Stress, coping, and depression: Testing a
new hypothesis in a prospectively studied general population sample of US-born Whites
and Blacks. Social Science & Medicine, 72(5), 650–659.
http://doi.org/10.1016/j.socscimed.2010.12.005
Lantz, P. M., House, J. S., Mero, R. P., & Williams, D. R. (2005). Stress, life events, and
socioeconomic disparities in health: Results from the Americans’ Changing Lives study.
Journal of Health and Social Behavior, 46(3), 274–288.
http://doi.org/10.1177/002214650504600305
Laska, M. N., Pasch, K. E., Lust, K., Story, M., & Ehlinger, E. (2009). Latent class analysis of
lifestyle characteristics and health risk behaviors among college youth. Prevention Science,
10(4), 376–386. http://doi.org/10.1007/s11121-009-0140-2
Levin, J. S., Taylor, R. J., & Chatters, L. M. (1994). Race and gender differences in religiosity
among older adults: Findings from four national surveys. Journal of Gerontology, 49(3),
S137–S145. http://doi.org/10.1093/geronj/49.3.S137
Lincoln, K. D. (2000). Social support, negative social interactions and psychological well-being.
Social Service Review, 74(2), 231–252. http://doi.org/10.1086/514478
Lincoln, K. D., Abdou, C. M., & Lloyd, D. (2014). Race and socioeconomic differences in
obesity and depression among Black and non-Hispanic White Americans. Journal of Health
Care for the Poor and Underserved, 25(1), 257–275. http://doi.org/10.1353/hpu.2014.0038
Hays
97

Lincoln, K. D., & Chae, D. H. (2012). Emotional support, negative interaction and major
depressive disorder among African Americans and Caribbean Blacks: Findings from the
National Survey of American Life. Social Psychiatry and Psychiatric Epidemiology, 47(3),
361–372. http://doi.org/10.1007/s00127-011-0347-y
Lincoln, K. D., Taylor, R. J., Watkins, D. C., & Chatters, L. M. (2011). Correlates of
psychological distress and major depressive disorder among African American men.
Research on Social Work Practice, 21(3), 278–288.
http://doi.org/10.1177/1049731510386122
Little, R. J. A., & Schenker, N. (1995). Missing data. In G. Arminger, C. C. Clogg, & M. E.
Sobel (Eds.), Handbook of statistical modeling for the social and behavioral sciences (pp.
39–75). New York, NY: Springer.
Lorant, V., Croux, C., Weich, S., Deliège, D., Mackenbach, J., & Ansseau, M. (2007).
Depression and socio-economic risk factors : 7-year longitudinal population study. British
Journal of Psychiatry, 190, 293–298. http://doi.org/10.1192/bjp.bp.105.020040
Mirowsky, J., & Ross, C. E. (1992). Age and depression. Journal of Health and Social Behavior,
33(3), 187–205. http://doi.org/10.2307/2137349
Mitchell, S. J., & Ronzio, C. R. (2011). Violence and other stressful life events as triggers of
depression and anxiety: What psychosocial resources protect African American mothers?
Maternal and Child Health Journal, 15(8), 1272–1281. http://doi.org/10.1007/s10995-010-
0668-6
Mizell, C. A. (1999). Life course influences on African American men’s depression: Adolescent
parental composition, self-concept, and adult earnings. Journal of Black Studies, 29, 467–
490. http://doi.org/10.1177/002193479902900401
Hays
98

Noël, P. H., Williams, J.W, J., Unutzer, J., Worchel, J., Lee, S., Cornell, J., … Hunkeler, E.
(2004). Depression and comorbid illness in elderley primary care patients: Impact on
multiple domains of health status and well-being. Annals of Family Medicine, 2, 555–562.
http://doi.org/10.1370/afm.143.
Perkins, D. E. . (2014). Challenges to traditional clinical definitions of depression in young black
men. American Journal of Men’s Health, 8(1), 74–81.
http://doi.org/10.1177/1557988313494234
Pickett, Y. R., Bazelais, K. N., & Bruce, M. L. (2013). Late-life depression in older African
Americans: A comprehensive review of epidemiological and clinical data. International
Journal of Geriatric Psychiatry, 28(6), 903–913. http://doi.org/10.1002/gps.3908
Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general
population. Applied Psychological Measurement, 1(3), 385–401.
http://doi.org/10.1177/014662167700100306
Roberts, T. J., & Ward, S. E. (2011). Using latent transition analysis in nursing research to
explore change over time. Nursing Research, 60(1), 73–79.
http://doi.org/10.1097/NNR.0b013e3182001c63
Salami, T. K., & Walker, R. L. (2014). Socioeconomic status and symptoms of depression and
anxiety in African American college students: The mediating role of hopelessness. Journal
of Black Psychology, 40(3), 275–290. http://doi.org/10.1177/0095798413486158
Santini, Z. I., Koyanagi, A., Tyrovolas, S., Mason, C., & Haro, J. M. (2015). The association
between social relationships and depression: A systematic review. Journal of Affective
Disorders, 175, 53–65. http://doi.org/10.1016/j.jad.2014.12.049
Schwarz, G. (1978). Estimating the dimension of a model. Annals of Statistics, 6(2), 461–464.
Hays
99

http://doi.org/10.1214/aos/1176344136
Substance Abuse and Mental Health Services Administration. (2015). Results from the 2015
National Survey on Drug Use and Health: Mental health findings. Rockville, MD.
Taylor, R. J., Chatters, L. M., & Abelson, J. M. (2012). Religious involvement and DSM-IV 12-
month and lifetime major depressive disorder among African Americans. Journal of
Nervous and Mental Disease, 200, 856–862.
http://doi.org/10.1097/NMD.0b013e31826b6d65
Taylor, R. J., Chatters, L. M., & Levin, J. (2003). Religion in the lives of African Americans:
Social, psychological, and health perspectives. Thousand Oaks, CA: Sage Publications.
Taylor, R. J., Mattis, J. S., & Chatters, L. M. (1999). Subjective religiosity among African
Americans: A synthesis of findings from five national samples. Journal of Black
Psychology, 25(4), 524–543. http://doi.org/10.1177/0095798499025004004
Turner, R. J., & Avison, W. R. (2003). Status variations in stress exposure: Implications for the
interpretation of research on race, socioeconomic status, and gender. Journal of Health and
Social Behavior, 44(4), 488–505. http://doi.org/10.2307/1519795
U.S. Department of Health and HumanServices. (2011). 2010 National healthcare disparities
report. AHRQ Publication No.11-0005. Rockville, MD.
Watkins, D. C. (2012). Depression over the adult life course for African American men: Toward
a framework for research and practice. American Journal of Mens Health, 6(3), 194–210.
http://doi.org/10.1177/1557988311424072
Whitfield, K. E., Allaire, J. C., Belue, R., & Edwards, C. L. (2008). Are comparisons the answer
to understanding behavioral aspects of aging in racial and ethnic groups? The Journals of
Gerontology. Series B, Psychological Sciences and Social Sciences, 63(5), P301–P308.
Hays
100

http://doi.org/10.1093/geronb/63.5.P301
Williams, D. R., González, H. M., Neighbors, H., Nesse, R., Abelson, J. M., Sweetman, J., &
Jackson, J. S. (2007). Prevalence and distribution of major depressive disorder in African
Americans, Caribbean blacks, and non-Hispanic whites: Results from the National Survey
of American Life. Archives of General Psychiatry, 64(3), 305–315.
http://doi.org/10.1001/archpsyc.64.3.305
Willson, A. E., Shuey, K. M., & Elder, G. H. (2007). Cumulative advantage processes as
mechanisms of inequality in life course health. American Journal of Sociology, 112(6),
1886–1924. http://doi.org/10.1086/512712
Xanthos, C., Treadwell, H. M., & Holden, K. B. (2010). Social determinants of health among
African–American men. Journal of Men’s Health, 7(1), 11–19.
http://doi.org/10.1016/j.jomh.2009.12.002

Hays
101

Chapter Four (Study 3): Variable Depression Trajectories among African Americans
Depression among African Americans is not well understood. Prevalence studies
suggested that African Americans have relatively low occurrence of clinical depression (Breslau
et al., 2006). Yet African Americans disproportionately suffer from excessive symptoms
(Bromberger, Harlow, Avis, Kravitz, & Cordal, 2004; Romero, Ortiz, Finley, Wayne, &
Lindeman, 2005), poor prognosis (Williams et al., 2007), and poor outcomes. The excess burden
of depression experienced by African Americans is a significant public health problem because
of its debilitating and costly consequences. Depression is associated with increased mortality due
to comorbid health conditions (Druss, Zhao, Von Esenwein, Morrato, & Marcus, 2011), poor
quality of life (Hirschfeld et al., 2000), lost economic potential, and burden on the health care
system (Donohue & Pincus, 2007). To improve depression outcomes in this population,
researchers must investigate heterogeneity both in the experience of depression and within the
African American population to identify subpopulations of African Americans who are most
likely to experience the poor depression outcomes cited in prior research. This information will
aid in development of prevention and treatment efforts to improve depression outcomes in this
group that are both targeted to the specific conditions that African Americans experience and
tailored to the subgroups most likely to suffer.
The literature on depression in the general population suggests that the course of
depression is variable with periods of remission, recovery, relapse, and recurrence (Richards,
2011). Studies have also suggested that the experience of depression is heterogeneous, with some
individuals having a more chronic and persistent course than others (Musliner, Munk-Olsen,
Eaton, & Zandi, 2016; Sutin et al., 2013). However, very little is known about the variable
experience of depression for African Americans, how this variation may differ from the general
population, and whether or not certain subgroups of African Americans experience a worse
Hays
102

course of the illness. This is a critical gap in the literature because identifying subpopulations of
African Americans who are likely to have a chronic or persistent course of depressive symptoms,
or depression trajectory, could provide direction to interventions seeking to focus on individuals
who are most likely to suffer excessively. The current study addressed these limitations of prior
research by investigating heterogeneity in depression trajectories in a nationally representative
sample of African American adults.
Background
African Americans and Depression
Approximately 5% of African Americans experienced depression in 2015 (Substance
Abuse and Mental Health Services Administration, 2015). Depression is associated with a host
of negative outcomes. For example, African Americans with depression are likely to experience
poor depression prognoses (Williams et al., 2007), chronic symptoms (U.S. Department of
Health and HumanServices, 2011), functional disability (Noël et al., 2004) and comorbid chronic
health conditions (Pickett, Bazelais, & Bruce, 2013). Further, depression is associated with lower
life expectancy (Colton & Manderscheid, 2006; Druss et al., 2011) particularly for African
Americans (Jackson, Knight, & Rafferty, 2010). Depression is a serious public health concern
and much that is still not understood about depression among African Americans. Several
limitations of existing research must be addressed to improve our understanding of, and ability to
intervene on, disparate depression outcomes experienced by African Americans.
First, research has suggested that the course of depression is highly variable, with periods
of symptom persistence, remission, and recurrence (Liang, Xu, Quiñones, Bennett, & Ye, 2011;
Lincoln & Takeuchi, 2010; Spence, Adkins, & Dupre, 2011). Longitudinal studies have
indicated that depressive symptoms can improve, worsen, or remain stable at high or low levels
over time (Lincoln & Takeuchi, 2010; Spence et al., 2011). General population studies have
Hays
103

confirmed that heterogeneity exists in the individual experience of depression (Richards, 2011).
However, variation in the course of depression within the African American population has not
been documented. This is important because based on this prior research, there is reason to
believe that there is heterogeneity in the experience of depression among African Americans that
may lead some individuals to have a worse course of illness than others. Subpopulations that
have a worse course of depressive symptoms cannot be identified through research that treats
African Americans as a monolithic group. Understanding how the course of depression varies
from one individual to another can provide insight into the excess burden of depression by
accounting for potentially divergent trajectories that might offset one another if considered in the
aggregate. Further, interventions that are targeted and tailored to subgroups with a worse course
have the potential to reduce the excess burden of depression among African Americans.
Second, many studies that investigated depression among African Americans only
considered individuals who meet criteria for clinical depression. However, evidence suggests
that many more African Americans experience depressive symptoms than are diagnosed with
clinical depression (Miller et al., 2004). Further, subclinical depression is often the precursor of
more severe depression or representative of a state of remission (Ji, 2012; Juruena, 2012).
Because depressive symptoms, in the absence of clinical depression continue to be a public
health concern that negatively affects the functioning of many African Americans they are
worthy of more research focus.
Third, many existing studies that investigated the course of depression used a race-
comparison approach that aggregates African Americans. This is a limitation because, although
many African Americans share a common socio-historical experience, the population is
heterogeneous, with great variation by region, socioeconomic status, health profile, and many
Hays
104

other factors (Williams & Jackson, 2000). Prior studies of depression that considered within-
group differences have identified subpopulations of African Americans with varying profiles
based on depressive symptoms (Lincoln, Chatters, Taylor, & Jackson, 2007), help-seeking
behavior (Hays & Gilreath, 2016; Woodward, Taylor, & Chatters, 2011), and level of distress
(Lincoln, Taylor, Watkins, & Chatters, 2011). Specifically, a study by Lincoln and associates
(2007) identified two distinct groups of Black Americans based on their depressive symptoms
profile. They concluded that heterogeneity exists in the Black population that can be
characterized by either high or low symptomatology and described by a specific
sociodemographic profile (Lincoln et al., 2007). Findings from this small body of research
highlight the importance of considering within-group heterogeneity among African Americans.
However, existing studies that investigated variable depression trajectories tended to use race-
comparison methods to study depression among African Americans. The current study directly
addressed these three limitations by investigating various trajectories of depressive symptoms in
a national community-based sample of African American adults.
Depression Course
No existing studies have investigated variable depression trajectories in the African
American population. However, depression studies on the general population or race-comparison
studies are useful for understanding the nature of depression during the life course. Richards
(2011) conducted a review of the depression literature, providing an overview of the history and
classification of depression and current knowledge regarding the prevalence and course of the
disease. He suggested that the course of depression involves periods of remission, recovery,
relapse, and recurrence and that although the majority of individuals with depression recover,
Hays
105

this recovery is not necessarily permanent and future episodes can turn into chronic depression
(Richards, 2011).
Several other researchers have investigated racial differences in depression trajectories
by comparing African Americans to other racial and ethnic groups. Liang and associates (2011)
studied racial differences in depression trajectories of White, Black, and Hispanic Americans
aged 50 or older. They identified six distinct depression courses: minimal depressive symptoms,
low depressive symptoms, moderate and stable depressive symptoms, high but decreasing
depressive symptoms, moderate but increasing depressive symptoms and, persistently high
depressive symptoms. In this study, Black and Hispanic Americans were significantly more
likely to be in trajectories marked by higher, increasing, or decreasing depressive symptoms
when compared to Whites (Liang et al., 2011). This pattern remained even when accounting for
differences in socioeconomic status, marital status, and health conditions.
Another investigation by Lincoln and Takeuchi (2010) examined depressive symptoms
by identifying latent trajectory classes of depression among African Americans and non-
Hispanic Whites. The authors attempted to go beyond simple race comparisons by examining
heterogeneity within and between racial groups and by considering the role of social location
factors such as race and how they interact with other factors such as age, marital status, gender,
socioeconomic status, and social relationships to influence mental health trajectories. Their
results revealed four depression trajectories, low symptoms, high symptoms, increasers, and
decreasers (Lincoln & Takeuchi, 2010). They also suggested that African Americans were as
likely as Whites to match any of the four trajectories. They found significant relationships
between trajectory membership and age, education, income, social support, and negative
interaction.
Hays
106

This body of work has revealed important information about depression. First, these
studies highlight the need to examine depressive symptoms during the life course because they
confirmed that the experience of depression is heterogeneous. Second, our understanding of
racial differences in the depression course, or the variable experience of depression among
African American, remains inconclusive because prior studies yielded mixed results and did not
seek to identify divergent depression trajectories among African Americans. No previous studies
investigated variable courses of depression within the African American population. Thus,
important questions remain, Are some African Americans likely to have a worse course of
depression than others? If so, can these subgroups be distinguished by specific demographic,
psychosocial, or sociocultural factors? Uncovering various courses of depression (i.e.,
persistently high, decreasing, or increasing symptoms) will help disaggregate what is known
about depression among African Americans from prior depression research and highlight the
courses that lead to potentially worse outcomes in this population.
This work has direct implications for the development of interventions aimed at
addressing the excess burden of depression among African Americans by pinpointing individuals
who are likely to experience a worse course of depression. For example, if study results suggest
that a subset of individuals (e.g., women younger than 29 with high incomes) experiences a
worse course of depression (i.e., increasing from low to high), interventions can be tailored to the
specific subset of African Americans who fit that demographic and depression profile.
Correlates of Depression
Previous cross-sectional studies have identified several individual, social, and behavioral
correlates of depression among African Americans including demographic characteristics,
psychosocial stress, social relationships, and religion. Characteristics like socioeconomic status,
Hays
107

age, gender, and marital status have been linked to depression (Chatters, Taylor, Woodward, &
Nicklett, 2015; Lincoln, Abdou, & Lloyd, 2014; Lincoln et al., 2011; Salami & Walker, 2014).
Social stress (e.g., financial stress) and negative life events (e.g., violence, trauma) have also
been found contribute to depressive symptoms (Cutrona et al., 2005; Lantz, House, Mero, &
Williams, 2005; Mitchell & Ronzio, 2011; Turner & Avison, 2003). Social relationships have
been identified as both a risk and protective factor for depression in African Americans (Chatters
et al., 2015; Lincoln & Chae, 2012). Additionally, religious involvement has been described as a
protective factor and linked to depression, with studies suggesting that African Americans who
attend religious services more frequently are less likely to experience depression than those who
attended church infrequently (Chatters et al., 2008).
Traditionally, research has investigated relationships between individual correlates (e.g.
stress or religious involvement) and depression instead of studying them concurrently. However,
a few researchers have made the case that some combinations of risk factors are more predictive
of psychopathology than others and that prior research that focused on the additive contributions
of individual risk factors, or interactions among two or three risk factors, cannot adequately
capture the complexity of what individuals face in the real world (Copeland, Shanahan, Costello,
& Angold, 2009). As such, this study employed a risk typology identified in Chapter 3 of this
dissertation that includes distinct constellations of risk and protective factors experienced by
African Americans.
Theoretical Background
The Cumulative Advantage/Disadvantage (CAD) model suggests that early life
advantages and disadvantages persist, or accumulate, into later life and are often magnified
during the life course (Crystal, Shea, & Reyes, 2016). This framework may help shed light on the
Hays
108

differential experience of depression in African Americans. However, studies are needed to
evaluate CAD theory’s applicability to depression among African American by investigating
various courses of depressive symptoms during the adult lifespan for African Americans, some
of which may be characterized by persistent or chronic symptomatology and related to negative
life conditions. This information will highlight individuals who are most likely to experience a
worse course of depression and according to CAD theory, are likely to suffer an excess burden of
depression over time. The conceptual model that guided this analysis incorporates aspects of
CAD theory to suggest that certain demographic factors and risk and protective types predict the
course of depression (see Figure 1).
Methods
Data
Five waves of data from the Americans’ Changing Lives Study were used in this study to
examine variable courses, or trajectories, of depressive symptoms in African Americans. This
ongoing nationally representative study focuses on differences between Black and White
Americans in middle and late life. The study features a range of sociological, psychological,
mental, and physical health items. Wave 1 of the study began in 1986 with face-to-face
interviews of 3,617 adults aged 25 or older. African Americans and people older than 60 were
oversampled. Data from Wave 2 (N = 2,867) were collected in 1989 via follow-up face-to-face
and telephone interviews with individuals who participated in Wave 1. During Wave 3 in 1994
(N= 2,562), Wave 4 in 2001 and 2002 (N = 1,787), and Wave 5 in 2011 (N = 1,427), participants
were re-interviewed by telephone, or when available, face to face. The average response rate for
these five waves was 79%. The analytic sample for this analysis features respondents who
identified as African American (N = 1,173).
Measures
Hays
109

Depressive symptoms. Depressive symptoms were measured with 11 items from the Center for
Epidemiological Studies Depression Scale (Radloff, 1977). Items measured the extent to which
respondents felt happy, lonely, sad, that everything was an effort, that their sleep was restless,
that people were unfriendly, that they did not feel like eating, that people dislike them, that they
could not get going, and that they enjoyed life. Responses were reported on a Likert-type scale (1
= hardly ever, 2 = some of the time, and 3 = most of the time). Positively worded items (i.e. felt
happy and enjoyed life) were reverse coded. Scores from the 11 items were averaged and higher
mean scores indicated more depressive symptoms. This abbreviated scale has been found to have
acceptable reliability (alpha=0.86) and a similar factor structure compared to the original 20-item
version and has been validated with African American samples (Foley, Reed, Mutran, &
DeVellis, 2002).
Demographic characteristics. The demographic variables used in this analysis included age,
gender, marital status, and education level at baseline (Wave 1). Age was a continuous measure
that was recoded into a categorical variable (25-39, 40-59, and 60 or older). Marital status was
categorized as married, separated, divorced, widowed, and never married. Education level was
recoded into a single variable with categories of below high school, high school education, some
college, and college degree.
Risk and protective types. Distinct types (or classes) of constellations of risk and protective
factors were identified in a previous latent class analysis (see Chapter 3). These types are
described as constellations of sociocultural and psychosocial factors at either high or low levels.
Specifically, respondents in the sample were asked about the degree to which they experienced
chronic financial stress, negative life events, social support, negative social interaction,
organizational religiosity, and subjective religiosity. The six contributing factors were recoded
Hays
110

into binary measures (0 = low and 1= high) (see Chapter 3). Latent class analysis results revealed
two distinct classes of risk and protection: (a) Low Protection/Low Risk and (b) High
Protection/Low Risk. Individuals in the High Protection/Low Risk type generally had low risk
factors (i.e., negative life events, financial stress, and negative social interaction) and high
protective factors (i.e., organizational and subjective religiosity, social support). Those in the
Low Protection/Low Risk type also had relatively low experience of risk factors, like financial
stress and negative life events; however, they were less likely to endorse experiencing protective
factors like religiosity and social support. For this current analysis, a risk variable was created
that included each respondent’s most likely risk type (1 = High Protective/Low Risk, 2 = Low
Protective/Low Risk).
Missing data. Maximum likelihood estimation was used to account for missing data in complex
samples with the ESTIMATOR = MLR command which is the recommended way of handling
missing data in Mplus (Muthén & Muthén, 2011). The covariance convergence described the
extent of missing data at each wave. The coverage range for CESD score for waves 1 through 5
were 1.0 to 0.32.
Analysis
Growth mixture modeling was employed to identify latent trajectory classes of
depression in the sample during a 25-year period. This is an appropriate modeling strategy
because of its ability to describe longitudinal change among unobserved subpopulations of
individuals (Ram & Grimm, 2009). In other words, a priori knowledge of, or identification of,
subgroups of individuals is not necessary which is arguably a more appropriate approach to
understanding phenomena among understudied populations like African Americans. Mplus 7
Hays
111

software was used for growth mixture modeling and SAS 9.3 was used for multinomial logistic
regression analysis.
Analysis proceeded in five steps similar to those described by Lincoln and Takeuchi
(2010). First, model testing began by comparing linear and quadratic models for depressive
symptoms at all five waves. Results indicated that a linear model fit the data better than a
quadratic model. Next, a series of linear growth mixture models were estimated with one to six
classes with intercept and slope starting values fixed to 0. The difference in time between waves
was accounted for using syntax to indicate unequal distances between time points. Second, fit
statistics were consulted to identify the number of trajectories needed to describe the data.
Specifically, the Akaike information criterion (AIC), Bayesian information criterion (BIC),
parametric bootstrapped likelihood ratio test (BLRT) p-value, entropy, and model interpretability
were considered (Akaike, 1974; Schwarz, 1978). Lower values of AIC and BIC, a significant
BLRT p-value, and higher entropy indicated improved model fit (see Table 2). Third, the best-
fitting model with the least number of classes was selected and a variable representing each
individual’s most likely class assignment was created and exported into SAS (see Jung &
Wickrama, 2008). Fourth, the mean depressive symptoms scores for each latent trajectory at each
time point were calculated, compared, and then plotted to provide a description of each
depression trajectory. This plot guided the interpretation and naming of the trajectories. Finally,
multinomial logistic regression was conducted to identify relationships among demographic
characteristics, pre-established risk types at baseline, and depression courses.
Results
The baseline demographic characteristics of the overall sample (N = 1,173) are described
in Table 1. Women comprised the majority of the sample (66%). Respondents aged 60 or older
represented a large proportion of the sample (43%). Overall, the respondents had low levels of
Hays
112

educational attainment, with more than half (51%) reporting less than a high school education.
Concerning marital status, many of the respondents were either married (39%) or widowed
(21%). Three quarters of sample participants (77%) were classified in the High Protective/Low
Risk type, which features high levels of religious involvement and social support and low levels
of stress and negative social interaction. The remaining 23% of respondents belonged to the Low
Protection/Low Risk type, characterized by low social support and religious involvement and
low stress and negative interaction.
After careful review of the various fit indexes (Table 2) and a plot diagram describing
depression trajectories (Figure 2), the 5-class solution was deemed to have the best overall fit to
the data. In other words, five distinct depression courses existed for African Americans in the
sample. The first course included the majority of respondents (n = 662); these individuals had
persistently low symptoms across all five waves of data and this class was termed the Low
Symptom trajectory. Few respondents (n = 40) were included in the High Symptom trajectory
characterized by persistently high depressive symptoms over time. The third trajectory was
named Increasers because individuals in this class experienced a steady increase in depressive
symptoms (n = 102). Those in the third depression course were described as Slow Decliners
because they had relatively high depressive symptoms that decreased moderately over time (n =
133). The second largest class (n = 233) was the Fast Decliners trajectory, which included
individuals with moderate depressive symptoms that decreased steadily before leveling off. The
estimated growth factor means for the five-class solution are shown in Table 3, which describes
the intercept and slope of each trajectory class. Estimated average probabilities of class
membership for each trajectory represent the probability of respondents to be in each of the five
classes. The average across-class probabilities were .86 for Low Symptoms, .78 for High
Hays
113

Symptoms, .69 for Increasers, .62 for Slow Decliners, and .61 for Fast Decliners, suggesting
acceptable model fit.
Multinomial logistic regression results suggest that several baseline demographic
variables predicted depression trajectory (see Table 4). First, compared to men, women were
more likely to be in the Slow Decliners (OR: 1.70; CI: 1.09, 2.68) or Fast Decliners (OR: 1.49;
CI: 1.06, 2.09) groups than the referent Low Symptom trajectory. Age was also a significant
predictor of class membership. Adults aged 40-59 were less likely than those aged 25-39 to be in
the Increasers (OR: 0.41; CI: 0.22, 0.75) and Slow Decliners (OR: 0.50; CI: 0.29, 0.87) classes
compared to the Low Symptoms class. Adults aged 60 or older were less likely than those aged
25-39 to be in each of the four trajectories marked by elevated depressive symptoms.
Educational attainment was also a predictor of depression trajectory. Those with a high school
education were significantly less likely than those with less than a high school education to be in
the High Symptoms (OR: 0.37; CI: 0.15, 0.92), Slow Decliners (OR: 0.41; CI: 0.24, 0.69, and
Fast Decliners (OR: 0.56; CI: 0.37, 0.85) trajectories compared to the referent Low Symptoms
trajectory. Similarly, those with some college education were less likely than those with less than
a high school education to be in each of the four trajectories characterized by increased
depressive symptoms. Respondents with a college degree were less likely than those with less
than a high school education to be in the Increasers (OR: 0.20; CI: 0.06, 0.68), Slow Decliners
(OR: 0.11; CI: 0.03, 0.36), and Fast Decliners (OR: 0.40; CI: 0.21, 0.76) classes. Additionally,
marital status predicted depression course membership. Separated individuals were more likely
than those who were married to be among each of the four trajectories with higher depressive
symptoms. Respondents who were divorced at baseline were more likely than their married
counterparts to be in the Fast Decliners group (OR: 1.70; CI: 1.05, 2.74). Widowhood was not a
Hays
114

significant predictor of class membership. However, those who were never married were more
likely that married individuals to be in the High Symptoms (OR: 3.34; CI: 1.33, 8.37) Slow
Decliners (OR: 2.86, CI: 1.62, 5.05), and Fast Decliners (OR: 2.50, CI: 1.57, 3.97) groups. Table
5 presents the overall demographic characteristics of each depression trajectory.
Finally, multinomial logistic regression was employed to assess relationships between
established types of risk and protective factors and depression course. Members of the Low
Protection/Low Risk Type were described as having low religious involvement, relatively low
social support, and low stress. These individuals were significantly more likely than those in the
High Protection/Low Risk group (with high religiosity, high social support, and low stress) to be
in the High Symptoms trajectory compared to the Low Symptoms trajectory (OR: 2.58; CI: 1.33,
4.99; see Table 6).
Discussion
This study examined variable courses of depressive symptoms among African American
adults during a 25-year period. Prior studies of depression among African Americans often used
cross-sectional data to understand the experience of depression at a single point in time or used
race-comparative methods to identify racial differences in the course of depression. Results of
this study have extended what is known about depression in African Americans by using a
within-group approach to uncover heterogeneity in the course of depressive symptoms without
race comparisons. This allowed for the identification of subgroups of African Americans with
distinct depression experiences over time. Further, this study suggests that specific individual
characteristics (i.e. age, gender, marital status, and education) and life conditions (i.e.
combinations of risk and protective factors) predict the course of depression for African
Americans. The four major findings that have emerged from this study will be discussed.
Depression Trajectories
Hays
115

First, results revealed five distinct depression trajectories during a 25-year period: 1) Low
Symptoms marked by few depressive symptoms, 2) High Symptoms featuring consistently high
depressive symptoms, 3) Increasers in which depressive symptoms steadily increased over time,
4) Slow Decliners featuring relatively high depressive symptoms that declined moderately over
time, and 5) Fast Decliners involving relatively high depressive symptoms that declined quickly
then leveled off. More than half of respondents in the sample were in the Low Symptom
depression trajectory (56%). However, the other 44% of respondents were in a depression
trajectory marked by elevated depressive symptoms at some point in the 25-year study period.
This is important to note in light of prevalence studies suggesting that only 5% of African
Americans experience depression (Substance Abuse and Mental Health Services Administration,
2015). Findings from this study suggest that a large proportion of African Americans suffer from
elevated symptoms at some point during their life, and although these symptoms may be sub-
clinical, they might result in impairment in daily functioning (Juruena, 2012). This study also
identified a fifth depression trajectory not identified in prior studies of depression course: the
Fast Decliners. More than 20% of respondents in this sample had a depression course marked by
relatively high symptoms that decreased rapidly before leveling off. This finding points to the
benefit of using a within-group approach to studying depression among African Americans
because the variable courses of depression in this population may differ from other racial and
ethnic groups in a way that race-comparison studies cannot capture.
The two trajectories of concern are High Symptom and Increasers because individuals in
these classes are likely to maintain or develop higher depressive symptoms over time. The High
Symptom group had persistently elevated depressive symptoms that seemed to decrease only
slightly from Wave 1 to Wave 2 (1989). Overall, however, individuals in this group did not
Hays
116

experience a period of remission or recovery from their symptoms during the measured
timepoints and their depressive symptoms should be considered chronic and long lasting. The
Increasers class is concerning because these individuals experienced a steady and gradual rise in
depressive symptoms over time. Past studies that considered African Americans in the aggregate
have not been able to capture individuals in these two divergent trajectories who may be more
likely to experience higher depressive symptoms because of the nature of their depression
course. This suggests that African Americans as a whole might not have poor depression
outcomes as the literature would suggest, but rather, a subset of individuals with chronic and
persistent symptoms suffers disproportionately. Individuals who fit to these depression
trajectories should be the focus of increased attention through research and intervention
development.
Demographic Correlates
Second, specific individual demographic characteristics are predictive of depression
course. First, women were more likely to be in the Slow Decliners and Fast Decliners groups
which means they were likely to experience higher depressive symptoms earlier in the 25-year
period that decreased over time. It may be the case that as African American women progress
through the life-course they become less likely to experience stressors related to childbearing and
establishing relationships and careers. It also may be that as African American women move
through life, they develop more support, coping skills, and resources that reduce their depressive
symptoms over time. In fact, research has suggested that social support has a moderating effect
on the course of depression among women, with smaller social networks being predictive of a
worse depression course (Byers et al., 2012). Studies have also suggested that African American
women tend to rely on their social networks to manage mental and emotional problems (Sosulski
Hays
117

& Woodward, 2013), which may help to explain why women in this sample were likely to
experience a reduction in depressive symptoms over time.
Age was also predictive of depression trajectory; individuals who were between the ages
of 25 and 39 at baseline were more likely than older respondents (60+) to be in all of the courses
marked by elevated depressive symptoms (i.e., High Symptoms, Increasers, Fast Decliners, and
Slow Decliners) compared to the Low Symptoms trajectory. In addition, adults aged 40-59 were
less likely than younger respondents to be among the Increasers and Slow Decliners. The
literature on age patterns of depression suggests that symptoms tend to be elevated during
adolescent years, followed by a stage of relatively low symptoms in early to late midlife, and
then a steady increase in symptoms into late life (Adkins, Wang, Dupre, van den Oord, & Elder,
2009; Kessler, Foster, Webster, & House, 1992; Yang & Lee., 2009). However, young African
Americans in this sample were likely to experience depression trajectories that diverge from this
general pattern as many had relatively high depressive symptoms in their early to midlife
followed by either persistent or decreasing symptoms.
This finding suggests the need for depression interventions for African Americans in
early adulthood because they may be currently experiencing depressive symptoms or are likely
to develop them at some point in their lives. Other researchers have noted that younger African
Americans, particularly young men, who experience depressive symptoms are likely to
underutilize mental health services due to stigma; however, social support from family plays a
pivotal role in buffering young African American men from depressive symptoms (Lindsey, Joe,
& Nebbitt, 2010). It is important to reach out to individuals in this demographic subgroup
through schools, parenting groups, churches, or other institutions to provide them, and their
Hays
118

families, with education, resources, and coping skills to reduce current symptoms or prevent the
future development of depression.
Educational attainment also predicted depression trajectory. In general, participants with
lower education at baseline (i.e. less than high school) were more likely than their more educated
counterparts to be in any of the trajectories marked by some period of elevated symptoms than in
the Low Symptom course. For some respondents, depressive symptoms increased over time,
whereas they decreased for others. Results from prior cross-sectional can help explicate these
findings. Prior research suggested that higher educational attainment is protective against
depressive symptoms (Hybels, Blazer, Pieper, Landerman, & Steffens, 2009). Individuals with
low education at baseline who were able to advance their educational status over time might
have experienced decreasing symptoms, whereas those who remained uneducated might have
experienced increasing symptoms. CAD theory suggests that higher educational attainment early
in life generates certain advantages such as better career options and higher incomes which are
likely to accumulate during the life course. These advantages and the life conditions they
engender are likely to protect against depressive symptoms.
However, an interesting point of digression from theory and prior research is the finding
that individuals with a college degree were no less likely to be in the High Symptoms trajectory
than respondents with less than a high school education. Based on CAD theory and the general
trend in the literature, a college degree would be expected to protect against higher depressive
symptoms, yet this study found no statistically significant relationship. Other researchers have
suggested that a paradoxical relationship exists between college education and health for African
Americans in that education does not always protect against worse health outcomes (Williams,
Mohammed, Leavell, & Collins, 2010). An inverse relationship between educational attainment
Hays
119

and improved health has been found among African American men, wherein those with a college
education have worse health then those with only some college experience (Williams, 2003).
Researchers have suggested that African Americans experience “diminishing returns” and
college education or higher socioeconomic status do not provide the same kinds of advantages to
health seen in White populations (Farmer & Ferraro, 2005). This might help explain the current
study’s findings that high school and some college education are protective against worse
depression but a college degree has no such relationship.
Marital status was also associated with depressive symptom trajectory. Respondents who
were married at baseline were less likely to have high depressive symptoms or experience
persistently high symptoms than those who were separated, divorced, or who had never been
married (at baseline). This is consistent with other studies suggesting that marriage protects
against depression and increases well-being (Lapierre, 2009). However, prior research has
suggested that marriage alone is not always protective, but rather that its positive effect on health
outcomes is likely a result of the increased social networks gained through marriage
(Schwarzbach, Luppa, Forstmeier, König, & Riedel-Heller, 2014). Although it is possible that
respondents’ marital status may have changed during the course of the 25-year study period (e.g.
separation or divorce), individuals who were married may have experienced sustained benefits of
increased social networks via marriage that they developed earlier in their lives.
Risk and Protection
Third, this study’s findings highlight risk and protective types linked to depression. It is
important to note that the risk types used in this study are primarily based on protective factors
instead of risk factors. Both of the identified risk types (Low Protective/Low Risk and High
Protective/Low Risk) were characterized by relatively low experiences of risk factors (i.e.
Hays
120

negative social interaction, negative life events, and financial stress). However, the experience of
protective factors (i.e. religiosity and social support) was markedly different in these two groups
and predictive of risk type.
Not surprisingly, individuals who were classified as having low protective factors were
more than twice as likely to be in the High Symptoms trajectory compared to those with high
protective factors. CAD theory suggests that having lower protective factors early in life serves
as a disadvantage that accumulates over time and it is this accumulation of disadvantages that
ultimately leads to worse health. In this study, low protective factors at baseline were linked to
high depressive symptoms early in life and these symptoms persisted over time. This result
provides valuable information that can be used to develop prevention programs for depression
among African Americans. Specifically, interventions should focus on promoting positive social
interactions through family-based programming or establish partnerships with clergy to find
ways to reach out to African Americans who could benefit from increased social and church-
based supports. These kinds of strategies may help to move individuals from a Low
Protective/Low Risk class into a High Protective/Low Risk group, thereby reducing their
likelihood of experiencing chronically high depressive symptoms.
Hays
121

Figure 4.1. Conceptual Model of depression trajectories in African American
adults, ages 25 and older
Hays
122

Table 4.1. Overall Demographic and Risk Type Distributions (N=1173)
Demographic & Risk Type Characteristics
% n
Sex
Men 33.67 395
Women 66.33 778
Age
25-39 31.60 370
40-59 25.88 303
60+ 42.53 498
Marital Status
Married 39.13 459
Separated 9.97 117
Divorced 13.04 153
Widowed 21.40 251
Never Married 16.45 193
Education
Below High School 51.32 602
High School 24.13 283
Some College 16.88 198
College Degree 7.67 90
Risk Type
High Protective/Low Risk 76.90
902
Low Protective/Low Risk 23.10
271

Hays
123

Table 4.2. Fit Statistics for Growth Mixture Models with 1 to 7 Latent Classes
Model Description
Risk Types for Depression
AIC BIC Entropy BLMRT
1 One-class 2467.02 2517.70
2 Two-class 2363.93 2429.81 .63 .000
3 Three-class 2342.97 2408.84 .71 .000
4 Four-class 2283.48 2364.56 .66 .000
5 Five-class 2264.88 2361.16 .63 .000
6 Six-class 2255.60 2367.08 .67 .000
7 Seven-class 2253.08 2379.76 .69 .000
Notes:
Bold text indicates selected model
AIC = Akaike information criterion
BIC = Bayesian information criterion
BLMRT = parametric bootstrapped likelihood ration test

Hays
124

Figure 4.2. Growth Curves for 5 Trajectory Classes of Depressive Symptoms for African Americans

1
1.2
1.4
1.6
1.8
2
2.2
2.4
1986 1991 1996 2001 2006 2011
CESD Mean
Low Symptom (n=662)
High Symptom (n=40)
Increasers (n=102)
Slow Decliners (n=133)
Fast Decliners (n=236)
Hays
125

Table 4.3. Estimated Growth Factor Means
Class Mean SE Est/SE
Low Symptoms (N=662)
Intercept 1.28 0.01 96.84***
Slope -0.01 0.01 -1.98*
High Symptoms (N=40)
Intercept 2.17 0.09 25.52***
Slope 0.02 0.03 0.46
Increasers (N=102)
Intercept 1.48 0.08 19.27***
Slope 0.13 0.04 3.61***
Slow Decliners (N=133)
Intercept 1.97 0.09 21.74***
Slope -0.07 0.03 -2.10*
Fast Decliners (N=236)
Intercept 1.75 0.05 38.36***
Slope -0.10 0.02 -5.26***
Note: *p<0.05. **p<0.01. ***p<.001.

Hays
126

Table 4.4. Multinomial Logistic Regression of Demographic Covariates on Depression Trajectories

Baseline Covariates
Depression Trajectories
High Symptoms
OR (95% CI)
Increasers
OR (95% CI)
Slow Decliners
OR (95% CI)
Fast Decliners
OR (95% CI)
Female 1.59 (0.76,3.35) 1.27 (0.78, 2.05) 1.70 (1.09, 2.68) 1.49 (1.06, 2.09)
40-59
a
0.41 (0.16, 1.04) 0.41 (0.22, 0.75) 0.50 (0.29, 0.87) 0.97 (0.63, 1.50)
60+
a
0.32 (0.14, 0.91) 0.32 (0.17, 0.59) 0.30 (0.17, 0.54) 0.56 (0.35, 0.90)
Separated
b
5.24 (1.89, 14.51) 3.38 (1.71, 6.68) 3.38 (1.83, 5.55) 3.19 (1.83, 5.55)
Divorced
b
1.67 (0.55, 5.08) 0.68 (0.30, 1.52) 1.46 (0.75, 2.86) 1.70 (1.05, 2.74)
Widowed
b
0.91 (0.28, 2.98) 1.16 (0.60, 2.26) 1.68 (0.92, 3.09) 1.34 (0.83, 2.15)
Never Married 3.34 (1.33, 8.37) 0.70 (0.34, 1.48) 2.86 (1.62, 5.05) 2.50 (1.57, 3.97)
High School
c
0.37 (0.15, 0.92) 0.90 (0.52, 1.54) 0.41 (0.24, 0.69) 0.56 (0.37, 0.85)
Some College
c
0.31 (0.11, 0.85) 0.31 (0.15, 0.67) 0.26 (0.14, 0.50) 0.42 (0.26, 0.69)
College Degree
c
0.36 (0.10, 1.32) 0.20 (0.06, 0.68) 0.11 (0.03, 0.36) 0.40 (0.21, 0.76)
Note: Trajectories assessed relative to Low Symptoms trajectory. Figures in bold are significant at p< .05
a
= Reference class is 25-39 years old
b
= Reference class is married
c
= Reference class is below high school

Hays
127

Table 4.5. Risk Types and Demographic Characteristics of Depression Trajectories
Low Symptoms
(n=662)
n (%)
High Symptoms
(n=40)
n (%)
Increasers
(n=102)
n (%)
Slow Decliners
(n=133)
n (%)
Fast Decliners
(n=236)
n (%)
Female 408 (61.63) 29 (72.5) 72 (70.59) 100 (75.19) 169 (71.61)
Male 254 (38.37) 11 (27.5) 30 (29.41) 33 (24.81) 67 (28.39)
25-39 181 (27.38) 19 (47.5) 42 (41.58) 54 (40.60) 74 (31.36)
40-59 169 (25.57) 8 (20) 22 (21.78) 31 (23.31) 73 (30.93)
60+ 311 (47.05) 13 (32.5) 37 (36.63) 48 (36.09) 89 (37.71)
Married 305 (46.07) 10 (25) 41 (40.20) 34 (25.56) 69 (29.24)
Separated 37 (5.59) 8 (20) 21 (20.59) 19 (14.29) 32 (13.56)
Divorced 89 (13.44) 5 (12.5) 8 (7.84) 15 (11.28) 36 (15.25)
Widowed 147 (22.21) 5 (12.5) 21 (20.59) 30 (22.56) 48 (20.34)
Never Married 84 (12.69) 12 (30) 11 (10.78) 35 (26.32) 51 (21.61)
Below High School 308 (46.53) 23 (57.5) 52 (50.95) 84 (63.16) 135 (57.20)
High School 157 (23.72) 8 (20) 36 (35.29) 29 (21.80) 53 (22.46)
Some College 131 (19.79) 6 (15) 11 (10.78) 17 (12.78) 33 (13.98)
College Degree 66 (9.97) 3 (7.5) 3 (2.94) 3 (2.26) 15 (6.36)
High
Protective/Low
Risk
526 (79.46) 24 (60) 80 (78.43) 96 (72.18) 176 (74.58)
Low
Protective/Low
Risk
136 (20.54) 16 (40) 22 (21.57) 37 (27.82) 60 (25.42)
Hays
128

Table 4.6. Multinomial Logistic Regression of Risk and Protective Types on Depression Trajectories
Types Depression Trajectories
High Symptoms
OR (95% CI)
Increasers
OR (95% CI)
Slow Decliners
OR (95% CI)
Fast Decliners
OR (95% CI)
Low Protection/Low Risk
vs.
High Protection/Low Risk
2.58 (1.33, 4.99) 1.06 (0.64, 1.77) 1.49 (0.98, 2.28) 1.32 (0.93, 1.87)
Note: Trajectories assessed relative to Low Symptoms trajectory. Figures in bold are significant at p< .05.

Hays
129

References
Adkins, D. E., Wang, V., Dupre, M. E., van den Oord, E. J. G. C., & Elder, G. H. (2009).
Structure and stress: Trajectories of depressive symptoms across adolescence and young
adult- hood.”. Social Forces, 88(1), 39–60. http://doi.org/10.1353/sof.0.0238
Akaike, H. (1974). A new look at the statistical model identification. IEEE Transactions on
Automatic Control, 19(6), 716–723. http://doi.org/10.1109/TAC.1974.1100705
Blau, P. M., & Duncan, O. D. (1967). The American occupational structure. New York, NY:
Free Press.
Breslau, J., Aguilar-Gaxiola, S., Kendler, K. S., Su, M., Williams, D. R., & Kessler, R. C.
(2006). Specifying race-ethnic differences in risk for psychiatric disorder in a USA
national sample. Psychological Medicine, 36(1), 57–68.
http://doi.org/10.1017/S0033291705006161
Bromberger, J. T., Harlow, S., Avis, N., Kravitz, H. M., & Cordal, A. (2004). Racial/ethnic
differences in the prevalence of depressive symptoms among middle-aged women: The
Study of Women’s Health Across the Nation (SWAN). American Journal of Public
Health, 94(8), 1378–1385. http://doi.org/10.2105/AJPH.94.8.1378
Byers, A. L., Vittinghoff, E., Lui, L.-Y., Hoang, T., Blazer, D. G., Covinsky, K. E., … Yaffe, K.
(2012). Twenty-year depressive trajectories among older women. Archives of General
Psychiatry, 69(10), 1073–1079. http://doi.org/10.1001/archgenpsychiatry.2012.43
Chatters, L. M., Bullard, K. M., Taylor, R. J., Woodward, A. T., Neighbors, H. W., & Jackson, J.
S. (2008). Religious participation and DSM IV disorders among older African
Americans: Findings from the National Survey of American Life. American Journal of
Geriatric Psychiatry, 16(12), 957–965. http://doi.org/10.1097/JGP.0b013e3181898081.
Chatters, L. M., Taylor, R. J., Woodward, A. T., & Nicklett, E. J. (2015). Social support from
Hays
130

church and family members and depressive symptoms among older African Americans.
American Journal of Geriatric Psychiatry, 23(6), 559–567.
http://doi.org/10.1016/j.jagp.2014.04.008
Colton, C. W., & Manderscheid, R. W. (2006). Congruencies in increased mortality rates , years
of potential life lost , and causes of death among public mental health clients in eight
states. Preventing Chronic Disease, 3(2), A.
Copeland, W., Shanahan, L., Costello, E. J., & Angold, A. (2009). Configurations of common
childhood psychosocial risk factors. Journal of Child Psychology and Psychiatry, 50(4),
451–459. http://doi.org/10.1111/j.1469-7610.2008.02005.x
Crystal, S., Shea, D. G., & Reyes, A. M. (2016). Cumulative advantage, cumulative
disadvantage, and evolving patterns of late-life inequality. Gerontologist, (Advanced
online publication), 1–11. http://doi.org/10.1093/geront/gnw056
Cutrona, C. E., Russell, D. W., Brown, P. A., Clark, L. A., Hessling, R. M., & Gardner, K. A.
(2005). Neighborhood context, personality, and stressful life events as predictors of
depression among African American women. Journal of Abnormal Psychology, 114(1),
3–15. http://doi.org/10.1037/0021-843X.114.1.3
DiPrete, T. A., & Eirich, G. M. (2006). Cumulative advantage as a mechanism for inequality: A
review of theoretical and empirical developments. Annual Review of Sociology, 32(1),
271–297. http://doi.org/10.1146/annurev.soc.32.061604.123127
Donohue, J., & Pincus, H. (2007). Reducing the societal burden of depression: A review of the
economic costs, quality of care and effects of treatment. Pharmaco Economics, 25(1), 7–
24. http://doi.org/10.2165/00019053-200725010-00003
Druss, B. G., Zhao, L., Von Esenwein, S., Morrato, E. H., & Marcus, S. C. (2011).
Hays
131

Understanding excess mortality in persons with mental illness: 17-year follow up of a
nationally representative US survey. Medical Care, 49(6), 599–604.
http://doi.org/10.1097/MLR.0b013e31820bf86e
Farmer, M. M., & Ferraro, K. F. (2005). Are racial disparities in health conditional on
socioeconomic status? Social Science and Medicine, 60(1), 191–204.
http://doi.org/10.1016/j.socscimed.2004.04.026
Ferraro, K. F., & Kelley-Moore, J. A. (2003). Cumulative disadvantage and health: long-term
consequences of obesity? American Sociological Review, 68(5), 707–729.
http://doi.org/10.2307/1519759
Foley, K. . L., Reed, P. S., Mutran, E. J., & DeVellis, R. F. (2002). Measurement adequacy of the
CES-D among a sample of older African–Americans. Psychiatry Research, 109(1), 61–
69. http://doi.org/10.1016/S0165-1781(01)00360-2
Garbarski, D. (2015). Racial/ethnic disparities in midlife depressive symptoms: The role of
cumulative disadvantage across the life course. Advances in Life Course Research, 23,
67–85. http://doi.org/10.1016/j.alcr.2014.12.006
Hays, K., & Gilreath, T. (2016). Profiles of depression help seeking among Black Americans: A
latent class approach. Journal of Nervous and Mental Disease, Advanced online
publication. http://doi.org/10.1097/NMD.0000000000000575
Hirschfeld, R. M. A., Montgomery, S. A., Keller, M. B., Kasper, S., Schatzberg, A. F., Möller,
H. J., … Bourgeois, M. (2000). Social functioning in depression: A review. Journal of
Clinical Psychiatry, 61(4), 268–275. http://doi.org/10.4088/JCP.v61n0405
Hybels, C. F., Blazer, D. G., Pieper, C. F., Landerman, L. R., & Steffens, D. C. (2009). Profiles
of depressive symptoms in older adults diagnosed with major depression: A latent cluster
Hays
132

analysis. American Journal of Geriatric Psychiatry, 17(5), 387–396.
http://doi.org/10.1097/JGP.0b013e31819431ff
Jackson, J. S., Knight, K., & Rafferty, J. A. (2010). Race and unhealthy behaviors: Chronic
stress, the HPA axis, and physical and mental health disparities over the life course.
American Journal of Public Health, 100(5), 933–939.
http://doi.org/10.2105/AJPH.2008.143446
Ji, J. (2012). Distinguishing subclinical (subthreshold) depression from the residual symptoms of
major depression. Shanghai Archives of Psychiatry, 24(5), 288–289.
http://doi.org/10.3969/j.issn.1002-0829.2012.05.007
Jung, T., & Wickrama, A. (2008). An introduction to latent class growth analysis and growth
mixture modeling. Social and Personality Psychology Compass, 2(1), 302–317.
http://doi.org/10.1111/j.1751-9004.2007.00054.x
Juruena, M. F. (2012). Understanding subthreshold depression. Shanghai Archives of Psychiatry,
24(5), 292–293. http://doi.org/10.3969/j.issn.1002-0829.2012.05.009
Kessler, R. C., Foster, C., Webster, P. S., & House, J. S. (1992). The relationship between age
and depressive symptoms in two national surveys. Psychology and Aging, 7, 119–126.
http://doi.org/10.1037/0882-7974.7.1.119
Lantz, P. M., House, J. S., Mero, R. P., & Williams, D. R. (2005). Stress, life events, and
socioeconomic disparities in health: Results from the Americans’ Changing Lives study.
Journal of Health and Social Behavior, 46(3), 274–288.
http://doi.org/10.1177/002214650504600305
Lapierre, T. A. (2009). Marital status and depressive symptoms over time: Age and gender
variations. Family Relations, 58(4), 404–416. http://doi.org/10.1111/j.1741-
Hays
133

3729.2009.00562.x
Liang, J., Xu, X., Quiñones, A. R., Bennett, J. M., & Ye, W. (2011). Multiple trajectories of
depressive symptoms in middle and late life: Racial/ethnic variations. Psychology and
Aging, 26(4), 761–777. http://doi.org/10.1037/a0023945
Lincoln, K. D., Abdou, C. M., & Lloyd, D. A. (2014). Race and socioeconomic differences in
obesity and depression among Black and non-Hispanic White Americans. Journal of
Health Care for the Poor and Underserved, 25, 257–275.
Lincoln, K. D., & Chae, D. H. (2012). Emotional support, negative interaction and major
depressive disorder among African Americans and Caribbean Blacks: Findings from the
National Survey of American Life. Social Psychiatry and Psychiatric Epidemiology,
47(3), 361–372. http://doi.org/10.1007/s00127-011-0347-y
Lincoln, K. D., Chatters, L. M., Taylor, R. J., & Jackson, J. S. (2007). Profiles of depressive
symptoms among African Americans and Caribbean Blacks. Social Science & Medicine,
65(2), 200–213. http://doi.org/10.1016/j.socscimed.2007.02.038
Lincoln, K. D., & Takeuchi, D. T. (2010). Variation in the trajectories of depressive symptoms:
results from the Americans’ Changing Lives study. Biodemography and Social Biology,
56(1), 24–41. http://doi.org/10.1080/19485561003709180
Lincoln, K. D., Taylor, R. J., Watkins, D. C., & Chatters, L. M. (2011). Correlates of
psychological distress and major depressive disorder among African American men.
Research on Social Work Practice, 21(3), 278–288.
http://doi.org/10.1177/1049731510386122
Lindsey, M. A., Joe, S., & Nebbitt, V. (2010). Family matters: The role of mental health stigma
and social support on depressive symptoms and subsequent help seeking among African
Hays
134

American boys. Journal of Black Psychology, 36(4), 458–482.
http://doi.org/10.1177/0095798409355796
Miller, D. K., Malmstrom, T. K., Joshi, S., Andresen, E. M., Morley, J. E., & Wolinsky, F. D.
(2004). Clinically relevant levels of depressive symptoms in community-dwelling
middle-aged African Americans. Journal of the American Geriatrics Society, 52(5), 741–
748. http://doi.org/10.1111/j.1532-5415.2004.52211.x
Mitchell, S. J., & Ronzio, C. R. (2011). Violence and other stressful life events as triggers of
depression and anxiety: What psychosocial resources protect African American mothers?
Maternal and Child Health Journal, 15(8), 1272–1281. http://doi.org/10.1007/s10995-
010-0668-6
Musliner, K. L., Munk-Olsen, T., Eaton, W. W., & Zandi, P. P. (2016). Heterogeneity in long-
term trajectories of depressive symptoms: Patterns, predictors and outcomes. Journal of
Affective Disorders, 199–211. http://doi.org/10.1016/j.jad.2015.12.030
Muthén, L. K., & Muthén, B. O. (2011). Mplus user’s guide (6th ed.). Los Angeles, CA: Muthén
& Muthén.
Noël, P. H., Williams, J.W, J., Unutzer, J., Worchel, J., Lee, S., Cornell, J., … Hunkeler, E.
(2004). Depression and comorbid illness in elderly primary care patients: Impact on
multiple domains of health status and well-being. Annals of Family Medicine, 2, 555–
562. http://doi.org/10.1370/afm.143.
Pickett, Y. R., Bazelais, K. N., & Bruce, M. L. (2013). Late-life depression in older African
Americans: A comprehensive review of epidemiological and clinical data. International
Journal of Geriatric Psychiatry, 28(6), 903–913. http://doi.org/10.1002/gps.3908
Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general
Hays
135

population. Applied Psychological Measurement, 1(3), 385–401.
http://doi.org/10.1177/014662167700100306
Ram, N., & Grimm, K. J. (2009). Growth mixture modeling: A method for identifying
differences in longitudinal change among unobserved groups. International Journal of
Behavioural Development, 33(6), 565–576.
http://doi.org/10.1177/0165025409343765.Growth
Richards, D. (2011). Prevalence and clinical course of depression: A review. Clinical Psychology
Review. http://doi.org/10.1016/j.cpr.2011.07.004
Romero, L. J., Ortiz, I. E., Finley, M., Wayne, S., & Lindeman, R. D. (2005). Prevalence of
depressive symptoms in New Mexico Hispanic and non-Hispanic White elderly.
Ethnicity & Disease, 15(4), 691–697.
Salami, T. K., & Walker, R. L. (2014). Socioeconomic status and symptoms of depression and
anxiety in African American college students: The mediating role of hopelessness.
Journal of Black Psychology, 40(3), 275–290. http://doi.org/10.1177/0095798413486158
Schwarz, G. (1978). Estimating the dimension of a model. Annals of Statistics, 6(2), 461–464.
http://doi.org/10.1214/aos/1176344136
Schwarzbach, M., Luppa, M., Forstmeier, S., König, H. H., & Riedel-Heller, S. G. (2014). Social
relations and depression in late life: A systematic review. International Journal of
Geriatric Psychiatry, 29(1), 1–21. http://doi.org/10.1002/gps.3971
Sosulski, M. R., & Woodward, A. T. (2013). African American women living with mental
disorders: factors associated with help seeking from professional services and informal
supports. Social Work in Public Health, 28(7), 660–671.
http://doi.org/10.1080/19371918.2011.593462
Hays
136

Spence, N. J., Adkins, D. E., & Dupre, M. E. (2011). Racial differences in depression trajectories
among older women: Socioeconomic, family, and health influences. Journal of Health
and Social Behavior, 52(4), 444–459. http://doi.org/10.1177/0022146511410432
Substance Abuse and Mental Health Services Administration. (2015). Results from the 2015
National Survey on Drug Use and Health: Mental health findings. Rockville, MD.
Sutin, A. R., Terracciano, A., Milaneschi, Y., An, Y., Ferrucci, L., & Zonderman, A. B. (2013).
The trajectory of depressive symptoms across the adult life span. JAMA Psychiatry,
70(8), 803–811. http://doi.org/10.1001/jamapsychiatry.2013.193
Turner, R. J., & Avison, W. R. (2003). Status variations in stress exposure: Implications for the
interpretation of research on race, socioeconomic status, and gender. Journal of Health
and Social Behavior, 44(4), 488–505. http://doi.org/10.2307/1519795
U.S. Department of Health and HumanServices. (2011). 2010 National healthcare disparities
report. AHRQ Publication No.11-0005. Rockville, MD.
Williams, D. R. (2003). The health of men: Structural inequalities and opportunities. American
Journal of Public Health, 93(5), 724–731.
http://doi.org/10.2105/AJPH.98.Supplement_1.S150
Williams, D. R., González, H. M., Neighbors, H., Nesse, R., Abelson, J. M., Sweetman, J., &
Jackson, J. S. (2007). Prevalence and distribution of major depressive disorder in African
Americans, Caribbean blacks, and non-Hispanic whites: Results from the National
Survey of American Life. Archives of General Psychiatry, 64(3), 305–315.
http://doi.org/10.1001/archpsyc.64.3.305
Williams, D. R., & Jackson, J. S. (2000). Race/ethnicity and the 2000 census: Recommendations
for African American and other Black populations in the United States. American
Hays
137

Journal of Public Health, 90(11), 1728–1730. http://doi.org/10.2105/AJPH.90.11.1728
Williams, D. R., Mohammed, S. A., Leavell, J. J., & Collins, C. (2010). Race, socioeconomic
status, and health: complexities, ongoing challenges, and research opportunities. Annals
of the New York Academy of Sciences, 1189, 69–101. http://doi.org/10.1111/j.1749-
6632.2009.05339.x.
Willson, A. E., Shuey, K. M., & Elder, G. H. (2007). Cumulative advantage processes as
mechanisms of inequality in life course health. American Journal of Sociology, 112(6),
1886–1924. http://doi.org/10.1086/512712
Woodward, A. T., Taylor, R. J., & Chatters, L. M. (2011). Use of professional and informal
support by Black men with mental disorders. Research on Social Work Practice, 21(3),
328–336. http://doi.org/10.1177/1049731510388668
Yang, Y., & Lee., L. C. (2009). Sex and race disparities in health: Cohort variations in life
course patterns. Social Forces, 87(4), 2093–2124. http://doi.org/10.1353/sof.0.0183

Hays
138

Chapter Five: Conclusion
The aim of this dissertation was to examine contributors to depression outcomes
experienced by African Americans to inform depression interventions that are targeted and
tailored to this population. Both person-centered and longitudinal methods were employed
using data from a nationally representative sample of African American adults and older
adults. The findings from these studies expand knowledge around the contributors to
depression outcomes which include a variety of psychosocial and sociocultural factors as well
as variations in the course of depressive symptoms. Study results have direct implications for
theory development, future research priorities, and depression interventions. This chapter will
present some of the key findings from the three empirical chapters, describe the limitations of
the studies, and discuss implications.
Summary of Major Findings

Findings from this study have contributed to our understanding of the correlates (e.g.
demographic factors), causes (e.g. risk and protective types), and longitudinal experience (e.g.
variable trajectories) of depression among African Americans. The first empirical study
(Chapter 2) provided insight into unique risk types associated with depression in African
Americans. A risk typology emerged that included three distinct constellations of risk and
protective factors that were predictive of later depressive symptoms (i.e. High Protective/Low
Risk, Moderate Protective/Low Risk, and Low Protective/Low Risk) which characterize
combinations of sociocultural and psychosocial risk.
Concerning depression, African Americans in the Moderate Protective/Low Risk and
Low Protective/Low Risk had significantly higher depressive symptoms than those in the
High Protective/Low Risk type. An analysis of the correlates of risk types revealed that
Hays
139

gender, age, and education level were significantly associated with risk type. Men were more
likely to be in the Moderate Protective/Low Risk and the Low Protective/Low Risk types
compared to women. Respondents aged 20–39 were more than twice as likely to be in the
Moderate Protective/Low Risk class compared to respondents aged 60 or older. Those with
less than a high school education or a high school diploma were 10 times more likely than
college graduates to be in the Moderate Protective/Low Risk type compared to the High
Protective/Low Risk types.
Results of this study contribute to the literature around depression in African
Americans by identifying a typology that illuminates the constellation of risk and protective
factors related to depressive symptoms. This is important because it suggests that when
investigating the influence of psychosocial and sociocultural risk factors on depression
researchers should consider the dynamic nature of the human experience by studying these
factors as a collective instead of singular occurrences. This study also provides further support
for the benefits of within-group analysis because of the potential to identify homogeneous
subpopulations among a heterogeneous population of African Americans.
The second empirical chapter (Chapter 3) was an extension of Chapter 2. This study
used the Cumulative Advantage and Disadvantage theory as a guide which suggests that
specific constellations of risk and protective factors might constitute a path toward depression
outcomes. As such, this study investigated the existence of distinct constellations of risk and
protective factors for depression and examined individual stability, or transitions, between risk
types over time. Latent transition analysis was conducted and results suggested that there were
two risk types, (a) Low Protection/Low Risk and (b) High Protection/Low Risk, and eight
distinct risk paths. Two paths represented stability in risk over the 25-year study period
Hays
140

(Stable High Protection/Low Risk and Stable Low Protection/Low Risk) while the other six
risk paths characterized transitions from one risk type to another at one or more time points.
Risk paths were predictive of depressive symptoms. Individuals who transitioned
between risk types were significantly more likely to have higher depressive symptoms than
those with stable High Protection or Low Protection risk. Also, there were several
demographic predictors of risk transitions. Women were more likely than men to have Stable
High Protection/Low Risk than either Stable Low Protection/Low Risk or Transitioning Risk.
Additionally, married respondents were more likely than those who were never married to
have Stable High Protection/Low Risk than Stable Low Protection/Low Risk.
The third study (Chapter 4) sought to uncover subpopulations of African Americans who
are most likely to suffer excessively from depression as evidenced by more chronic and
persistent symptoms. This study examined heterogeneity in the course of depressive symptoms
among African American adults and older adults and identified variable depression trajectories
and their demographic, psychosocial, and sociocultural correlates. Growth mixture modeling was
employed and results suggested five distinct depression trajectories for African Americans: Low
Symptoms, High Symptoms, Increasers, Slow Decliners, and Fast Decliners. In general, younger
age and lower educational attainment were predictive of a depression course marked by elevated
symptoms while marriage was protective against a worse course of depression. Also, individuals
in the Low Protection/Low Risk type were more than two and a half times more likely to be in
the High Symptoms trajectory than their High Protection/Low Risk counterparts. The results of
this study demonstrated the benefits of using a within-group approach to investigate
heterogeneity within African Americans and in the course of depressive symptoms in this group.
More specifically, this analysis disaggregated African Americans and identified divergent
Hays
141

trajectories of depressive symptoms which is typically obscured in race-comparison methods.
Further, this study used a longitudinal approach to suggest specific depression courses of
concern.
Key Dissertation Themes
The first major theme that emerged from this study involves the influence of collective
risk and protective factors on depression in African Americans. Prior research of depression in
African Americans tended to focus on individual correlates of depression. However, this study
investigated the collective role of various psychosocial and sociocultural factors in
contributing to depression in this population. This approach revealed a risk typology
characterized by distinct classes of religious involvement, social support, negative social
interaction, stress, and negative life events which differentiated subpopulations of African
Americans. These findings support the idea that the human experience is dynamic and the
collection of co-occurring risk and protective factors should be considered to more fully
understand the life conditions that contribute to depression in African Americans. Not only are
these factors co-occurring but they are likely interrelated (e.g. religious involvement and social
support) and function collectively to influence depressive symptoms and outcomes.
Additionally, it is important to note that what most distinguished the various risk types
was the experience of protective factors, not necessarily risk factors. There were pronounced
differences in religious involvement and social support among the identified risk types
however the experience of risk factors (i.e. negative interaction, chronic stress, and negative
life events) was quite comparable across types. The importance of collective protective factors
is not surprising given the sociohistorical context of the African American experience. African
Americans have long endured discrimination, racism, violence, and oppression at both the
Hays
142

individual and institutional levels. To cope with these life stressors, African Americans have
used naturally occurring resources such as family support and religion. Thus, African
Americans who cannot benefit from protective factors like social support and religion may be
especially vulnerable to the effects of common negative life conditions which may lead to
depressive symptoms. These findings suggest the importance of emphasizing the collective
role of protective factors in influencing health outcomes. This approach is consistent with
strengths-based models that focus on increasing individual and community-level resilience,
social connectedness, and well-being as a way to improve health outcomes instead of focusing
solely on psychosocial deficits or other risks (Hays & Lincoln, 2016; Walsh, 2003).
The second key finding of this study involves the identification of risk types and
depression courses of concern as well as demographic subpopulations of African Americans
likely to suffer from excessive depressive symptoms. Specifically, there are particular
constellations of risk and protective factors that are related to worse depressive symptoms.
Individuals whose risk type was characterized by lower collective protective factors (Moderate
Protection/Low Risk and Low Protection/Low Risk) were likely to experience increased
depressive symptoms. Another group of concern was those African Americans who
experienced instability in risk and protective factors as they transitioned from one risk type to
another over time. Those with transitioning risk paths were more likely than those with a
stable risk type (either high protection or low protection) to have higher depressive symptoms.
By identifying the specific life conditions that are linked to worse depression, interventions
can be developed that target these contributors in an attempt to reduce the burden of
depression on subpopulations of African Americans.
Hays
143

In addition to risk types and transitions of concern there are depression trajectories
marked by high or increasing symptoms that should be the focus of attention. A subset of
respondents in this study had depression courses characterized by persistently high or
increasing symptoms. Individuals in these two trajectories were likely to suffer in excess
because their depressive symptoms were long lasting, chronic, or worsened over time. These
results have confirmed that there is heterogeneity in the life course experience of depression.
Results also suggest heterogeneity in depression among African Americans in that some
individuals had a worse course than others. These findings are particularly helpful in
understanding prior research that suggests that although prevalence rates of depression are low
among African Americans, disease burden is high. Race comparison studies that aggregate
African Americans suggest that African Americans in general experience worse outcomes than
whites (Richards, 2011; Williams et al., 2007). However, it may be the case that these
subgroups of African Americans who have chronic depression are truly the ones who suffer
disproportionately but are lost when considered in the aggregate. These subgroups with a
persistently high or increasing depression trajectory should be the focus of intervention efforts
as they are likely to suffer in excess.
This study also identified demographic subpopulations of African Americans with worse
depressive symptoms. In general, men were identified as a subgroup likely to experience worse
depression. Across the studies men were more likely to belong to the risk types characterized by
lower levels of protective resources and experience transitioning risk which was associated with
higher depressive symptoms. This finding is particularly important given what is known about
gender and depression. Specifically, many prevalence studies suggest that women have higher
rates of depression and thus are often the focus of depression intervention efforts (Richards,
Hays
144

2011). Although prevalence rates of depression in African American males may be low it is
plausible that men are also less likely than women to benefit from protective factors like
religiosity and social support that buffer against worse depressive symptoms. This study
confirms suggestions made by researchers that posit that African American men are often
underdiagnosed and underserved when it comes to depressive symptoms (Lincoln, Taylor,
Watkins, & Chatters, 2011; Ward & Mengesha, 2013; Watkins, 2012).
Additionally, younger age and lower educational attainment were associated with worse
risk types and depression trajectories. These associations are generally consistent with prior
research suggesting that low education is associated with depression (Barnes, Keyes, & Bates,
2013) and that depressive symptoms are elevated earlier in the life course (Lindsey, Joe, &
Nebbitt, 2010; Richards, 2011). However, in this particular study being male, young, and having
low education were all associated with increased depressive symptoms. Collectively, findings
identify several profiles that each require different approaches to intervention. Further, the results
of this study suggest that demographic characteristics alone do not lead African American men to
have high depressive symptoms but rather they are likely to experience a risk profile
characterized by a lack of protective factors (particularly religious involvement and social
support) and transitioning risk. This is a subpopulation of African Americans worthy of
increased attention because of their potential to experience excessive depressive symptoms and
because they are an underserved group in term of depression research and treatment.
Implications

Implications for Research and Theory Development
This study was guided by aspects of the Cumulative Advantage/Disadvantage theory
(CAD). The theory suggests that early life advantages and disadvantages accumulate over
Hays
145

time to affect individual health outcomes. Once specific mechanism of the theory (the path-
dependent) was particularly helpful in guiding the analyses in Chapter 3. Chapter 3 provided
support for the path-dependent mechanism by confirming that there were distinct paths of
risk and protective factors for African Americans which were predictive of their depression
outcomes. However, this study contributed to our understanding of the theory in a few
specific ways. First, this study suggests that, for African Americans, risk factors
(disadvantages) and protective factors (advantages) were co-occurring and the experience of
protective factors (or the lack thereof) was more deterministic of depression outcomes than
risk. This suggests that when considering contributors to health outcomes advantages and
disadvantages must be considered as co-occurring and dynamic. Thus, the negative impact of
risk factors (or disadvantages) may be offset by the experience of protective factors
(advantages). Other researchers have suggested that social support and wellbeing buffer
against the adverse effects of stressful life events (Cohen & McKay, 1984). Incorporating the
stress buffering hypothesis and the CAD theory may more adequately explain the
contributors to depression in African Americans.
Second, this study adds to theory by suggesting that paths towards depression
outcomes are dynamic and heterogeneous. Although the path-dependent mechanism of the
CAD suggests variable paths towards depression it does not specify whether or not these
paths are linear or if transitions in paths occur. In this study, there was heterogeneity in the
paths towards depression with two subgroups having stable risk types over time. However, a
third subgroup of African Americans did not have a linear accumulation of risk and
protective factors but was likely to experience a path characterized by transitions between
risk types over time. This means that a subgroup of African Americans experienced changes
Hays
146

in certain life conditions (advantages and disadvantages) or changes in the impact of these
life conditions throughout their lives; at some points in time they enjoyed high levels of
social support and religiosity and at other times they did not. This extends our understanding
of the CAD by suggesting that although some, and perhaps most, experience an accumulation
of advantages and disadvantages there is another subpopulation of African Americans whose
risk transitions. In fact, it is this subgroup of individuals who transition between risk types
that is most likely to suffer from elevated depressive symptoms. Thus, the contribution to
theory is the suggestion that heterogeneity within the population and in the experience of
advantages and disadvantages over time must be explored. Again, the existence of advantages
could offset or buffer against the negative effects of disadvantages. Theory must consider the
interaction between advantages and disadvantages over time and how this dynamic
interaction contributes to paths towards depressive symptoms.
The results of this study also lay a foundation for future research that seeks to
examine collective risk and protective factors for depression in African Americans. Now that
it has been established that risk types for depression in African Americans exists, subsequent
studies should examine the extent to which risk types might differ across gender. This is
important considering study findings that suggest that men experience lower protective
factors and are likely to transition between risk types over time. Thus, research should seek to
uncover variation among African American men that could help explain why they have lower
prevalence of depression than African American women (Substance Abuse and Mental
Health Services Administration, 2015) but have a risk profile linked to more depressive
symptoms.
Hays
147

Studies that investigate varying risk typologies for African American men and women
separately might further inform interventions using an intersectional approach that accounts
for variations in experiences across race and gender. Specifically, the addition of factors such
as having children, discrimination, or health status may provide more information into
potentially modifiable contributing factors to depression for African American men and
women separately. Further, this study restricted its examination to a few risk and protective
factors known from prior literature to be associated with depression in African Americans.
However, future research should go beyond this study by investigating other contributors
such as health status, region, and neighborhood conditions. Adding these factors to the risk
typology will make for a more robust constellation of life conditions. This is particularly
important for analyses of specific demographic subgroups like older adults for whom
physical health is closely related to mental health (Pickett, Bazelais, & Bruce, 2013).
Additionally, Chapter 4 of this dissertation explored the influence of risk type on
depression course and confirmed that there is a relationship between baseline risk type and
depression trajectory. Future studies should take these results a step further by examining the
influence of the longitudinal experience of risk on depression course. For example, a future
study might ask, does someone’s experience of risk and protective factors over their life
course determine their depression trajectory? Do African Americans with transitioning risk
patterns have a worse course of depressive symptoms? Answers to these questions would
serve to further illuminate the entre points for intervention so that the specific points during
the life course at which protective factors increase or decrease can be targeted.
Implications for Social Work Practice

Hays
148

One of the main objectives of this dissertation was to increase knowledge that can aid
in the development of targeted and tailored interventions to reduce the burden of depression
for African Americans. The identification of specific combinations of risk and protection that
are linked to depression outcomes (i.e. religiosity, social support, and financial stress) serves
as a guide to the development of interventions that target the collection of factors associated
with more depressive symptoms. For example, in Chapter 2, individuals in the Moderate
Protective/Low Risk and Low Protective/Low Risk types were at the greatest risk for worse
depressive symptoms. An intervention targeting these groups should be based in a religious
and family context, not only focusing on the individual’s experience of depression, but on the
church, family, and friends that surround the individual. Because the experience of protective
factors was key in distinguishing risk types, interventions should focus on helping collateral
supports increase their capacity to provide adequate social support and avoid negative social
interaction with loved ones. Focusing on improving social and religious connections may be a
way of reducing, or preventing, depressive symptoms for African Americans.
Further, specific interventions are needed that target younger adults and seek to connect
them to social and religious resources that will increase their experience of protective factors
earlier in life. This is particularly important because this study suggests that the risk type
someone belongs to at an early age will likely continue throughout their adult lives. Interventions
are needed that are tailored to younger African Americans and focus on enhancing social
support, increasing connection to religion, and promoting high marital quality as a way of
increasing protective factors early in life that, if sustained, may buffer them from experiencing
depressive symptoms.
Hays
149

Findings revealed particular profiles of African Americans that should be the focus of
intervention efforts. These profiles include men with low protective factors, young African
Americans, and those with low levels of education because they are most likely to face increased
depressive symptoms. This information is key in tailoring interventions towards subpopulations
of African Americans. An approach that considers the intersection of gender, race, and risk
profile will be helpful in addressing the unique life experiences of African American men;
especially considering that they are not traditionally considered to be at risk for depression
(Watkins, 2012). Issues such as interactions with police and the criminal justice system,
institutional racism, and neighborhood conditions should not be ignored as these are issues that
may impact African American men (Xanthos, Treadwell, & Holden, 2010). Programs should
consider age specific information relevant to relationships, education and careers, and economic
stability which are all things young African Americans are likely to be concerned about.
Interventions should also focus on enhancing the protective factors of men, young adults, and
those with low education as they are likely to experience low social support and religious
involvement. Increasing their ability to develop and maintain stable social supports and
connections will be important as changes, gains, and losses of these protective factors are
particularly damaging. Further, for the population as a whole, any intervention curriculum or
literature should also be formatted in a way that is accessible to those with less than a high
school education and culturally tailored to include language that African Americans can relate to
and be staffed by persons who understand their cultural experiences (Bernal & Rodríguez, 2012).
Results of this study not only suggest who (e.g. demographic subgroups) and what (e.g.
low protective factors) to target through interventions but they also suggest when to target them.
Using longitudinal data this study was able to examine risk and protective factors as well as
Hays
150

depressive symptoms over the life course. Findings provide clues into critical intervention entre
points. For example, based on the results from chapter 3 interventions should target younger
African Americans who have high protective factors and help them to develop the skills
necessary to maintain these supports. This is important because some individuals in the High
Protective/Low Risk type transitioned into the Low Protective/Low Risk type and individuals
with transitioning risk experienced the highest depressive symptoms. Helping African Americans
maintain social supports and connections to religion earlier in life may help them remain in the
Stable High Protection/Low Risk group and thus reduce their likelihood of experiencing
depressive symptoms.
Church Based Mental Health Promotion provides a model for the development of
interventions that seek to improve mental health, reduce mental illness, and increase social
support within a church or faith-based context (Campbell, Hudson, Resnicow, Blakeney, &
Baskin, 2007). This model has the potential to improve depression outcomes among African
Americans by focusing on reducing symptoms as well as improving overall mental health while
integrating family and church within a strengths-based approach (Hays & Lincoln, 2016).
Although this approach has shown promise, there are very few documented interventions that
focus on addressing the mental health of African Americans while incorporating faith
(Hankerson & Weissman, 2012; Hays & Aranda, 2016). Partnerships between mental health
professionals and African American clergy have the potential to increase psychoeducation, skills
building, and depression prevention efforts among African Americans in a way that is tailored to
their cultural norms (Hays, 2015). More attention should be given to research leading to the
development of Church Based Mental Health Promoting interventions given the importance of
Hays
151

religious involvement and social support in distinguishing risk types for depression among
African Americans.
Limitations

Findings from this dissertation should be considered in light of its contributions and
limitations. Responses were self-reported and may be subject to underreporting. For example,
chronic financial stress was self-reported and may not capture experiences of poverty and
lack of financial resources in the same way that more objective measures might (e.g.
enrollment in public assistance programs). Survey respondents were community dwelling and
non-institutionalized and results are generalizable to this population but not those who are
incarcerated or hospitalized. Also, because the data are from 1986 some of the results may
not represent current trends in behavior among African Americans (particularly in Chapter 2).
However, experiences, like religiosity, have been found to be fairly stable across time for
African Americans (Pew Research Center, 2015) which suggests that these data might still be
quite relevant in capturing contributors to poor depression outcomes experienced by African
Americans at later time points. Further, some researchers have noted the benefits to
measuring both time-variant and time-invariant correlates of health outcomes (Liang, Xu,
Quiñones, Bennett, & Ye, 2011). This study has established relationships between
demographic characteristics, risk types, and depression courses. Accordingly, future studies
can investigate the degree to which changes in demographic factors over time influence risk
and protective factors for depression in African Americans.
Also, there were discrepant time intervals between waves. Specifically, in Chapter 2,
there was a large span of time between wave 2 and wave 5 (22 years). It may be possible that
individuals experienced transitions in their risk during this time period that could not be
Hays
152

captured. Further, although Chapter 3 accounted for the variability in time between waves
through its analysis, changes in risk types and depressive symptoms are possible particularly
in the larger distances between waves. Also, Chapter 3 required the use of dichotomous
variables for the LTA which forwent the nuanced information available in Chapter 2 which
utilized variables with more response categories. The difference in variable construction
between the two chapters likely was the reason for the difference in the number of identified
latent classes. Although Chapter 3 did not include the Moderate Protective/Low Risk group
the results of the two dissertation chapters were still quite consistent. Despite these
limitations this is the first study to examine longitudinal risk and protective factors for
depression and depression trajectories in a national sample of African American adults. The
results of this study provide information necessary for the development of interventions that
can target the constellation of life conditions associated with worse depression outcomes (i.e.
low social support and religiosity) and be tailored to the subpopulations of African
Americans most likely to suffer (i.e. men and young adults).
In summary, African Americans experience excess depression burden. Social work
research and practice have been limited in the ability to address this problem, partially due to
lack of information about contributors to depression in the population. This current study has
shed light on a previously obscured area of research related to risk factors for depression and
variable depression trajectories in African Americans. Researchers and practitioners can use
the results of this study to guide the development of depression interventions for African
Americans which target specific subpopulations based on risk type, course of symptoms, and
demographic profiles to reduce the excess depression burden.

Hays
153

References
Barnes, D., Keyes, K., & Bates, L. (2013). Racial differences in depression in the United States:
how do subgroup analyses inform a paradox? Social Psychiatry and Psychiatric
Epidemiology, 48(12), 1941–1949. http://doi.org/10.1007/s00127-013-0718-7
Bernal, G., & Rodríguez, M. D. (2012). Cultural adaptations: Tools for evidence-based practice
with diverse populations. American Psychological Association.
Campbell, M. K., Hudson, M. A., Resnicow, K., Blakeney, N., & Baskin, M. (2007). Church-
Based Health Promotion Interventions : Evidence and Lessons Learned. Annual Rev. Public
Health, 28, 213–234.
Cohen, S., & McKay, G. (1984). Social support, stress and the buffering hypothesis: A
theoretical analysis. In & J. E. S. (Eds. ). A. Baum, S. E. Taylor (Ed.), Handbook of
psychology and health (pp. 253–267). Hillsdale, NJ: England: Lawrence Erlbaum
Associates, Inc.
Hankerson, S. H., & Weissman, M. M. (2012). Church-based health programs for mental
disorders among African Americans: A review. Psychiatric Services, 63(3), 243–249.
http://doi.org/10.1176/appi.ps.201100216
Hays, K. (2015). Black churches’ capacity to respond to the mental health needs of African
Americans. Social Work & Christianity, 42(3), 296–312.
Hays, K., & Aranda, M. P. (2016). Faith-based mental health interventions with African
Americans: A review. Research on Social Work Practice, 26, 777–789.
http://doi.org/10.1177/1049731515569356
Hays, K., & Lincoln, K. D. (2016). Mental health promotion among African Americans. In E. M.
Hays
154

P. Schott & E. L. Wiess (Eds.), Transformative social work practice. Thousand Oaks, CA:
Sage.
Liang, J., Xu, X., Quiñones, A. R., Bennett, J. M., & Ye, W. (2011). Multiple trajectories of
depressive symptoms in middle and late life: Racial/ethnic variations. Psychology and
Aging, 26(4), 761–777. http://doi.org/10.1037/a0023945
Lincoln, K. D., Taylor, R. J., Watkins, D., & Chatters, L. M. (2011). Correlates of psychological
distress and major depressive disorder among African American men. Research on Social
Work Practice, 21(3), 278–288. http://doi.org/10.1177/1049731510386122
Lindsey, M. A., Joe, S., & Nebbitt, V. (2010). Family matters: The role of mental health stigma
and social support on depressive symptoms and subsequent help seeking among African
American boys. Journal of Black Psychology, 36(4), 458–482.
http://doi.org/10.1177/0095798409355796
Pew Research Center. (2015). America’s changing religious landscape. Retrieved from
http://www.pewforum.org/2015/05/12/americas-changing-religious-landscape/
Pickett, Y. R., Bazelais, K. N., & Bruce, M. L. (2013). Late-life depression in older African
Americans: A comprehensive review of epidemiological and clinical data. International
Journal of Geriatric Psychiatry, 28(6), 903–913. http://doi.org/10.1002/gps.3908
Richards, D. (2011). Prevalence and clinical course of depression: A review. Clinical Psychology
Review. http://doi.org/10.1016/j.cpr.2011.07.004
Substance Abuse and Mental Health Services Administration. (2015). Results from the 2015
National Survey on Drug Use and Health: Mental health findings. Rockville, MD.
Hays
155

Walsh, F. (2003). Family resilience: A framework for clinical practice. Family Process, 42(1),
1–18. http://doi.org/10.1111/j.1545-5300.2003.00001.x
Ward, E., & Mengesha, M. (2013). Depression in African American men: A review of what we
know and where we need to go from here. American Journal of Orthopsychiatry, 83(2
PART 3), 386–397. http://doi.org/10.1111/ajop.12015
Watkins, D. (2012). Depression over the adult life course for African American men: Toward a
framework for research and practice. American Journal of Mens Health, 6(3), 194–210.
http://doi.org/10.1177/1557988311424072
Williams, D. R., González, H. M., Neighbors, H., Nesse, R., Abelson, J. M., Sweetman, J., &
Jackson, J. S. (2007). Prevalence and distribution of major depressive disorder in African
Americans, Caribbean blacks, and non-Hispanic whites: Results from the National Survey
of American Life. Archives of General Psychiatry, 64(3), 305–315.
http://doi.org/10.1001/archpsyc.64.3.305
Xanthos, C., Treadwell, H. M., & Holden, K. B. (2010). Social determinants of health among
African American men. Journal of Men’s Health, 7(1), 11–19.
http://doi.org/10.1016/j.jomh.2009.12.002

Abstract (if available)

Abstract Although rates of clinical depression for African Americans are low, research suggests that disease burden is high as African Americans are likely to face chronic, debilitating, severe, and persistent depressive symptoms. To alleviate the excess burden of depression in African Americans we must better understand the various factors that contribute to depressive symptoms in this population and how these factors vary within the population. The Cumulative Advantage/Disadvantage (CAD) theory suggests that there may be specific constellations of risk and protective factors that constitute paths toward better or worse depression outcomes. It also suggests that early life exposure to certain life conditions, and the advantages and disadvantages they engender, will impact depressive symptoms over the life course. The overall purpose of this dissertation is to fill gaps in the current literature and uncover contributors to the excess burden of depression in African Americans by exploring heterogeneity within the population and in the experience of depressive symptoms. Specifically, this study examines distinct profiles of risk and protective factors (types), transitions in risk and protective factors over time, and various depression courses (trajectories) among African American adults aged 25 or older. ❧ Using data from the Americans’ Changing Lives study, and the CAD theory as a conceptual framework, this dissertation examines three primary research questions: 1) Are there specific risk profiles (i.e. types) that can be characterized by constellations of psychosocial, sociocultural factors, and individual demographic characteristics? 2) Do individual African Americans transition or remain stable in their risk type over time? and 3) Are there multiple courses of depression for African Americans? If so, are these courses (i.e. trajectories) differentiated by psychosocial, sociocultural, and demographic factors? ❧ The dissertation is presented in a multiple manuscript format, including three studies that are distinct but related. Empirical study 1 (chapter 2) investigated unique constellations of multiple co-occurring risk and protective factors that offer greater explanatory potential regarding depression burden for African Americans than those measures treated independently. Study results revealed a risk typology with three distinct risk patterns for African Americans and identified risks types that were associated with worse depressive symptoms. Study 2 (chapter 3) identified distinct constellations of risk and protection for depression and investigated individual stability, or transitions, between risk types over time using Latent Transition Analysis. Results confirmed that there are distinct classes of risk and protection and further identified eight distinct risk paths. Those who transitioned between risk types were found to have higher depressive symptoms than those with stable risk. Empirical paper 3 (chapter 4) examined heterogeneity in the course of depressive symptoms among African American adults and older adults to identify variable depression trajectories and their demographic, psychosocial, and sociocultural correlates. Growth mixture modeling was employed and results suggested that there are 5 distinct depression trajectories for African Americans: Low Symptoms, High Symptoms, Increasers, Slow Decliners, and Fast Decliners. Several factors were found to predict trajectory membership including gender, age, education, marital status, and their psychosocial/sociocultural risk type. ❧ Overall, this dissertation provides information needed to develop depression interventions that are aimed at a moving target and tailored to the unique needs of subpopulations of African Americans likely to experience poor depression outcomes. This study’s results suggest specific combinations of protective factors that should be promoted among African Americans because they buffer against depression. Also, a subpopulation of African Americans exists that is likely to experience persistent or increasing depression. Information on the demographic profile of those African Americans who are likely to suffer excessively should be consulted when designing tailored interventions.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Cultural risk and protective factors for tobacco use behaviors and depressive symptoms among American Indian adolescents in California

PDF

The impact of psychosocial rehabilitation services on recovery outcomes for racial/ethnic minorities with severe mental illness

PDF

The link between maternal depression and adolescent daughters' risk behavior: the mediating and moderating role of family

PDF

Social network influences on depressive symptoms among Chinese adolescents

PDF

Cultural resources and health among Asian Americans: results from the National Latino and Asian American Study

PDF

A sociocultural and developmental approach to intimate partner violence among a sample of Hispanic emerging adults

PDF

A study of depression disclosure among Latino older adults in Los Angeles

PDF

The role of depression symptoms on social information processing and tobacco use among adolescents

PDF

Depression severity, self-care behaviors, and self-reported diabetes symptoms and daily functioning among low-income patients receiving depression care

PDF

An exploratory study of the role of religion and religious involvement among Latinos with schizophrenia and family caregivers

PDF

Implementation of peer providers in integrated health care settings

PDF

Mindfulness and resilience: an investigation of the role of mindfulness in post-9/11 military veterans' mental health-related outcomes

PDF

Incarceration trajectories of mothers in state and federal prisons and their relation to the mother’s mental health problems and child’s risk of incarceration

PDF

Factors and correlates of sexual behaviors among young adults from continuation high schools

PDF

Childhood cancer survivorship: parental factors associated with survivor's follow-up care behavior and mental health

PDF

Macroeconomic conditions, systematic risk factors, and the time series dynamics of commercial mortgage credit risk

PDF

The impact of family factors on the functioning of African-American consumers living with schizophrenia

PDF

U.S. Latinx youth development and substance use risk: adversity and strengths

PDF

Properties of human motor control under risk and risk aware control

PDF

Gendered issues in alcohol abuse and dependence among HIV-positive African Americans

Asset Metadata

Creator Hays, Krystal (author)

Core Title Depression trajectories and risk typology among African Americans

School School of Social Work

Degree Doctor of Philosophy

Degree Program Social Work

Publication Date 06/28/2017

Defense Date 05/08/2017

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

Tag African Americans,Depression,Mental Health,OAI-PMH Harvest,risk

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Lincoln, Karen, D. (committee chair), Baezconde-Garbonati, Lourdes (committee member), Cox, Robynn (committee member), Lloyd, Donald (committee member), Murray, Cecil (committee member)

Creator Email krystalw43@gmail.com,kswillia@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-391796

Unique identifier UC11264226

Identifier etd-HaysKrysta-5462.pdf (filename),usctheses-c40-391796 (legacy record id)

Legacy Identifier etd-HaysKrysta-5462.pdf

Dmrecord 391796

Document Type Dissertation

Rights Hays, Krystal

Type texts

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

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

Repository Name University of Southern California Digital Library

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