University of Southern California Dissertations and Theses

Evaluating a cultural process model of depression and suicidality in Latino adolescents

(USC Thesis Other)

Evaluating a cultural process model of depression and suicidality in Latino adolescents

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
EVALUATING A CULTURAL PROCESS MODEL OF DEPRESSION AND
SUICIDALITY IN LATINO ADOLESCENTS

by

Yolanda Maria Céspedes

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

August 2008

Copyright 2008 Yolanda Maria Céspedes

ii

Acknowledgements
I would not have been able to complete this dissertation without the guidance and
support of my mentor and advisor, Dr. Stanley Huey. I thank him sincerely for inspiring
me to pursue my research interests and for his unwavering encouragement through this
process.
I also thank the members of my dissertation committee, Drs. Margaret Gatz, Gayla
Margolin, Jennifer Unger, and Rand Wilcox, who provided important insights that greatly
strengthened my project. I give special thanks to Dr. Wilcox for his unending patience
and commitment to my research.
I cannot find words to express my gratefulness to family and friends. I could not
have completed this dissertation without their love and support. I thank them for
standing by me when I felt defeated and for celebrating with me each step forward.
This research project was completed with financial support from the American
Foundation for Suicide Prevention.

iii

Table of Contents

Acknowledgements ii

List of Tables iv

List of Figures v

Abstract vii

Chapter One: Background and Significance 1

Chapter Two: Experimental Design and Methods 21

Chapter Three: Analyses 31

Chapter Four: Results 39

Chapter Five: Discussion 66

References 84

iv

List of Tables

Table 1: Correlations Between Male and Female Caregiver Scores 30
on Indices of Cultural Discrepancy

Table 2: Comparison of Male and Female Scores on Cultural 40
Discrepancy, Family Functioning, Fatalism, and Depression
Measures

Table 3: Comparison of Male and Female Probabilities on Suicide 44
Measures

Table 4: Correlations Between Cultural Discrepancy, Family 45
Functioning, Fatalism, Depression, and Suicidality

Table 5: Confidence Intervals for Differences Between Regression 52
Coefficients (t- t') for Hypothesized Mediational Models

Table 6: Confidence Intervals for Adjusted Regression Coefficients (t') 52
for Hypothesized Mediational Models

Table 7: Correlations Between True and Perceived Measures of Cultural 55
Discrepancy

Table 8: Correlations Between True Cultural Discrepancy, Family 56
Functioning, Fatalism, Depression, and Suicidality Variables

Table 9: Confidence Intervals for the Slope of the Quantile Regression 58
Path For Significant Predictors of Suicidality

Table 10: Analyses of Explanatory Power for Outcomes in Mediation 64
Models

v

List of Figures

Figure 1: Mediational model illustrating hypothesized regression paths 19
among cultural discrepancy, family functioning, fatalism, and
outcome variables.

Figure 2: Model illustrating hypothesized moderated relationship between 20
cultural discrepancy and outcome variables.

Figure 3: Plot of the shift function for male and female machismo 41
discrepancy scores.

Figure 4: Plot of the shift function for male and female cohesion scores. 41

Figure 5: Plot of the shift function for male and female RADS scores. 42

Figure 6: Plot of the shift function for male and female CES-D scores. 42

Figure 7: Regression paths for familism discrepancy conflict 47
depression composite mediational model.

Figure 8: Regression paths for familism discrepancy conflict 48
future ideation mediational model.

Figure 9: Regression paths for familism discrepancy cohesion 48
depression composite mediational model.

Figure 10: Regression paths for conflict fatalism depression 49
composite mediational model.

Figure 11: Regression paths for conflict fatalism future ideation 49
mediational model.

Figure 12: Regression paths for conflict fatalism negative thoughts 50
mediational model.

Figure 13: Regression paths for cohesion fatalism depression 50
composite mediational model.

Figure 14: Regression paths for cohesion fatalism negative thoughts 51
mediational model.

Figure 15: Models summarizing mediational results for hypothesized 53
cultural process model.

vi

Figure 16: Machismo discrepancy and gender interaction effect on CES 54
depression.

Figure 17: Two-dimensional plot of the regression surface illustrating 57
the pattern of association between true familism discrepancy
and future suicide attempt.

Figure 18: Three-dimensional plot of the regression surface illustrating the 59
pattern of association among familism discrepancy, conflict,
and depression (based on composite of RADS and CES-D).

Figure 19: Three-dimensional plot of the regression surface illustrating the 59
pattern of association among familism discrepancy, cohesion,
and depression (based on composite of RADS and CES-D).

Figure 20: Three-dimensional plot of the regression surface illustrating the 60
pattern of association among familism discrepancy, conflict,
and future ideation.

Figure 21: Three-dimensional plot of the regression surface illustrating the 61
pattern of association among conflict, fatalism, and depression
(based on composite of RADS and CES-D).

Figure 22: Three-dimensional plot of the regression surface illustrating the 61
pattern of association among cohesion, fatalism, and depression
(based on composite of RADS and CES-D).

Figure 23: Three-dimensional plot of the regression surface illustrating the 62
pattern of association among conflict, fatalism, and future
suicidal ideation.

Figure 24: Three-dimensional plot of the regression surface illustrating the 62
pattern of association among conflict, fatalism, and negative
suicidal thoughts.

Figure 25: Three-dimensional plot of the regression surface illustrating the 63
pattern of association among cohesion, fatalism, and negative
suicidal thoughts.

vii

Abstract
Adolescent Latinos report high levels of depression and suicidality compared to
youth from other cultural backgrounds, with Latinas reporting the highest rates. Despite
this well-documented ethnic disparity, few studies have investigated why Latinos are at
disproportionately high risk. Cultural process models argue that intercultural discrepancy
between Latino youth and their immigrant caregivers contributes to poor mental health
outcomes through its mediating effects on family functioning. This study provided an
empirical test of this model, with a primary focus on youth depression and suicidality.
This study also evaluated the contributing role of fatalism, a variable excluded from prior
cultural process models. Self-report measures assessing acculturation, gender role
beliefs, familism, family functioning, fatalism, depression, and suicidality were
administered to 395 Latino students in a Los Angeles high school. Results showed that
youth-caregiver discrepancy in familism was significantly associated with depression,
with this effect partially mediated by family conflict and cohesion. Discrepancy in
familism was also significantly associated with suicidal ideation, with this effect partially
mediated by family conflict. Results also showed that family conflict was significantly
associated with depression, suicidal ideation, and negative thoughts about suicide, with
these effects partially mediated by fatalism. Moderator analyses suggested that the
relationship between discrepancy in gender role beliefs and depression may be more
robust for females than for males. Treatment implications and directions for future
research are discussed.

1

Chapter One: Background and Significance
Depression and suicidality (suicidal ideation and suicide attempts) are fairly common
in the adolescent population, but reported at higher rates among Latino adolescents
(Blazer, Kessler, McGonagle, & Swartz, 1994; Emslie, Weinberg, Rush, Adams, &
Rintelmann, 1990; Joiner, Perez, Wagner, Berenson, & Marquina, 2001; Lester &
Anderson, 1992; Roberts & Chen, 1995; Roberts, Roberts, and Chen, 1997; Roberts &
Sobhan, 1992; Twenge & Nolen-Hoeksema, 2002; Weinberg & Emslie, 1987).
Researchers have identified many correlates of adolescent depression and suicidality;
however, few studies have addressed the cultural factors that may account for increased
risk among Latino adolescents. Latino youth often experience stresses resulting from the
acculturation experience and pressures to conform with parental expectations (Hovey &
King, 1996; Zayas & Dyche, 1995; Zayas, Kaplan, Turner, Romano, & Gonzalez-Ramos,
2000), which may increase risk for depressive symptomatology and suicidal behavior.
The current study provides an empirical test of such a model. Uncovering cultural
mechanisms may be helpful in understanding why Latino adolescents are at increased
risk for depression and suicidality and identifying treatments that might be most
appropriate for this population.
Ethnic Differences in Youth Depression
In a recent national survey of almost 14,000 9
th
through 12
th
graders (Centers for
Disease Control and Prevention, 2006), Latino adolescents (36.2%) were considerably
more likely than African-American (28.4%) and Anglo-American (25.8%) adolescents to
have experienced persistent sadness and hopelessness. Among all students surveyed,
females reported higher levels of sadness and hopelessness than males. In particular,

2

Latinas (46.7%) were considerably more likely to report these feelings than African-
American (36.9%) and Anglo-American (33.4%) females.
Other studies of high school students have yielded similar results. A survey of Latino
and African-American high school students showed that Latinos reported significantly
higher levels of depression (Lester & Anderson, 1992). Roberts and Sobhan (1992)
found similar outcomes when they compared Mexican-American and Anglo-American
students between 12 and 17 years of age. Mexican-American adolescents were
approximately 1.5 times more likely than their Anglo-American counterparts to report
depressive symptoms, even when controlling for age, gender, and socioeconomic status.
Overall, depression was higher among females compared to males, and this held true for
those of Mexican descent as well (Roberts & Sobhan, 1992). Indeed, there is consistent
evidence that Latinas report the highest levels of depression in high school samples
(Emslie et al., 1990; Weinberg & Emslie, 1987).
Similar findings have been found among pre-adolescents as well. Mexican-American
middle school students report higher levels of depression compared to African-Americans
and Anglo-Americans (Roberts & Chen, 1995; Roberts, Roberts, & Chen, 1997). These
studies also found that females were more likely to report depression than males, with
Mexican-American females reporting the highest levels. A recent study including more
than 1,000 middle school children found that those of Latino background reported higher
levels of depression compared to those of Anglo-American, African-American,
Asian/Pacific Islander, Native American, and Middle Eastern descent (McLaughlin, Hilt,
& Nolen-Hoeksema, 2007).

3

Finally, a meta-analysis reviewed 310 samples of children and adolescents (N =
61,424) who completed the Children’s Depression Inventory (CDI). Overall, Latino
youth scored significantly higher on the CDI than both African-Americans and Anglo-
Americans (Twenge & Nolen-Hoeksema, 2002).
Thus, the existing literature strongly suggests that Latino youth, particularly Latinas,
are at increased risk for depression relative to other ethnic groups.
Ethnic Differences in Youth Suicidality
Studies also suggest that Latino adolescents are at increased risk for suicidality. For
example, Mexican-American adolescents report more suicidal thoughts than their Anglo-
American counterparts (Roberts & Chen, 1995). Similarly, Latino students are more
likely to report serious suicidal ideation when compared with African-Americans (Lester
& Anderson, 1992). Furthermore, females generally report more suicidal ideation than
males, with Mexican-American females reporting the highest levels of ideation (Roberts
& Chen, 1995).
These findings are consistent with results from the most recent Youth Risk Behavior
Surveillance (YRBS) administered by the Centers for Disease Control and Prevention
(CDC, 2006). The survey found that Latino students (17.9%) were more likely than
African-American (12.2%) and Anglo-American (16.9%) students to have seriously
considered suicide, with Latinas reporting the highest rates (24.2%). Latino (11.3%)
students were also more likely than both African-American (7.6%) and Anglo-American
(7.3%) students to have made a suicide attempt, with Latinas (14.9%) again reporting the
highest rates. Across all survey questions about suicide, Latino students were among

4

those reporting the highest rates and Latinas tended to report higher rates than Latino
males (CDC, 2006).
Similar statistics were cited in a report on the nation’s health status, compiled by the
National Center for Health Statistics (NCHS, 2000). The report, which provides gender-
based comparisons, indicates that among male adolescents in grades nine through twelve,
Latinos (13.6%) were more likely than Anglo-Americans (12.5%) and African-
Americans (11.7%) to have seriously considered suicide. However, African-American
males (7.1%) had higher rates of suicide attempts compared to Anglo-American (4.5%)
and Latino (6.6%) males. African-American males (3.4%) also had the highest rate of
injurious attempts compared to Anglo-American (1.6%) and Latino (1.4%) males.
Among females, Latinas (26.1%) were more likely than Anglo-Americans (23.2%) and
African-Americans (18.8%) to have considered suicide. Furthermore, the percentage of
Latinas who had attempted (18.9%) was considerably higher than the percentage for
Anglo-American (9.0%) and African-American (7.5%) females. Latinas (4.6%) were
also twice as likely as Anglo-American (2.3%) and African-American (2.4%) females to
have made a suicide attempt that resulted in injury.
The results of these reports indicate that Latino adolescents are at higher risk for
suicidal ideation and attempts compared to other ethnic groups. Furthermore, data across
multiple studies suggest that Latinas are at greatest risk and report rates of suicidality up
to twice as high as other ethnic groups. When comparisons are restricted to males and
females separately, however, there is evidence that African-American males are at greater
risk for suicide attempts compared to Latino and Anglo-American males. While this
difference in prevalence may not be statistically significant, it highlights the fact that

5

overall group differences in suicidality are heavily impacted by the comparatively high
rates among Latinas.
Etiological Factors and Correlates of Youth Depression and Suicidality
Child psychopathology researchers have uncovered many factors that elevate risk
for youth depression. Factors range from negative cognitive style (Simmons, Cooper,
Drinkwater, & Stewart, 2006; Southall & Roberts, 2002; Turner & Cole, 1994) to poor
problem-solving skills (Rudolph, Hammen, & Burge, 1994) to stressful life events
(Bouma, Ormel, Verhulst, & Oldehinkel, 2008; DuBois, Felner, Brand, Adan, & Evans,
1992; Hankin, Mermelstein, & Roesch, 2007). Various etiological models have been
proposed to explain adolescent depression (Hammen & Rudolph, 1996) and a more
integrative approach has been recommended given evidence for multiple etiological
pathways (Compas, 1997). Indeed, some of the more specific models of adolescent
depression have not received support in prospective studies, and authors have argued that
multivariate risk models be used (Burton, Stice, & Seeley, 2004). Birmaher and
colleagues (1996) reviewed the literature on childhood and adolescent depression over a
ten-year period and identified four factors that are repeatedly shown to be associated with
depression in youth: genetic diathesis, family environment, stressful life events, and
negative cognitive style.
Similarly, research has identified multiple factors that increase risk for suicidal
ideation and suicide attempts among youth, including depression (Chabrol & Moron,
1988; Fortuna, Perez, Canino, Sribney, & Alegria, 2007; Lewinsohn, Rohde, & Seeley,
1993; Lewinsohn, Rohde, & Seeley, 1996; Locke & Newcomb, 2005; Negron, Piacentini,
Graae, Davies, & Shaffer, 1997; Rutter & Behrendt, 2004; Spirito, Brown, Overholser, &

6

Fritz, 1989). A developmentally based etiological model has been proposed, where
interactions among cognitive, emotional, behavioral, family, and school factors trigger
youth suicidal behavior (Pfeffer, 1997). The pathways and risk factors included in this
multivariate model overlap greatly with those contributing to risk for depression.
While research indicates that depression increases risk for suicidality, and that
depression and suicidality share a similar etiological pathway, there is also empirical
evidence that suicidality occurs in the absence of depression and that specific risk factors
may uniquely contribute to suicidality and depression. For example, research shows that
large proportions of suicidal adolescents do not meet clinical criteria for depression
(Hollis, 1996; Shaffer et al., 1996). Furthermore, objective measures of depression do
not distinguish between suicide attempters and those who make no suicide attempt
(Mann, Waternaux, Haas, & Malone, 1999). A subset of suicidal behavior appears to be
associated with impulsive tendencies and not with depression (Simon et al., 2001), which
may account for these findings.
There is also evidence that certain family characteristics differentiate depressed
youth from suicidal youth (Wagner, 1997). Furthermore, research shows that while some
psychosocial risk factors (e.g., conflict with parents) are associated with suicidal behavior
because of their association with depression, other psychosocial risk factors (e.g., family
social support) continue to be significantly associated with suicidal behavior once
depression is controlled for (Lewinsohn, Rohde, Seeley, 1993; Lewinsohn, Rohde,
Seeley, 1994).
These findings suggest that the association between depression and suicidality is not
straightforward and that, contrary to clinical expectations, depression does not necessarily

7

precede suicidality. For this reason, additional research is warranted in order to further
explore etiological pathways to both types of mental health problems. The present study
explores the role of family functioning in contributing to depression and suicidality in
Latino youth. Family functioning is repeatedly implicated as an important component in
multivariate models of depression and suicidality (Birmaher et al., 1996; Pfeffer, 1997)
and literature reviews support the claim that family processes have an important
etiological role in these mental health problems (Sheeber, Hops, Alpert, Davis, &
Andrews, 1997; Wagner, 1997).
Family conflict and low family cohesion are strongly linked to both depression
(Asarnow, Carlson, & Guthrie, 1987; Cole & McPherson, 1993; Constantine, 2006;
Cumsille & Epstein, 1994; Fendrich, Warner, & Weissman, 1990; Hops, Lewinsohn,
Andrews, & Roberts, 1990; Kaslow, Rehm, & Siegel, 1984) and suicidality (Asarnow &
Carlson, 1988; Carlton-Ford, Paikoff, & Brooks-Gunn, 1991; Chioqueta & Stiles, 2007;
Rubenstein, Heeren, Housman, Rubin, & Stechler, 1989) in samples of children,
adolescents, and college students. De Wilde, Kienhorst, Diekstra, and Wolters (1993)
evaluated family conflict and family cohesion among groups of depressed adolescents,
adolescent suicide attempters, and non-depressed/non-suicidal adolescents. Results
showed that the depressed and suicide attempt groups reported significantly higher levels
of family conflict and significantly lower levels of family cohesion than the non-
depressed/non-suicidal group.
The quality of parent-child interactions may contribute to overall conflict and
cohesion in the family, thereby impacting youth’s risk for subsequent depression or
suicidality (Wagner, 1997). Indeed, adolescent suicide attempters frequently report

8

conflict with parents as the primary cause of their suicide attempts (Miller, Chiles, &
Barnes, 1982; Negron et al., 1997), and also report significantly more conflict with
parents/guardians than matched controls (Corder, Shorr, & Corder, 1974). Furthermore,
positive parent-child interactions appear to serve as a protective factor against suicidality,
as “closeness” between parents and youth is associated with lower risk for suicidal
ideation and attempts (Tobin, 2000; Wagner, Silverman, & Martin, 2003).
The present study explored how high levels of family conflict and low levels of
family cohesion may increase Latino adolescents’ risk for depression and suicidality.
Given evidence that such family dysfunction may be attributable to the quality of parent-
child interactions (Corder, Shorr, & Corder, 1974; Miller, Chiles, & Barnes, 1982;
Negron et al., 1997; Tobin, 2000; Wagner, 1997; Wagner, Silverman, & Martin, 2003),
this study also explored culturally-based sources of parent-child disagreement that may
impact parent-child relationships.
Cultural Discrepancy
Cultural factors related to the immigration experience may contribute to depression
and suicidality in Latino youth because of their impact on parent-child interactions.
Clinical accounts suggest that family conflict results when Latino youth acculturate to the
host culture more rapidly than their immigrant parents (Szapocznik et al., 1989; Zayas &
Dyche, 1995). During the acculturation process, these youth adopt new attitudes, values,
and behaviors that may be incompatible with the traditional beliefs of their parents and
native culture, causing a cultural and generational gap between youth and parents.
Although these processes are theorized to operate broadly in terms of acculturation
status (Szapocznik et al., 1986), generational disparities may also arise in connection with

9

culturally-based gender and family role beliefs. Ways in which generational differences
in marianismo, machismo, and familismo beliefs might influence mental health outcomes
for Latino youth are discussed below.
Marianismo and machismo beliefs. The terms marianismo and machismo describe
traditional sex role expectations commonly found in many Latin-American societies
(Falicov, 1998). The term marianismo derives from Catholic representations of the
Virgin Mary, who is venerated as a symbol of saintliness, self-sacrifice, and emotional
suffering (Nieto, 1997; NietoGomez, 1997). Marianismo refers to the expectation for
Latinas to be acquiescent, humble, and modest (Comas-Díaz, 1987). Within a
marianismo conceptualization, Latinas are viewed as caretakers who are able to foster
harmonious relationships with others (Comas-Díaz, 1987; Soto, 1983). In addition,
Latinas are often expected to devote themselves to their families, and may be discouraged
from pursuing academic or professional goals (Senour, 1977).
Marianismo contrasts with the value of machismo, which describes Latino men as
strong, assertive, and dominant figures committed to protecting their families (Grebler,
Moore, & Guzmán, 1970). Latino men are expected to command respect, particularly
from their children, and to be resolute (Falicov, 1998). Machismo also signifies a “strong
sense of personal honor and family loyalty” (Ruiz, 1977, p. 243).
While marianismo and machismo are terms used to delineate specific gender role
ideologies in Latino culture, the values and expectations captured by these terms are not
necessarily unique to those of a Latino heritage. For example, comparisons have been
made between Latino and Asian beliefs, showing a parallel emphasis on obedience and
respect with regard to child rearing and parental discipline (Maker, Shah, & Agha, 2005).

10

Thus, while different ethnic and cultural groups may share commonalities in fundamental
beliefs, the terms used to describe these beliefs may be group-specific.
The cultural and generational gap that forms between Latino youth and their
immigrant parents during the acculturation process (Szapocznik, et al., 1986; Szapocznik
et al., 1989; Zayas & Dyche, 1995) may correspond with a similar generational
discrepancy in gender role ideologies. Indeed, gender role beliefs tend to correlate
significantly with acculturation level. Among a group of Latina adults, greater
acculturation was associated with less traditional female attitudes (Kranau, Green, &
Valencia-Weber, 1982). This suggests that as Latinas become more acculturated,
mainstream attitudes toward women begin to replace more traditional attitudes. Highly
acculturated Mexican-American parents are also more likely to endorse egalitarian
gender role beliefs compared to their less acculturated counterparts (Leaper & Valin,
1996).
As Latino youth and their immigrant parents acculturate, differential beliefs
concerning marianismo/machismo roles may lead to family conflicts over gender-related
expectations and responsibilities, thereby increasing depression and suicidal risk for
Latino youth. Indeed, preliminary research suggests that (1) gender role discrepancy may
contribute to depression among adolescent Latinos, and (2) this association is mediated
by increases in family dysfunction (Céspedes & Huey, 2008).
Familismo beliefs. The term familismo describes the strong emphasis placed on the
family unit among Latinos. It captures the sense of support and reliance on one another
among family members (Baca Zinn, 1994; Marín & Marín, 1991), as well as sentiments
of mutual obligation, loyalty, and family unity (Cortes, 1995; Rodriguez & Kosloski,

11

1998; Sabogal, Marín, Otero-Sabogal, Marín, & Perez-Stable, 1987). Familism
incorporates a wide array of culturally-nuanced attitudes among Latinos, including
beliefs that family members are obligated to assist one another when they are in need
(Burgess, Locke, & Thomes, 1963) and that family members are obligated to remain in
close proximity to each other (George, 1986). Steidel and Contreras (2003) recently
elaborated on the definition of familism, describing four primary beliefs: that the family
comes before the individual, that family members should maintain strong emotional and
physical bonds, that family members should support each other in times of need, and that
family members should protect the family name and honor. While similar to the concept
of family cohesion, familism not only captures the sense of closeness among family
members, but also emphasizes the primacy of the family and the roles, responsibilities,
and obligations assigned to family members.
Research has yielded mixed findings with regard to the association between familism
beliefs and an individual’s acculturation status. Beliefs emphasizing mutual and familial
obligations, parental respect, and family honor are negatively associated with
acculturation level (Gil, Wagner, & Vega, 2000; Rodriguez & Kosloski, 1998; Sabogal et
al., 1987; Steidel & Contreras, 2003), suggesting that these beliefs decrease as one
becomes more assimilated to mainstream culture. However, beliefs focused on using
family members as referents (i.e., as models of appropriate behaviors and attitudes) have
yielded inconsistent findings, with research showing both a negative association (Sabogal
et al., 1987) and a nonsignificant association (Rodriguez & Kosloski, 1998) with
acculturation level. Research exploring beliefs that emphasize family support are equally
inconsistent, with findings showing no association (Sabogal et al., 1987; Steidel &

12

Contreras, 2003) and a positive association (Rodriguez & Kosloski, 1998) with
acculturation. Rodriguez et al. (2007) suggest that differences in the conceptualization
and measurement of familism and acculturation, as well as an overemphasis on the
supportive aspects of the family, may contribute to inconsistent associations between
these constructs (Rodriguez, Mira, Paez, & Myers, 2007).
Although the association between familism and acculturation appears complex,
certain aspects of traditional familism beliefs diminish as an individual acculturates (Gil,
Wagner, & Vega, 2000; Rodriguez & Kosloski, 1998; Sabogal et al., 1987; Steidel &
Contreras, 2003). As acculturation gaps develop between Latino youth and their
caregivers during the immigration process, a corresponding discrepancy in familism
beliefs may emerge. As the divergence in family-based expectations becomes more
pronounced, increased family conflict and decreased family cohesion may contribute to
depression and suicidality in Latino youth.
Zayas and his colleagues have discussed how changes in familism beliefs may have a
particularly adverse affect on the mental health of adolescent Latinas by increasing their
risk for suicidal behavior (Zayas, Fortuna, Lester, & Cabassa, 2005). Several aspects of
familism are emphasized in this framework: deference to parental authority,
interdependence among family members, and restriction of female autonomy. According
to Zayas et al., familism beliefs may come into conflict with the mainstream cultural
environment, which encourages independence. As acculturating Latinas become more
autonomous, they may increasingly challenge core beliefs of family unity and
interconnectedness, thereby generating conflict with parental figures. Zayas et al.
theorize that such conflict may precipitate suicidal behavior among adolescent Latinas.

13

Fatalism
Maladaptive family interactions may predispose Latino adolescents to increased risk
for depression and suicidality through their effects on fatalism. Fatalism is generally
conceptualized as a belief in external control, wherein factors such as luck and fate
determine the outcomes of events (Joiner et al., 2001; Roberts, Roberts, & Chen, 2000;
Ross, Mirowsky, & Cockerham, 1983; Rotter, 1966; Wheaton, 1980). Fatalism has also
been viewed as an inability to alter or control one’s destiny (Cuéllar, Arnold, &
González, 1995; Flores, 2000). These conceptualizations suggest that fatalistic
individuals tend to relinquish control over life events.
Madsen (1973) described individuals of Latino descent as highly fatalistic. Research
confirms that Latino adults are more fatalistic than Anglo-American adults (Farris &
Glenn, 1976; Mirowsky & Ross, 1984; Ross, Mirowsky, & Cockerham, 1983) and that
Latino youth are more fatalistic than youth of other ethnic backgrounds (Jamieson &
Romer, 2008). However, it has also been theorized that fatalistic beliefs are not unique to
a specific ethnic or cultural group, but rather to low socioeconomic status (Sue & Sue,
1990). It may therefore be the case that Latinos are more likely to report fatalistic beliefs
compared to other ethnic groups because of their ongoing struggles against those with
more power and limited opportunities to excel (Sue & Sue, 1990).
Some posit that a tendency towards fatalism may lead to a sense of resignation in the
face of negative life events, thus precipitating episodes of depression and other mental
health problems (Domino, Fragoso, & Moreno, 1991; Mirowsky & Ross, 1984; Pearlin,
Menaghan, Lieberman, & Mullan, 1981; Roberts, Roberts, & Chen, 1997; Roberts,
Roberts, & Chen, 2000; Ross, Mirowsky, & Cockerham, 1983). Fatalistic beliefs are

14

argued to increase risk for mental health problems through their effect on active coping
(Mirowsky & Ross, 1984; Wheaton, 1980). As the individual relinquishes control over
event outcomes, feelings of helplessness may reduce motivation to cope, thereby
precipitating psychological distress.
As argued earlier, acculturation gaps that emerge between Latino youth and their
parents during the immigration process may increase parent-child discrepancy in
culturally-based beliefs, creating negative parent-child interactions and overall family
dysfunction. Exposure to a negative family environment may consequently increase
fatalistic beliefs among Latino youth, reducing youths’ ability to cope and increasing
susceptibility to depression and suicidality (Sheeber et al., 1997).
Proposed Mechanism and Model of Depression and Suicidality for Latino Youth
The proposed process model describes a culturally-based etiological pathway to
depression and suicidality among Latino youth. The cultural and generational gap that
emerges between Latino youth and their parents during the immigration process
(Szapocznik et al., 1986; Szapocznik et al., 1989) is hypothesized to play an important
contributing role in increasing risk for these mental health problems. Immigrant Latinos
often migrate from cultures in which familial and generational roles are more clearly
defined and enforced than in the U.S. (Baptiste, 1993), with Latino parents endorsing
more restrictive, conformity-oriented, and obedience-focused child-rearing practices than
Anglo-Americans (Harwood et al., 1996; Okagaki & Sternberg, 1993; Varela et al.,
2004). As second-generation Latino youth acculturate, generational differences may
emerge as immigrant parents hold onto their native language and culture while their
children adopt the values, language, and conventions of mainstream society

15

(Christiansen, 1997; Elliott, 2001; Szapocznik et al., 1989). Such generational and
intercultural conflict may generate maladaptive family interaction patterns (Szapocznik et
al., 1989). Faced with a negative family environment, Latino youth may become
increasingly fatalistic and lose motivation to actively cope with the situation (Mirowsky
& Ross, 1984; Wheaton, 1980). Risk for depression and suicidal behavior subsequently
increases. The current study provided an empirical test of this theoretical model,
evaluating family dysfunction and fatalism as mediators of the association between
cultural discrepancy and adolescent depression/suicidality.
This study also evaluated a related theoretical perspective offered by Zayas and
colleagues in relation to adolescent Latinas. Acculturating Latinas may have a
particularly difficult time navigating tensions between the gender-typed
expectations of their culture of origin and the broader roles of the host American culture
(Zayas & Dyche, 1995; Zayas et al., 2000, Zayas et al., 2005). In flexible, “well-
functioning” Latino households, parents may permit a limited degree of female
autonomy, allowing for some measure of coexistence between traditional and mainstream
values thus easing potential sources of family tension (Canino, 1982; Zayas et al., 2000).
However, in more rigid, authoritarian households, acculturating Latinas may actively
challenge family hierarchy and parental restrictions, leading to greater family dysfunction
(Zayas et al., 2000). Given concerns that adherence to traditional gender roles may
constrain social opportunities for females (Garcia-Preto, 1998; Hines, Garcia-Preto,
McGoldrick, & Rhea, 1992), parent-child discrepancy in this domain may have greater
mental health consequences for girls than for boys. The current study provided an

16

empirical test of this theoretical model, evaluating gender as a moderator of the
association between gender role discrepancy and adolescent depression/suicidality.
The present study was an expansion of previous research exploring the contributing
role of cultural discrepancy to increased risk for depression among Latino adolescents
(Céspedes & Huey, 2008). Prior results showed that females reported greater levels of
gender role discrepancy and depression compared to males. Results also showed that the
relationship between gender role discrepancy and depression was mediated by increases
in family dysfunction (i.e., greater family conflict and reduced family cohesion).
Moderator analyses indicated that gender role discrepancy effects were most pronounced
for adolescent Latinas, although this interaction was only marginally significant.
Findings from this earlier study provided preliminary evidence that cultural discrepancy
plays an important contributing role to depressive symptomatology among Latino
adolescents and that discrepancy in gender role beliefs may have a more adverse effect on
females than males.
The present study sought to replicate these findings and to evaluate additional factors
previously excluded from the cultural process model. First, familism beliefs were tested
as a potential source of cultural discrepancy in immigrant Latino families. Second, the
effects of cultural discrepancy on both adolescent depression and suicidal behavior were
explored. Finally, fatalism was evaluated as a mediating variable of the association
between cultural discrepancy and adolescent depression/suicidality. In addition to
expanding the previous theoretical model in these three ways, the present study improved
upon the research design used by Céspedes and Huey (2008) by incorporating both youth
and parent reports and recruiting a larger and more representative student sample.

17

In summary, the present study had two primary aims addressing the relationship
between cultural factors and adolescent depression/suicidality. First, this study examined
whether family dysfunction and fatalism mediated the relationship between cultural
discrepancy and adolescent depression/suicidality (Figure 1). Second, the role of gender
as a moderator of the relationship between gender role discrepancy and
depression/suicidality was examined (Figure 2). These aims were tested using the
youth’s own reports of parent-child disparity in cultural values (i.e., perceived cultural
discrepancy). Moreover, supplementary analyses evaluated whether true cultural
discrepancy (i.e., using youth and caregiver report) yielded similar results. The specific
aims and hypotheses were as follows:
Aim 1. To evaluate factors mediating the relationship between cultural discrepancy and
depression/suicidality among adolescent Latinos.
Hypothesis 1a. Cultural discrepancy (i.e., acculturation discrepancy, gender role
discrepancy, and familism discrepancy) between adolescents and their caregivers
will be associated with diminished family functioning, higher fatalism, and increased
depression/suicidality.
Hypothesis 1b. Family dysfunction will be associated with higher fatalism and
increased depression/suicidality.
Hypothesis 1c. Fatalism will be associated with increased depression/suicidality.
Hypothesis 1d. Cultural discrepancy will contribute indirectly to increased
depression/suicidality through its effects on family dysfunction and fatalism.
Aim 2. To determine whether gender moderates the effects of gender role discrepancy on
adolescent depression/suicidality.

18

Hypothesis 2a. Gender role discrepancy will be greater for females than for males.
Hypothesis 2b. Family dysfunction will be greater for females than for males.
Hypothesis 2c. The association between gender role discrepancy and
depression/suicidality will be greater for females than for males.

19

Figure 1. Mediational model illustrating hypothesized regression paths among cultural discrepancy, family functioning, fatalism, and
outcome variables.

Outcome
-- RADS Depression
-- CES Depression
-- Past Ideation
-- Future Ideation
-- Future Attempt
-- Total Attempts Made
-- Overall Suicidality
-- Negative Thoughts

Family Dysfunction
-- High Conflict
-- Low Cohesion

Cultural Discrepancy
-- Assimilation
-- English Language Use
-- Familism
-- Machismo
-- Marianismo

Fatalism

20

Figure 2. Model illustrating hypothesized moderated relationship between cultural
discrepancy and outcome variables.

Outcome
-- RADS Depression
-- CES Depression
-- Past Ideation
-- Future Ideation
-- Future Attempt
-- Total Attempts Made
-- Overall Suicidality
-- Negative Thoughts

Cultural Discrepancy
-- Machismo
-- Marianismo

Gender

21

Chapter Two: Experimental Design and Methods
Participants
Participants were 395 ninth through twelfth grade Latino students enrolled at a high
school in Los Angeles. Although data was collected from an ethnically-diverse group of
457 youth, this study included only those students of Latino descent. Participants ranged
in age from 13 to 18 years (M = 15.25, SD = 1.21), with 50% of the sample being female
and 45% male. Five percent of youth did not report their gender. Forty-six percent of
youth self-identified as Mexican, 30% as Salvadoran, 10% as Guatemalan, 2% as
Honduran, and 3% as mixed Latino. Also, 4% percent self-identified as “Other Latino,”
and 4% percent did not further specify their ethnic background. Approximately 95% of
youth reported that both parents were immigrants to the United States, with another 3%
of youth reporting one immigrant parent. Of youth born outside the United States (n =
89), age of arrival ranged from 0 to 17 years (M = 8.90, SD = 5.41).
Approximately 18% of student caregivers participated in the study (n = 70).
Caregivers were defined as any parent or adult figure living in the home. Seventy-nine
percent of participating caregivers self-identified as the biological mother, 16% as the
biological father, 4% as a sibling, and 1% as the step-mother. Caregivers ranged in age
from 20 to 56 years (M = 40.78, SD = 7.38), with 67% of the sample being female and
19% male. Fourteen percent of participating caregivers did not report their gender.
Forty-three percent of caregivers self-identified as Mexican, 29% as Salvadoran, 13% as
Guatemalan, 3% as Honduran, and 1% as “Other Latino.” Eleven percent did not specify
their ethnic background. Ninety-four percent of caregivers were born outside the United
States, with age of arrival ranging from 4 to 39 years (M = 22.09, SD = 7.14).

22

Procedure
Consent for participation was obtained from each youth’s parent or primary
guardian. Consent packets were distributed to students in non-core academic classes
(e.g., health, physical education, electives), and students who returned a signed consent
form were invited to participate. These students were assented and administered the
questionnaire packet in a group setting during class. Permission was requested in
advance from all teachers to allow students to miss a portion of class.
Participating youth underwent a one-time assessment where they were asked to
complete a packet of questionnaires in a group setting. Youth completed measures of
depression, suicidality, family functioning, fatalism, acculturation, gender role beliefs,
and familism. Youth completed the measures of acculturation, gender role beliefs, and
familism from their own perspective and from the perspectives of both a female and male
caregiver. By having adolescents complete these measures for themselves and their
caregivers, a perceived discrepancy score between caregiver and youth was calculated for
each student (see below for details). The questionnaire packet took approximately 75
minutes to complete and youth were paid $10 for their participation.
As students submitted their completed questionnaire packets, each packet was
screened for acute suicidality by both an undergraduate research assistant and the
graduate student principal investigator. Questions from the depression and suicidality
measures were used as trigger items to identify students who were at high-risk for
suicidal behavior. These students were immediately referred to the Psychiatric Social
Worker at the high school, who agreed to meet with all referred students within 24 hours.

23

Upon completion of the questionnaires, students were given a packet of materials to
take home to their caregivers containing a cover letter, several questionnaires, and a
stamped envelope. The cover letter thanked caregivers for allowing their child to
participate in the study and asked caregivers to complete the enclosed questionnaires
(while clearly stating that participation was voluntary). Questionnaires assessed
acculturation status, gender role beliefs, and familism. Caregivers who returned a
completed questionnaire packet were paid $5 for their participation. By having
caregivers complete these measures, a true discrepancy score between parent and youth
reports was derived (see below for details).
Measures
Acculturation. Acculturation level was evaluated through self-report on the
Acculturation, Habits, and Interests Multicultural Scale for Adolescents (AHIMSA;
Unger et al., 2002) and the English Language Usage Scale (ELUS; Unger et al., 2002).
The AHIMSA is an eight-item, orthogonal measure of acculturation designed for use
with adolescents. It evaluates the degree to which an individual is assimilated, separated,
integrated, or marginalized. Items from this measure include “I am most comfortable with
people from…,” “The holidays I celebrate are from…,” and “The food I eat at home is
from….” For each item, subjects choose from one of four responses: “The United
States,” “The country my family is from,” “Both,” and “Neither.” The assimilation
subscale of the AHIMSA was utilized for this study as an index of acculturation to
mainstream American society. An assimilation score was obtained by summing the
number of “United States” responses. The AHIMSA has been shown to have high
concurrent validity, correlating with the Acculturation Rating Scale for Mexican-

24

Americans-II and the ELUS (Unger et al., 2002). Additionally, the AHIMSA correlates
highly with measures of generation status. Third-generation adolescents reported greater
assimilation than first- or second- generation adolescents (Unger et al., 2002). In the
present study, reliability analyses produced cronbach alphas of .74, .86, and .81 for the
youth, female caregiver, and male caregiver versions (i.e., versions based on youth
report), respectively.
The ELUS is a five-item scale assessing the degree to which an individual speaks
English or another primary language. Items from this scale include “What language do
you usually think in?” and “What language do you usually speak with your friends?”
Item responses are rated on a five-point Likert scale ranging from “English only” to
“Another language only.” An ELUS score was obtained by taking the average of all item
responses. The items on the ELUS were adapted from the Brief Acculturation Scale for
Hispanics, a scale that correlated highly with generation status and time spent in the
United States (Norris, Ford, & Bova, 1996). The ELUS was found to correlate highly
with the AHIMSA (Unger et al., 2002), and this correlation was also evident in the
present study. Reliability analyses produced alphas of .81, .95, and .93 for the youth,
female caregiver, and male caregiver versions, respectively.
Gender role beliefs. Gender role beliefs were evaluated using the short version of the
Attitudes toward Women Scale (AWS; Spence, Helmreich, & Stapp, 1973) and the
Machismo Scale (Cuéllar, Arnold, & González, 1995). The AWS is a 25-item measure
that assesses the extent to which individuals hold traditional views of the female role.
Items are rated on a four-point Likert scale. Sample items include “Women should worry
less about their rights and more about becoming good wives and mothers” and “Women

25

should have the same place in the job market as men.” Because the original scale was
designed for adults, item phrasing was simplified to facilitate adolescent comprehension
(e.g., “There should be a strict merit system in job appointment and promotion without
regard to sex” was changed to “The same rules should apply to both men and women in
getting a job or getting a raise”). In all cases, efforts were made to retain the original
meaning of the item. A few items were also altered to make the language more modern.
For example, the phrase “darn socks” was changed to “sew socks.” An AWS score was
obtained by taking the mean of all item responses. Reliability analyses in the present
study produced original alphas of .71, .59, and .67 for the youth, female caregiver, and
male caregiver versions, respectively. All alphas were improved by removing three items
from the AWS scale (“Being in charge of their money and relationships means more to
women than being feminine,” “Women who make as much money as their dates should
split the bill when they go out,” and “Women should be more responsible for solving
problems in today’s world”), yielding new alphas of .75, .66, and .72 for the youth,
female caregiver, and male caregiver versions, respectively. This adaptation of the AWS
scale was used for analyses.
The Machismo Scale is composed of 17 items measuring traditional male gender role
beliefs. Sample items include “There are many jobs that should go to men instead of
women,” “Men are smarter than women,” and “Wives should respect the man’s position
as head of the household.” As with the AWS scale, items from the Machismo Scale were
simplified in wording to facilitate adolescent comprehension. Although the original
measure used a true-false response format, a 4-point Likert scale was used for this study
to allow for consistency with the AWS and greater variation in student responses. A

26

Machismo score was obtained by taking the mean of all item responses. Cuéllar, Arnold,
and González (1995) found that the Machismo Scale was negatively correlated with
acculturation (r = -.24), although this finding was not replicated in the present study.
Reliability analyses in the present study produced alphas of .86 for the youth and female
caregiver versions, and .85 for the male caregiver version.
Familism. Familism was evaluated using the Familism Scale (Steidel & Contreras,
2003). The Familism Scale is an 18-item measure assessing familism beliefs. Items are
rated on a four-point Likert scale and include “A person should live near his or her
parents and spend time with them on a regular basis” and “A person should always be
expected to defend his/her family’s honor no matter what the cost.” A Familism score
was obtained by taking the mean of all item responses. Reliability analyses produced
alphas of .77 for the youth version and .79 for both the female and male caregiver
versions.
Family functioning. Family conflict and cohesion were evaluated through self-report
on the Conflict subscale of the Family Environment Scale (FES; Moos & Moos, 1981)
and the Cohesion subscale of the Family Adaptability and Cohesion Evaluation Scale
(FACES-III; Olson, Portner, & Lavee, 1985). The Conflict subscale consists of nine
items with a true/false response format. Items from this subscale include “We fight a lot
in our family” and “Family members often criticize each other.” A Conflict score was
obtained by taking the sum of all item responses. A reliability analysis produced an alpha
of .71 for the Conflict subscale.
The Cohesion subscale consists of ten items rated on a five-point Likert scale. Items
from this subscale include “We like to do things with just our immediate family” and

27

“Family members consult other family members on their decisions.” A Cohesion score
was obtained by taking the mean of all item responses. A reliability analysis produced an
alpha of .85.
Both the Conflict and Cohesion subscales have been shown to discriminate between
adolescents with and without psychopathology such as depression and suicidality
(Asarnow, Carlson, & Guthrie, 1987; De Wilde et al., 1993; Asarnow & Carlson, 1988;
Rubenstein et al., 1989).
Fatalism. Fatalism was evaluated through the Relinquished Control subscale of the
Primary-Secondary Control Scale (PSQ; Weisz, 1992). The Relinquished Control
subscale contains ten items rated on a four-point Likert scale. Items include “When
things are bad, there is usually nothing I can do about them” and “When things don’t go
my way, I usually just give up.” A fatalism score was obtained by taking the mean of all
item responses. Relinquished control has been identified as a maladaptive form of coping
and predicts poor psychosocial adjustment in children (Weisz, McCabe, & Dennig,
1994). A reliability analysis in the present study produced an alpha of .82 for the
Relinquished Control subscale.
Depression. Depression was evaluated through self-report on the Reynolds
Adolescent Depression Scale-2 (RADS-2; Reynolds, 2002) and the Center for
Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977). The RADS-2 is a 30-
item measure rated on a four-point Likert scale. Items include “I feel happy” and “I feel
life is unfair.” A total depression score was obtained by taking the mean of item
responses. The scale has been used in other studies to assess depression in Latino

28

populations (Aisenberg, Trickett, Mennen, Saltzman, & Zayas, 2007; Hovey & King,
1996). A reliability analysis in the present study yielded a coefficient alpha of .89.
The CES-D consists of 20 items rated on a four-point Likert scale. Items include
“During the past week I felt depressed,” “During the past week I felt hopeful about the
future,” and “During the past week my sleep was restless.” A total depression score was
obtained by taking the mean of all item responses. The CES-D has been shown to be
suitable for high school populations (Radloff, 1991) and appropriate for use with Latino
youth (Crockett, Randall, Shen, Russell, & Driscoll, 2005). A reliability analysis in the
present study produced an alpha of .88.
Suicidality. Suicidality was evaluated through self-report on the Suicidal Behaviors
Questionnaire (SBQ-14; Linehan, 1996) and the Negative Thoughts subscale of the
Positive and Negative Suicide Ideation Inventory (PANSI; Osman et al., 1998). An
abbreviated version of the SBQ-14 was utilized for this study. This version contains nine
core questions -- five with follow-up questions that probe suicidality at specific time
intervals (e.g., “in the last year”). Item stems include “How often have you thought about
killing yourself” and “What chance is there that you will attempt suicide?” The SBQ-14
yields five indices of suicidality: past suicidal ideation, likelihood of future suicidal
ideation, likelihood of future suicide attempt, number of total attempts made, and an
overall index of suicidality. The overall index of suicidality was obtained by summing
item responses across multiple suicide domains following scoring guidelines provided by
Marsha Linehan. The full version of the SBQ-14 was found to have good convergent
validity in a sample of high school adolescents, correlating with measures of depression,
hopelessness, and potential suicide (Cole, 1989). In a sample of adolescent psychiatric

29

inpatients, current ideation and likelihood of future suicidal behavior were negatively
associated with survival and coping beliefs (Osman et al., 1996). Reliability analyses
produced an alpha of .89 for the past ideation, future ideation, and future attempt indices
and an alpha of .95 for the overall suicidality index. Cronbach’s alpha was not calculated
for the total number of attempts index, as the total attempts score was simply the sum of
all previous attempts made.
The Negative Thoughts subscale of the PANSI consists of eight items rated on a five-
point Likert scale. Items include “During the past two weeks, how often have you felt
hopeless about the future and wondered if you should kill yourself” and “During the past
two weeks, how often have you felt so lonely or sad you wanted to kill yourself so that
you could end your pain.” The PANSI was shown to have good predictive validity for a
group of adolescent psychiatric inpatients, with suicide attempters scoring significantly
higher on negative thoughts than non-suicidal or at-risk youth (Osman et al., 2002). A
reliability analysis in the present study produced an alpha of .90.
Cultural discrepancy. Measures of cultural discrepancy were derived using a
generalized difference approach (Lau, et al., 2005; Crane, Ngai, Larson & Hafen, 2005),
in which caregiver values were subtracted from youth values. Measures of true
discrepancy were based on youth self-report and caregiver self-report. Measures of
perceived discrepancy were based on youth self-report and youth report on caregivers.
Youth reported on both male and female caregivers, yielding two parent scores from
which to calculate perceived discrepancy. Because the two parent scores in each cultural
domain (e.g., assimilation) were significantly correlated (Table 1), scores were collapsed
using a robust principal components analysis (Maronna, 2005). The resulting joint

30

caregiver composite was utilized to calculate perceived cultural discrepancy for each
cultural domain. Unless otherwise noted, use of the term “cultural discrepancy” refers to
perceived cultural discrepancy.
Table 1. Correlations Between Male and Female Caregiver Scores on Indices of Cultural
Discrepancy
Male
Caregiver
Assimilation
Male
Caregiver
ELU
Male Caregiver
Familism
Score
Male
Caregiver
Machismo
Male
Caregiver
Marianismo
Female Caregiver
Assimilation Score
.929* --- --- --- ---
Female Caregiver
ELU Score
--- .909* --- --- ---
Female Caregiver
Familism Score
--- --- .842* --- ---
Female Caregiver
Machismo Score

--- --- --- .588* ---
Female Caregiver
Marianismo Score
--- --- --- --- .636*
Note. Male and female caregiver scores are based on youth report. ELU = English Language Usage.
*p < .05.

31

Chapter Three: Analyses
Robust statistical methods were utilized for this study as they perform well under
conditions of non-normality and heteroscedasticity. Such methods are generally
preferable to classical hypothesis testing techniques based on least squares estimators as
they offer greater power and stronger control over Type I error (Hampel, Ronchetti,
Rousseeuw, & Stahel, 1986; Huber, 1981; Staudte & Sheather, 1990; Wilcox, 2003;
Wilcox 2005).
Bootstrap methods were utilized to estimate the sampling distribution of various
statistics (e.g., sample median). A bootstrap sample is obtained by randomly sampling,
with replacement, values from the original sample (Wilcox, 2003). Analyses utilized a
minimum of 100 and a maximum of 2000 bootstrap samples when calculating a sampling
distribution. This bootstrap technique was also utilized to estimate confidence intervals
associated with the statistic being tested.
Prior to statistical analyses, all data were examined for missing values and outliers.
Missing data was generally minimal and non-systematic for any given item in the
questionnaire packet. However, four total scores could not be calculated for a substantial
proportion of participants (>10%): male caregiver ELUS (14.4%), male caregiver
familism (14.7%), male caregiver machismo (17.5%), and male caregiver AWS (16.2%).
Total scores for these measures could not be calculated because all items on the measures
were left blank. This was largely attributed to students’ written statements that a male
caregiver was not living in the household and/or they had never known the male
caregiver.

32

Although there are a variety of strategies for imputing missing data (e.g., Afifi &
Elashoff, 1966), no published papers support the use of robust methods when such
techniques are applied. Robust estimators are not designed to deal with violations of the
random sampling assumption, which requires values to be independent and identically
distributed. Therefore, missing values were excluded from analyses.
R software was used to conduct statistical analyses using functions described in
Wilcox (2003, 2005) and one function developed by Wilcox specifically for this study.
Outliers were identified using two statistical techniques: a projection method and the
minimum volume ellipsoid estimator, both of which take into account the overall
structure of the data (Wilcox, 2003).
Group Differences
Independent group differences were explored through comparisons of sample
medians and probabilities. The R function MEDPB was used to compare group scores on
cultural discrepancy, family functioning, fatalism, and depression variables. This
function uses a percentile bootstrap method to compute the confidence interval for the
difference between two sample medians. If the confidence interval does not contain zero,
the group scores are significantly different at the .05 level.
The R functions TWOBINOM and TWOBICI were used for probability comparisons
on suicidality variables. Robust estimates of the interquartile range, a measure of
variation derived from the center portion of the data, reflected values of zero on all six
measures of suicidality. This finding indicates that most students endorsed few, if any,
suicidality items. Such limited variability in student responses prevented valid
comparisons on measures of central tendency (e.g., median). Thus, each suicidality index

33

was dichotomized, with a score of zero indicating no suicidality and a score greater than
zero indicating endorsement of at least one suicidality item. Suicidality variables were
dichotomized for group comparisons only; correlation and regression analyses tested
suicidality as continuous variables.
TWOBINOM tests the hypothesis that two independent binomials have an equal
probability of success (e.g., the hypothesis that males and females have an equal
probability of endorsing at least one suicidality item on a given suicidality measure). A
significance level is generated indicating whether the probabilities are different.
TWOBICI computes the confidence interval for the difference between the probabilities,
which provides an alternative measure of the extent to which the probabilities differ.
To explore group differences in greater detail, the R function SBAND was utilized.
SBAND computes a confidence band for the difference in quantile scores between two
independent groups. A plot of the shift function (i.e., the differences between all quantile
scores) is also created, with the median and quartiles indicated. If independent group
scores were significantly different, the confidence band and plot were used to locate at
what values the effects were strongest.
Associations among Key Variables
Correlations among variables were computed using R functions SCOR and PCORB.
Both functions compute Pearson’s coefficient; however, PCORB does not remove
outliers in the distribution. Given the limited variation in responses to suicidality items
(i.e., most students endorsed little to no suicidality), including extreme values on
measures of suicidality was necessary to test for associations with key variables in the
study. For this reason, PCORB was utilized when computing correlations with

34

suicidality data. SCOR, which removes outliers using a projection method, was utilized
to calculate correlations for all other variables.
Mediational Model
In order to test the hypothesized mediational model (Figure 1), a series of regression
analyses using individual variables was conducted. This method of analysis was
preferred over structural equation modeling (SEM) given concerns that correlation
matrices utilized in classical SEM techniques are not robust to violations of assumptions
about the distribution of the data (Wilcox, 2003). The magnitude of the correlation
between two variables is greatly impacted by outliers and non-normal distributions,
which are common in the social sciences (Wilcox, 1996). While some procedures have
been proposed to address the issue of outliers when using SEM (Yuan & Bentler, 1998),
the assumption of elliptical distributions continues to pose problems.
By choosing to run multiple regression equations, the full mediational model
hypothesized in Figure 1 could not be tested simultaneously. Instead, the model was
broken down into four separate mediator analyses: 1) cultural discrepancy family
dysfunction fatalism, 2) cultural discrepancy family dysfunction
depression/suicidality, 3) cultural discrepancy fatalism depression/suicidality, and
4) family dysfunction fatalism depression/suicidality. Cultural discrepancy was
represented by five variables: assimilation discrepancy, English language usage
discrepancy, familism discrepancy, machismo discrepancy, and marianismo discrepancy.
Family dysfunction was represented by two variables: family conflict and family
cohesion. Fatalism was represented by one variable only: fatalistic beliefs.
Depression/suicidality was represented by eight variables: RADS depression, CES

35

depression, past suicidal ideation, likelihood of future suicidal ideation, likelihood of
future suicide attempt, total attempts made, overall suicidality, and negative thoughts
about suicide. As RADS and CES depression were highly correlated (r = .821), and to
reduce the total number of regression paths tested, a robust principal components analysis
was utilized to collapse the depression measures into a composite variable. Given the
limitations of SEM mentioned above, robust principal components analysis is a preferred
method for collapsing highly correlated variables (Maronna, 2005). The depression
composite was utilized for subsequent regression analyses.
Baron and Kenny (1986) identified four criteria that must be met for mediation to
occur: a) a significant association must exist between the predictor and outcome
variables, b) a significant association must exist between the predictor and mediator
variables, c) a significant association must exist between the mediator and outcome
variables, and finally, d) there must be a subsequent reduction in significance in the
association between the predictor and outcome when the mediating variable is controlled.
R function LSFITCI was utilized to test for the first three criteria. This function uses an
adjusted percentile bootstrap method to compute a confidence interval for the slope
parameter. If the confidence interval does not contain zero, the regression path is
significant at the .05 level. Eighty-two regression paths, based on all possible pairs of
key variables (e.g., assimilation discrepancy and family conflict, assimilation discrepancy
and family cohesion, etc.), were tested.
Because the R program did not have a function for testing Baron and Kenny’s (1986)
final criterion, Wilcox developed the function MEDIATE to approximate mediation
testing. MEDIATE uses a modified percentile bootstrap method to compare the

36

association between the predictor and outcome variables before and after adjusting for the
mediator. This technique is based on a model proposed by MacKinnon and colleagues
(MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002). The function MEDIATE
estimates the confidence interval for the difference between the original (t) and adjusted
(t') regression coefficients, thereby providing an estimate of whether the regression
coefficient is reduced once adjusting for the mediator. If the regression coefficient is
reduced, the difference between the original and adjusted coefficients should be positive
(t - t' > 0). If the confidence interval generated contains only positive values, a
mediational effect is inferred. The R function REGCI was subsequently used to compute
a confidence interval for the adjusted regression coefficient (t') in order to determine
whether results were consistent with partial or full mediation. If the confidence interval
is centered around zero, it is inferred that the adjusted regression coefficient was reduced
to zero. Such results would be consistent with full mediation. While a confidence
interval containing zero, but not centered around zero, does not exclude the possibility of
full mediation, such a result is more consistent with partial mediation.
Moderator Model
In order to test for moderator effects, the R functions REG2CI and RUNMEAN2G
were utilized. REG2CI computes a confidence interval for the difference between the
slopes corresponding to two independent groups. If the confidence interval does not
contain zero, the slopes are significantly different at the .05 level, indicating a moderated
effect. RUNMEAN2G creates a plot of the running interval smoother for the two
independent groups, which provides a visual illustration of the interaction effect.

37

Supplementary Analyses
Associations with true cultural discrepancy. Supplementary analyses were conducted
to explore associations between true cultural discrepancy and other key variables in order
to evaluate whether measures of true discrepancy yielded results similar to those of
perceived discrepancy. Correlational analyses using the SCOR and PCORB functions
were utilized.
Predictors of suicidality. Given the limited variation in responses to suicidality items
and the severe skewness of the distribution, supplementary analyses were utilized to
identify significant predictors of high values on suicidality measures. A quantile
regression method, R function QREGCI, was utilized for this purpose. This function
provides a confidence interval for the slope parameter at specified quantiles. In other
words, this approach evaluates the extent to which a variable X predicts an outcome Y,
when restricting Y values to a specific quantile. The .75 quantile was utilized in order to
restrict analyses to high values. If the confidence interval does not contain zero, the
regression path is significant at the .05 level.
Regression surface. Plots of the regression surface were examined to identify
potential nonlinear relationships among key variables that might have influenced
mediator and moderator analyses. R function LPLOT was used to create three-
dimensional plots illustrating the pattern of associations among sets of three variables
(i.e., predictor, mediator, and outcome).
Explanatory power. Traditional mediation testing as proposed by Baron and Kenny
(1986) and MacKinnon et al. (2002) makes several assumptions about the distribution of
the data, such as normality, homoscedasticity, and linear associations among the

38

variables. The bootstrap mediation test used in this study addresses issues of
heteroscedasticity, and to some extent, non-normality. However, this method continues
to assume that a linear model best fits the data. A technique based on explanatory power,
R function EPMOD, was utilized to provide an alternative way of exploring hypothesized
relationships without assuming linear associations. This function estimates the
explanatory power for a set of predictor variables, similar to providing an estimate of the
proportion of variance explained by a predictor variable. Greater explanatory power
reflects a greater proportion of variance accounted for.

39

Chapter Four: Results
Group Differences
Analyses were first conducted to determine whether Mexican and Central-American
youth differed on key variables. Marginally significant differences were found for
assimilation discrepancy and ELU discrepancy, although no consistent pattern was found:
Mexican youth reported somewhat higher ELU discrepancy (p = .079) and Central-
American youth reported somewhat higher assimilation discrepancy (p = .068). Analyses
were also conducted to compare families who returned a parent questionnaire with those
who did not. No significant differences were found on any key variables.
Perceived cultural discrepancy. Analyses were conducted to determine whether
males and females differed significantly on measures of perceived cultural discrepancy.
Consistent with hypotheses, significant gender differences were found for machismo
discrepancy, with females reporting greater discrepancy with caregivers compared to
males (Table 2). A plot of the shift function (Figure 3) illustrated that there was little
difference in discrepancy quantiles at low values, but that the difference became more
pronounced and uniform at values greater than the median. No gender effects were found
for other measures of perceived cultural discrepancy.
Family functioning and fatalism. Group comparisons were conducted to determine
whether males and females differed significantly on measures of family functioning and
fatalism. Significant gender differences were found for family cohesion, with females
reporting lower levels of cohesion compared to males (Table 2). A plot of the shift
function (Figure 4) indicated that the difference in quantiles was most pronounced at
values around the median. However, the plot also showed that the difference in quantiles

40

Table 2. Comparison of Male and Female Scores on Cultural Discrepancy, Family
Functioning, Fatalism, and Depression Measures
Measure
Male
Median
Female
Median
Difference of
Medians
Lower CI Upper CI Sig.
Familism
Discrepancy
0.278 0.278 0.000 -0.003 0.089 .106
Marianismo
Discrepancy
0.160 0.200 .040 -0.060 0.060 .982
Machismo
Discrepancy
0.294 0.529 .235 0.021 0.254 .024
Assimilation
Discrepancy
1.000 1.200 .200 -1.000 0.500 .313
ELU
Discrepancy
1.300 1.000 -.300 -0.400 0.100 .266
Cohesion 3.400 3.100 -.300 -0.450 -0.100 .007
Conflict 2.000 3.000 1.000 0.000 1.000 .218
Fatalism 2.000 2.100 0.100 0.000 0.300 .174
RADS Depression 1.821 2.016 .195 .074 .326 .005
CES Depression .750 1.000 .250 .100 .400 .001
Note. ELU = English Language Usage; RADS = Reynolds Adolescent Depression Scale; CES = Center for
Epidemiologic Studies; CI = Confidence Interval; Sig. = Significance Level.

became negative as extreme values were approached, suggesting higher cohesion scores
among females at high and low values. Group comparisons on a second measure of
central tendency (20% trimmed mean) confirmed that female cohesion scores were lower
than male scores (p = .013). No significant gender effects were found for family conflict
or fatalism.
Depression. Analyses were conducted to determine whether males and females
differed significantly on measures of depression and suicidality. Consistent with
previous research, significant gender differences were found for both the RADS and
CES-D, with females reporting significantly higher levels of depression on both measures
(Table 2). Plots of the shift functions for RADS and CES depression (Figures 5 and 6,

41

Figure 3. Plot of the shift function for male and female machismo discrepancy scores.
Female scores were subtracted from male scores to create this plot. The shift function
(solid line) shows that the difference between the quantiles increases as you approach the
median (indicated by a + along the x-axis) and remains generally uniform at higher
values. The dashed lines mark the confidence bands for the shift function.

Figure 4. Plot of the shift function for male and female cohesion scores. Female scores
were subtracted from male scores to create this plot. The shift function (solid line)
shows that the difference in quantiles is positive at values near the median (indicated by a
+ along the x-axis), but negative at low and high values. This suggests that female scores
are lower than males at values near the median, but are higher than males as you
approach extreme values. The dashed lines mark the confidence bands for the shift
function.

42

Figure 5. Plot of the shift function for male and female RADS scores. Female scores
were subtracted from male scores to create this plot. The shift function (solid line) shows
that the difference between the quantiles increases as you approach the median (indicated
by a + along the x-axis) and remains generally uniform at higher values. The dashed
lines mark the confidence bands for the shift function.

Figure 6. Plot of the shift function for male and female CES-D scores. Female scores
were subtracted from male scores to create this plot. The shift function (solid line) shows
that the difference between the quantiles increases as you approach the median (indicated
by a + along the x-axis) but decreases as you approach higher values. The dashed lines
mark the confidence bands for the shift function.

43

respectively) illustrate that the difference in depression scores is small at low values and
becomes most pronounced and uniform at values greater than the median.
Separate analyses were conducted to determine whether males and females differed
significantly in the extent to which they met clinical criteria for depression on the CES-D.
Scores greater than 16 on the CES-D indicate a clinically significant level of depression
(Radloff, 1977). The probability of obtaining a score greater than 16 was compared for
males and females. A significant gender difference was found, with estimated
probabilities of .574 for females and .408 for males (p<.001). These results indicate that
females have a 57.4% probability of meeting the clinical cutoff, whereas males have a
40.8% probability.
Suicidality. Analyses were conducted to compare male and female probabilities of
endorsing at least one item on each suicidality scale. A significant gender difference was
found for two indices of suicidality, with females more likely to endorse past ideation and
overall suicidality (Table 3). Marginally significant gender differences were found for
both measures of attempted suicide, with females somewhat more likely to make a future
suicide attempt and to have made at least one attempt (Table 3). No significant gender
differences were found on an index assessing the likelihood of future ideation or an index
of negative thoughts about suicide.
Associations among Key Variables
Correlational analyses were conducted to identify significant associations among key
variables (Table 4). Assimilation discrepancy was significantly associated with higher
family conflict. Familism discrepancy was significantly associated with higher family
conflict, lower family cohesion, higher RADS and CES depression, and greater number

44

Table 3. Comparison of Male and Female Probabilities on Suicide Measures
Measure Male
Est. Prob.
Female
Est. Prob.
Lower
CI
Upper
CI
Sig.
SBQ
Overall Suicidality

.318

.388

.042

.178

.000
Past Ideation .267 .471 .090 .313 .000
Future Ideation .212 .254 -.058 .140 .262
Future Attempt .094 .159 -.014 .141 .052
Total Attempts Made .171 .264 -.020 .202 .068
PANSI
Negative Thoughts 1.000 1.000 .000 .002 1.00
Note. Probabilities reflect the likelihood that males (females) endorse at least one item on the specified
suicidality measure. SBQ = Suicidal Behaviors Questionnaire; PANSI = Positive and Negative Suicide
Ideation Inventory; Est. Prob. = Estimated Probability; CI = Confidence Interval; Sig. = Significance Level.

of suicide attempts made. Machismo discrepancy was significantly associated with
higher CES depression. AWS discrepancy was significantly associated with higher
family conflict and greater likelihood of future suicidal ideation. ELU discrepancy was
not associated with other key variables in the study.
Family conflict was significantly associated with increased fatalism, greater RADS
and CES depression, greater likelihood of future suicidal ideation, and greater negative
thoughts about suicide. Family cohesion was significantly associated with increased
conflict, increased fatalism, greater RADS and CES depression, and greater negative
thoughts about suicide. Fatalism was significantly associated with greater RADS and
CES depression, higher levels of past suicidal ideation, greater likelihood of future
suicidal ideation, greater likelihood of future suicide attempts, and greater negative
thoughts about suicide.

45

Table 4. Correlations Between Cultural Discrepancy, Family Functioning, Fatalism, Depression, and Suicidality Variables
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
(1) Assimilation
Discrepancy --- .206 .066 .005 .129* -.032 .120* .004 .070 -.100 .088 -.088 -.063 -.106 -.038 -.055
(2) ELU
Discrepancy --- .025 .072 .051 -.074 -.014 .000 .039 -.059 -.041 -.037 -.041 -.027 -.036 .002
(3) Familism
Discrepancy

---

.113

.127* -.225* .195* .081 .126* .118* -.014 .106 .139 .138* -.016 .136
(4) Machismo
Discrepancy

---

.412* -.110 .105 .029 .067 .121* -.011 .059 .034 .012 -.019 .061
(5) Marianismo
Discrepancy

--- -.031 .122* .027 .070 -.020 -.031 .226* .019 .098 -.001 .029
(6) Family
Cohesion

--- -.384* -.247* -.289* -.345* .104 .088 .098 -.130 .107 -.172*
(7) Family
Conflict

--- .237* .438* .390* .064 .137* .034 .134 .063 .270*
(8) Fatalism
--- .397* .385* .124* .165* .203* .032 .002 .218*
(9) RADS
Depression

--- .821* .094 .039 .031 .247* .042 .467*
(10) CES
Depression

--- .038 -.009 -.018 .265* -.001 .544*
(11) Past Suicidal
Ideation

--- .997* .996* -.008 .778* .005
(12) Future Suicidal
Ideation

--- .998* -.007 .939* .016
(13) Future Suicidal
Attempt

--- -.007 .939* .005
(14) Total Attempts
Made

--- -.002 .350*

46

Table 4, Continued
(15) Overall
Suicidality

--- .013
(16) Negative
Thoughts

---
Note. Discrepancy variables reflect perceived discrepancy between youth and caregivers. Discrepancy scores were calculated using joint caregiver
scores. ELU = English Language Usage

47

Mediator Analyses
Eight sets of relationships were found to meet the first three mediation criteria (Baron
& Kenny, 1986), in which significant regression paths were found among the predictor,
mediator, and outcome variables (Figures 7-14): 1) Familism discrepancy Conflict
Depression composite, 2) Familism discrepancy Conflict Future suicidal ideation,
3) Familism discrepancy Cohesion Depression composite, 4) Conflict Fatalism
Depression composite, 5) Conflict Fatalism Future suicidal ideation, 6) Conflict
Fatalism Negative suicidal thoughts, 7) Cohesion Fatalism Depression
composite, and 8) Cohesion Fatalism Negative suicidal thoughts.

Figure 7. Regression paths for familism discrepancy conflict depression composite
mediational model. Values outside parentheses are unstandardized regression
coefficients. Values inside parentheses are the confidence interval of the slope. *p < .05.

Conflict
Familism
Discrepancy
Depression
Composite
2.116 (1.280, 3.074)* .126 (.099, .153)*
.308 (.043, .594)*

48

Figure 8. Regression paths for familism discrepancy conflict future ideation
mediational model. Values outside parentheses are unstandardized regression
coefficients. Values inside parentheses are the confidence interval of the slope.
*p < .05.

Figure 9. Regression paths for familism discrepancy cohesion depression
composite mediational model. Values outside parentheses are unstandardized regression
coefficients. Values inside parentheses are the confidence interval of the slope. *p < .05.

Conflict
Familism
Discrepancy
Future Ideation
2.116 (1.280, 3.074)* .943 (.194, 2.425)*
3.809 (.867, 7.843)*
Cohesion

Familism
Discrepancy
Depression
Composite
-.808 (-1.109, -.503)* -.278 (-.363, -.190)*
.308 (.043, .594)*

49

Figure 10. Regression paths for conflict fatalism depression composite mediational
model. Values outside parentheses are unstandardized regression coefficients. Values
inside parentheses are the confidence interval of the slope.
*p < .05.

Figure 11. Regression paths for conflict fatalism future ideation mediational model.
Values outside parentheses are unstandardized regression coefficients. Values inside
parentheses are the confidence interval of the slope. *p < .05.

Fatalism
Conflict
Depression
Composite
.047 (.022, .074)* .360 (.250, .485)*

.126 (.099, .153)*

Fatalism
Conflict
Future Ideation
.047 (.022, .074)* 1.411 (.300, 2.921)*

.943 (.194, 2.425)*

50

Figure 12. Regression paths for conflict fatalism negative thoughts mediational
model. Values outside parentheses are unstandardized regression coefficients. Values
inside parentheses are the confidence interval of the slope. *p < .05.

Figure 13. Regression paths for cohesion fatalism depression composite
mediational model. Values outside parentheses are unstandardized regression
coefficients. Values inside parentheses are the confidence interval of the slope.
*p < .05.

Fatalism
Cohesion
Depression
Composite
-.134 (-.221, -.053)*

-.278 (-.363, -.190)*
.176 (.087, .284)*
Fatalism
Conflict Negative Thoughts
.047 (.022, .074)* .176 (.087, .284)*

.056 (.033, .082)*

51

Figure 14. Regression paths for cohesion fatalism negative thoughts mediational
model. Values outside parentheses are unstandardized regression coefficients. Values
inside parentheses are the confidence interval of the slope.
*p < .05.

A modified percentile bootstrap method was used to determine whether the
association between the predictor and outcome was reduced when controlling for the
mediator, the final criterion for mediation proposed by Baron and Kenny (1986). The
eight sets of relationships were tested using this bootstrap mediation technique. Results
showed that the confidence intervals for the difference between the original (t) and
adjusted (t') regression coefficients contained only positive integers for six of the eight
models (Table 5). Confidence intervals containing only positive values are consistent
with a mediated effect, as the regression coefficient was reduced (i.e., t – t' > 0) when the
mediator was controlled.
In order to determine the extent of mediation (i.e., partial versus full mediation) for
these six models, confidence intervals for the adjusted regression coefficient (Table 6)
were calculated. Confidence intervals were not centered around zero, suggesting that
Fatalism
Cohesion Negative Thoughts
-.134 (-.221, -.053)*

-.105 (-.178, -.037)*
.176 (.087, .284)*

52

Table 5. Confidence Intervals for Differences Between Regression Coefficients (t- t') for
Hypothesized Mediational Models
Mediational Model
Lower
CI
Upper
CI
Mediated
Effect?
Familism discrepancyConflictDepression composite .148 .393 Yes
Familism discrepancyConflictFuture suicidal ideation .089 1.261 Yes
Familism discrepancyCohesionDepression composite .114 .347 Yes
ConflictFatalismDepression composite .006 .023 Yes
ConflictFatalismFuture suicidal ideation .002 .142 Yes
ConflictFatalismNegative suicidal thoughts .002 .013 Yes
CohesionFatalismDepression Composite -.077 -.016 No
CohesionFatalismNegative suicidal thoughts -.045 -.006 No
Note. Confidence intervals containing positive integers only are consistent with a mediated effect.
CI = Confidence Interval.

Table 6. Confidence Intervals for Adjusted Regression Coefficients (t') for Hypothesized
Mediational Models
Mediational Model
Lower
CI
Upper
CI
Full
Mediation?
Familism discrepancyConflictDepression composite -.221 .325 No
Familism discrepancyConflictFuture suicidal ideation No
Familism discrepancyCohesionDepression composite
.038
-.175
.523
.366 No
ConflictFatalismDepression composite .083 .136 No
ConflictFatalismFuture suicidal ideation .069 .586 No
ConflictFatalismNegative suicidal thoughts .029 .071 No
Note. Confidence intervals centered around zero are consistent with full mediation. CI = Confidence
Interval.

findings for all six models were consistent with partial mediation. Figure 15 presents a
summary of the mediational findings.
Moderator Analyses
Gender was hypothesized to moderate the relationship between gender role
discrepancy and depression/suicidality. In order to evaluate moderator effects, the slopes

53

Figure 15. Models summarizing mediational results for hypothesized cultural process
model. Solid lines represent paths that received empirical support. Dashed line
represents path that was not supported. Faded areas reflect the portion of the
hypothesized cultural process model excluded from that set of mediated relationships.
Note. The association between familism discrepancy and future ideation was mediated by
conflict only.

Family Functioning
-- High Conflict
-- Low Cohesion
Cultural Discrepancy
-- Familism

Fatalism
Outcome
-- Depression Composite
-- Future Ideation

Family Functioning
-- High Conflict

Cultural Discrepancy

Fatalism
Outcome
-- Depression Composite
-- Future Ideation
-- Negative Thoughts

54

of the regression lines for males and females were compared to detect the presence of an
interaction effect (Holmbeck, 1997; Wilcox, 2003). Gender was tested as a moderator for
all combinations of gender role discrepancy and outcome associations (e.g., marianismo
discrepancy and RADS depression, marianismo discrepancy and CES depression, etc.)
The only significant interaction effect was found for the association between machismo
discrepancy and CES depression. A .95 confidence interval for the difference between
the two slopes ranged from .000 to .529 (p = .045), indicating unequal slopes. A plot
illustrating the slopes of the regression lines showed that, while there is a negative
association between discrepancy and depression for males, there is a positive association
for females (Figure 16).

Figure 16. Machismo discrepancy and gender interaction effect on CES depression.

____ Female x = Machismo Discrepancy
------ Male y = CES Depression

55

Supplemental Analyses
Associations with true cultural discrepancy. Supplementary analyses were conducted
to explore associations with measures of true cultural discrepancy. True discrepancy
scores in each cultural domain were significantly correlated with the corresponding
perceived discrepancy score, except for machismo discrepancy (r = -.066, ns, Table 7).

Table 7. Correlations Between True and Perceived Measures of Cultural Discrepancy
Perceived
Assimilation
Discrepancy
Perceived
ELU
Discrepancy
Perceived
Familism
Discrepancy
Perceived
Machismo
Discrepancy
Perceived
Marianismo
Discrepancy
True Assimilation
Discrepancy
.777* --- --- --- ---
True ELU
Discrepancy
--- .673* --- --- ---
True Familism
Discrepancy
--- --- .320* --- ---
True Machismo
Discrepancy

--- --- --- -.066 ---
True Marianismo
Discrepancy
--- --- --- --- .455*
Note. Perceived discrepancy was calculated using joint caregiver scores. ELU = English Language
Usage. *p < .05.

True familism discrepancy was associated with decreased family cohesion, higher levels
of previous suicidal ideation, greater likelihood of future ideation, greater likelihood of
future attempt, greater overall suicidality, and higher levels of negative thoughts about
suicide (Table 8). However, scatter plots of the individual X and Y values suggest that
these associations are generally weak, with Pearson’s coefficient driven by the effect of
one or two outlying points (e.g., Figure 17). Other measures of true cultural discrepancy
were generally unassociated with key variables in the study with one exception: true
ELUS discrepancy was associated with lower CES depression (Table 8). The overall

56

Table 8. Correlations Between True Cultural Discrepancy, Family Functioning, Fatalism,
Depression, and Suicidality Variables
(1) (2) (3) (4) (5)
(1)
Assimilation Discrepancy
---
(2)
ELU Discrepancy
-.001 ---
(3)
Familism Discrepancy
.010 .033 ---
(4)
Machismo Discrepancy
.132 -.091 .106 ---
(5)
Marianismo Discrepancy
-.044 .136 .023 .042 ---
(6)
Family Cohesion
.087 -.093 -.432* -.055 -.007
(7)
Family Conflict

.038 -.192 .123 -.046 -.001
(8)
Fatalism
.078 -.065 -.099 -.086 -.030
(9)
RADS Depression
.113 -.211 .015 -.236 -.041
(10)
CES Depression
-.083 -.329* .176 -.071 -.051
(11)
Past Suicidal Ideation
-.066 .133 .327* -.054 .070
(12)
Future Suicidal Ideation
-.099 .149 .469* .001 .052
(13)
Future Suicidal Attempt
-.042 .201 .548* .163 -.010
(14)
Total Suicide Attempt
-.034 .190 -.007 .020 .079
(15)
Overall Suicidality
-.103 .202 .470* .009 .082
(16)
Negative Thoughts
-.110 .222 .329* -.009 .039
Note. Discrepancy variables reflect true discrepancy between youth and caregivers. ELU = English
Language Usage; RADS = Reynolds Adolescent Depression Scale; CES = Center for Epidemiologic
Studies. *p < .05.

lack of significant findings is not surprising given the low rate of caregiver participation
(n = 70) and resulting low power (1-β = .550; Faul & Erdfelder, 2007).

57

Figure 17. Two-dimensional plot of the regression surface illustrating the pattern of
association between true familism discrepancy and future suicide attempt.

Predictors of suicidality. Quantile regression analyses were utilized to further
explore associations between key variables and suicidality. Associations were tested at
the .25, .5, and .75 quantiles. Significant regression paths were found for fourteen pairs
of relationships (Table 9), but only when suicidality values were restricted to the .75
quantile (i.e., highest values). These results indicate that the variables identified in Table
9 predict high values of suicidality but do not predict low or typical values (i.e., values at
the .5 quantile).
Regression surface plots. Three-dimensional plots of the regression surface were
examined for each set of relationships tested for mediation. These plots illustrate the
pattern of associations among the variables and can be used to identify potential
nonlinear relationships.
The plots of the regression surface for familism discrepancy, family functioning, and
depression provided preliminary evidence of nonlinear associations. Specifically,

58

Table 9. Confidence Intervals for the Slope of the Quantile Regression
Path for Significant Predictors of Suicidality
Predictor Suicidality Variables
Lower
CI
Upper
CI
Familism Discrepancy Past Ideation 2.423 11.512
Familism Discrepancy Future Ideation .014 2.866
Familism Discrepancy Negative Thoughts .059 .806
Marianismo Discrepancy Past Ideation .722 6.611
Marianismo Discrepancy Future Ideation .229 2.014
Conflict Past Ideation .330 1.004
Conflict Future Ideation .039 .294
Conflict Total Attempts Made .012 .274
Conflict Overall Suicidality .437 .896
Conflict Negative Thoughts .047 .092
Cohesion Overall Suicidality -1.467 -.272
Cohesion Negative Thoughts -.181 -.046
Fatalism Past Ideation .140 3.194
Fatalism Negative Thoughts .110 .306
Note. Confidence intervals that do not contain zero are significant at the .05 level.
All regression paths were restricted to the .75 quantile. CI = Confidence Interval.

depression levels appeared lower at moderate levels of familism discrepancy but higher at
the tail ends (i.e., low and high values) of familism discrepancy (Figures 18-19).
Furthermore, while depression appeared to rise with increased conflict at low levels of
family discrepancy, depression decreased somewhat with increased conflict at high levels
of discrepancy (Figure 18). Depression appeared to generally increase with decreased
cohesion and increased familism discrepancy (Figure 19).
A plot of the regression surface for the associations among familism discrepancy,
family conflict, and future suicidal ideation showed that ideation scores appeared to

59

Figure 18. Three-dimensional plot of the regression surface illustrating the pattern of
association among Familism Discrepancy, Conflict, and Depression (based on composite
of RADS and CES-D).

Figure 19. Three-dimensional plot of the regression surface illustrating the pattern of
association among Familism Discrepancy, Cohesion, and Depression (based on
composite of RADS and CES-D).

60

climb sharply at high values of familism discrepancy and conflict (Figure 20). There
appeared to be little variation in ideation scores at low to moderate values of familism
discrepancy and conflict.
Plots of the regression surface for family functioning, fatalism, and depression
showed that depression generally appeared to increase with higher conflict, higher
fatalism, and lower cohesion (Figures 21-22). Plots of the regression surface for family
functioning, fatalism, and suicidality variables showed minimal variation in suicidality at
low to moderate levels of family functioning and fatalism (Figures 23-25). However, at
high values of both fatalism and family dysfunction, future suicidal ideation and negative
thoughts about suicide appeared to increase.
Figure 20. Three-dimensional plot of the regression surface illustrating the pattern of
association among Familism Discrepancy, Conflict, and Future Ideation.

61

Figure 21. Three-dimensional plot of the regression surface illustrating the pattern of
association among Conflict, Fatalism, and Depression (based on composite of RADS and
CES-D).

Figure 22. Three-dimensional plot of the regression surface illustrating the pattern of
association among Cohesion, Fatalism, and Depression (based on composite of RADS
and CES-D).

62

Figure 23. Three-dimensional plot of the regression surface illustrating the pattern of
association among Conflict, Fatalism, and Future Suicidal Ideation.

Figure 24. Three-dimensional plot of the regression surface illustrating the pattern of
association among Conflict, Fatalism, and Negative Suicidal Thoughts.

63

Figure 25. Three-dimensional plot of the regression surface illustrating the pattern of
association among Cohesion, Fatalism, and Negative Suicidal Thoughts.

The results of these supplementary analyses suggest that caution is warranted when
drawing conclusions about the associations among familism discrepancy, family
functioning, and depression, given evidence that these variables may have a nonlinear
relationship. If this is indeed the case, findings from mediational analyses based on linear
models may be misleading. Plots of the associations among other key variables were
generally consistent with a linear model of association.
Explanatory power. Explanatory power analyses were subsequently conducted to
further explore the nature of the associations among variables in the eight sets of
relationships tested for mediation. This technique estimates the explanatory power for a
set of predictor variables without assuming linear associations among the variables. For
each outcome (i.e., depression composite, future suicidal ideation, and negative

64

thoughts), significant predictor (i.e., familism discrepancy) and mediator (i.e., conflict,
cohesion, fatalism) variables were tested.
Results showed that 1) family conflict yielded the greatest explanatory power when
predicting depression, 2) familism discrepancy and conflict together yielded the greatest
explanatory power when predicting future suicidal ideation, and 3) conflict and fatalism
together yielded the greatest explanatory power when predicting negative thoughts about
suicide (Table 10). These findings indicate that family conflict is consistently among the
set of variables yielding greatest explanatory power for depression and suicidal outcomes.

Table 10. Analyses of Explanatory Power for Outcomes in Mediation Models

Explanatory
Power
Depression composite
Familism Discrepancy .023
Conflict .326
Cohesion .110
Fatalism .105
Familism Discrepancy + Conflict .175
Familism Discrepancy + Cohesion .091
Familism Discrepancy + Fatalism .075
Conflict + Cohesion .225
Conflict + Fatalism .224
Cohesion + Fatalism .134
Familism Discrepancy + Conflict + Cohesion .212
Familism Discrepancy + Conflict + Fatalism .191
Familism Discrepancy + Cohesion + Fatalism .179
Conflict + Cohesion + Fatalism .239
Familism Discrepancy + Conflict + Cohesion + Fatalism .253

65

Table 10, Continued

Explanatory
Power
Future Suicidal Ideation
Familism Discrepancy .002
Conflict .005
Fatalism .000
Familism Discrepancy + Conflict .556
Familism Discrepancy + Fatalism .359
Conflict + Fatalism .491
Familism Discrepancy + Conflict + Fatalism .511
Negative Thoughts
Conflict .006
Cohesion .002
Fatalism .001
Conflict + Cohesion .387
Conflict + Fatalism .514
Cohesion + Fatalism .331
Conflict + Cohesion + Fatalism .491

In the case of depression, family conflict alone yielded the greatest explanatory power,
suggesting that conflict may be the most important variable in explaining depression
variance, above and beyond familism discrepancy and fatalism.

66

Chapter Five: Discussion
This study evaluated the extent to which cultural processes were associated with
depression and suicidality in adolescent Latinos from immigrant families. Etiological
perspectives on Latino youth psychopathology (Szapocznik, et al., 1989; Zayas & Dyche,
1995; Zayas, et al., 2000; Zayas, et al., 2005) were empirically tested. The proposed
cultural process model (Figure 1) hypothesized sequentially mediated effects wherein
perceived cultural discrepancy between youth and caregivers contributes to adolescent
depression/suicidality through its effects on family functioning and fatalistic beliefs. To
evaluate, the cultural process model was divided into four mediated relationships: 1)
perceived cultural discrepancy family dysfunction fatalism, 2) perceived cultural
discrepancy family dysfunction depression/suicidality, 3) perceived cultural
discrepancy fatalism depression/suicidality, and 4) family dysfunction fatalism
depression/suicidality. No support was found for the first or third sets of relationships,
as perceived cultural discrepancy variables were not significantly associated with
fatalistic beliefs. Partial support was found for the second and fourth sets of
relationships, as described below.
Consistent with hypotheses, perceived familism discrepancy was associated with
greater youth depression. This relationship was partially mediated by increases in family
conflict and decreases in family cohesion. Familism discrepancy was also associated
with increased likelihood of future suicidal ideation, with this relationship partially
mediated by increases in family conflict. Contrary to expectations, no mediated effects
were found for associations between gender role discrepancy and adolescent
depression/suicidality. Also contrary to expectations, no associations were found

67

between acculturation discrepancy and depression/suicidality. Findings did show that the
association between family conflict and youth depression/suicidality was partially
mediated by fatalism. However, contrary to expectations, fatalism did not mediate the
association between family cohesion and outcome measures.
These mediational results provided limited support for the hypothesized cultural
process model. Based on existing theoretical perspectives, the model assumes that
cultural ideologies and traditions derived from the country of origin are challenged by
acculturating youth, generating internal clashes within immigrant families (Szapocznik,
1989; Zayas & Dyche, 1995). These clashes contribute to maladaptive family
interactions (Szapocznik et al., 1989), thereby precipitating episodes of depression and
suicidality. This study sheds light on this cultural process by identifying family-based
ideologies as potential sources of tension for these families.
The model predicted that acculturation discrepancy and gender role discrepancy
would also contribute to adolescent depression and suicidality through a similar
disruption in family functioning; however no evidence for these effects were found.
While contrary to hypotheses, non-significant findings for acculturation discrepancy are
consistent with results from a prior study in which acculturation discrepancy was
unassociated with family functioning or adolescent depression (Céspedes & Huey, 2008).
These findings also converge with results from a study by Lau et al. (2005), in which
acculturation gaps in high-risk Mexican American families were not associated with
family conflict or with youth conduct problems. It may be that the acculturation
measures used in these studies may not have captured aspects of the acculturation process
that are most salient for Latino adolescents. These measures emphasized domains such

68

as language, social environment, and behavioral practices/preferences. As a result,
sources of discrepancy that were meaningfully related to adolescent psychosocial
functioning may not have been evaluated. Discrepancy in more culturally salient
domains (e.g., beliefs about expectations of family members) may prove to have a greater
impact on youth mental health, as evidenced by our findings with familism discrepancy.
Youth-caregiver disparity in such beliefs, which involve themes relevant to self-concept
and self-identity, may be experienced more negatively by acculturating youth than
differences in language or social environment, given the importance of identity formation
in adolescence (Erikson, 1968).
Non-significant findings for gender role discrepancy were unexpected, as results
from a prior study showed that composite effects of machismo and marianismo
discrepancy on depression were mediated by family dysfunction (Céspedes & Huey,
2008). It may be that gender role discrepancy effects are driven by specific parent-child
(e.g., mother-daughter, mother-son, etc.) relationships. Indeed, the gender composition
of parent-child dyads can have differential impacts on youth adjustment and psychosocial
functioning (Barber & Delfabbro, 2000; Liu, 2005; Marcus & Betzer, 1996). Disrupted
mother-daughter relationships have been argued to play an important contributing role in
Latinas’ risk for suicidal behavior (Zayas & Dyche, 1995; Zayas et al., 2000).
Furthermore, parental resistance to adolescent individuation may reduce the mother’s
ability to effectively guide her daughter in navigating the cultural clashes between
traditional and mainstream roles (Zayas et al., 2000).
Given these arguments, gender effects on the association between gender role
discrepancy and depression/suicidality may be heavily impacted by mother-daughter

69

relations, in particular. This view may help explain the significant mediated effects
found by Céspedes and Huey (2008). Seventy percent of the youth sample was female
and 61% of youth reported from a mother’s perspective. Therefore, gender effects were
likely driven by mother-daughter dyads. Because male and female caregiver scores were
collapsed into a joint measure of discrepancy in the present study, and the influence of
specific parent-child dyads (e.g., mother-daughter) was not tested, significant mediation
effects may have been missed.
An alternative explanation for the inconsistency between results in the present study
and those found by Céspedes and Huey (2008) is the technique used to calculate youth-
caregiver discrepancy. The present study followed precedence in this literature by
utilizing a generalized difference approach to calculate discrepancy (Lau, et al., 2005;
Crane, Ngai, Larson & Hafen, 2005), whereas Céspedes and Huey (2008) used a
standardized difference procedure (Reynolds, 1995). Although the generalized difference
approach may suffer from measurement error, no alternative measure of difference is able
to overcome this problem (R. Wilcox, personal communication, May 8, 2006). A
generalized difference score is therefore a reasonable approach to use in estimating
youth-caregiver discrepancy.
In addition to identifying familism discrepancy as a source of tension for immigrant
families, this study contributes to our understanding of the cultural process model by
identifying fatalism as a mediator of the association between family conflict and
adolescent depression/suicidality. However, contrary to hypotheses, fatalism did not
mediate the relationship between family cohesion and depression/suicidality. Competing
conceptualizations of fatalism may help explain these inconsistent findings. The model

70

tested in this study argues that fatalistic beliefs reduce an individual’s ability to cope
(Mirowsky & Ross, 1984; Wheaton, 1980), thereby acting as a risk factor for youth
depression and suicidality. However, fatalism has also been described as an adaptive
response to uncontrollable events (Falicov, 1998; Parker & Kleiner, 1966) that may
buffer against mental health problems (Neff & Hoppe, 1993). Indeed, fatalism has been
found to act simultaneously as a risk factor and protective factor, increasing risk for
depression but decreasing risk for anxiety following a negative event (Mirowsky & Ross,
1984).
Joiner and colleagues (2001) argue that fatalism can be conceptualized as either
pessimistic attribution or external attribution. Pessimistic attribution is characterized by
a cognitive style in which negative events are attributed to stable and global factors.
External attribution is characterized by an external locus of control. Joiner et al. (2001)
found that Latino adolescents reported greater pessimistic attribution but not greater
external attribution than adolescents of other backgrounds. Furthermore, they found that
pessimistic attribution partly accounted for the variance in depressive symptoms between
ethnic groups.
These competing conceptualizations of fatalism may help explain why fatalism
mediated the association between family conflict and depression/suicidality but not the
association between family cohesion and depression/suicidality. Items from the
Relinquished Control Scale tapped sentiments of pessimism (e.g., “When I have trouble
with something, I usually just quit trying,” “When things are bad, there is usually nothing
I can do about them,” and “When things don’t go my way, I usually just give up”), but
did not tap attributions to external factors. Pessimistic attribution may be more likely to

71

fluctuate in response to acute stressors, such as family conflict. However, external
attribution may be a more stable construct that fluctuates only in response to chronic
stressors. In other words, while family conflict may occur as a proximal response to
parent-youth discrepancy, reductions in family cohesion may occur gradually over time.
Thus the association between family cohesion and depression may be more likely
mediated by levels of external attribution, resulting in the non-significant results found.
This dual conceptualization of fatalism may also explain the lack of significant
associations between cultural discrepancy and fatalism, as cultural discrepancy may be
more highly associated with external attribution than with “pessimistic fatalism” for the
same reasons noted above.
In addition to testing mediated effects on the association between cultural
discrepancy and depression/suicidality, this study evaluated the perspective that Latinas
may have a particularly difficult time navigating the inconsistencies between parental
expectations and mainstream values (Zayas & Dyche, 1995), and may therefore be at
greater risk for negative mental health outcomes. Consistent with previous research
(CDC, 2006; Emslie et al., 1990; NCHS, 2000; Roberts & Chen, 1995; Roberts, Roberts,
& Chen, 1997; Roberts & Sobhan, 1992; Weinberg & Emslie, 1987), results showed that
females reported higher depression and suicidality compared to males. Females also
reported significantly higher machismo discrepancy and lower family cohesion than
males. Consistent with hypotheses, a discrepancy X gender interaction effect was found,
showing a positive association between machismo discrepancy and depression for
females, but a negative association for males. These results provide preliminary evidence
that gender role discrepancy effects are more robust for females than for males.

72

The pressures of reconciling parental expectations with mainstream values and the
ensuing family deterioration may be felt most acutely by adolescent Latinas for at least
two reasons. First, compared to males, women and adolescent girls are particularly
vulnerable to depression in the face of interpersonal stressors such as family conflict
(Nolen-Hoeksema & Girgus, 1994; Rudolph & Hammen, 1999). Second, given that
adolescent Latinas experience greater constraints than boys over dating, sexual
expression, and other normative adolescent activities (Hovell et al., 1994; Raffaelli &
Ontai, 2004), their efforts to individuate may violate traditional role expectations,
accelerate emotional disengagement from parents, and bring about conflict within the
family (Baptiste, 1993). Furthermore, as Latinas are often expected to care for other
members of the family (Zayas & Dyche, 1995), their push for independence may be met
with increased opposition from parents (Zayas et al., 2005). Thus, not only are
adolescent Latinas disproportionately affected by depression, but culturally-influenced
gender role strain may indirectly contribute to differential risk.
The negative association between machismo discrepancy and depression among
males is difficult to interpret, as a positive association was predicted by the cultural
process model. One speculation is that adolescent males may receive greater peer
support compared to females as they struggle against conforming with parents’ gender-
based expectations. School-related social support has been found to decrease risk for
maladjustment in Latino adolescents (e.g., Demaray & Malecki, 2002), which may
explain the differential impact of machismo discrepancy on male and female depression
levels. Empirical support for this view could not be found, however, and it remains
unclear why machismo discrepancy would be associated with lower depression for males.

73

The non-significant gender effects for the association between marianismo
discrepancy and depression were also unexpected, given previous results that composite
effects of machismo and marianismo discrepancy on depression were moderated by
gender (Céspedes & Huey, 2008). The AWS scale used to measure marianismo was
reduced by three items in order to improve reliability. It may be that the deletion of these
three items accounts for the null findings with marianismo discrepancy. Alternatively,
despite attempts to improve reliability, the AWS female caregiver version yielded an
alpha of .66. Inadequate scale reliability may also account for null findings, as alphas for
the machismo scale were considerably higher.
Implications
The findings from this study contribute to the developing literature on immigration,
acculturation, and Latino mental health. Research suggests that mental health problems,
including depression and suicidality, fall along a continuum with Mexicans living in
Mexico and recent immigrants at lower risk, and native-born Mexican-Americans and
long-term immigrants at higher risk (Reyes-Pulliza, 1990; Sorenson & Golding, 1988;
Swanson et al., 1992; Ungemack & Guarnaccia, 1998; Vega et al., 1998). Research in
this area has focused primarily on the role of acculturation in explaining these differential
prevalence rates. However, associations between acculturation status and mental health
outcomes are inconsistent, with positive, negative, and null associations found (Berry &
Kim, 1988; Rogler, Cortes, & Malgady, 1991). Research has moved towards an
emphasis on acculturative stress – the extent to which the immigration and acculturation
process is experienced as stressful – in order to elucidate the mechanisms by which the
acculturation process may impact immigrant mental health (Hovey, 2000; Hovey & King,

74

1997). A model based on acculturative stress has received empirical support in
explaining Latino adolescents’ risk for depression and suicidal behavior (Hovey, 2000;
Hovey & King, 1996), arguing that the stress of trying to reconcile conflictual values and
beliefs between traditional and mainstream cultures increases risk for youth depression
and suicidal behavior (Hovey & King, 1996).
Results from the present study contribute to our understanding of the acculturative
stress model by identifying potential sources of acculturative stress that may be
experienced by Latino youth. Youth may struggle to reconcile traditional family-based
ideologies and gender role expectations that appear incompatible with those of the
mainstream culture. As adolescents contend with family pressures to maintain traditional
values, their push for autonomy may be met with parental resistance (Zayas et al., 2000;
Zayas et al., 2005), generating maladaptive family interaction patterns (Szapocznik et al.,
1989; Zayas et al., 2000) and contributing to acculturative stress. Given more restrictive
socialization practices for females (Garcia-Preto, 1998; Hines et al., 1992), Latinas may
experience increased levels of acculturative stress, perhaps contributing to their
comparatively higher risk for depression and suicidality.
This acculturative stress model provides a framework for understanding the
continuum effect described above, where immigrants who reside for longer periods of
time in the U.S. may feel increasing stress as the values and beliefs of the mainstream
culture come into conflict with their traditional mores, thereby predisposing them to
psychopathology.

75

Limitations
Although the results of this study have implications for theory and research on the
psychosocial functioning of Latino youth, several limitations should be noted. First, the
analytic method chosen to test for mediated effects had several drawbacks. Unlike
traditional tests of mediation, the percentile bootstrap method did not permit an accurate
estimate of the extent of mediation (i.e., partial or full). This technique generated
confidence intervals for the difference between the original and adjusted regression
coefficients, permitting inferences about whether results were consistent with partial or
full mediation, but preventing confirmatory tests of this hypothesis. The percentile
bootstrap method also did not permit a simultaneous test of the overarching mediation
model, requiring a series of regression analyses to test various combinations of the
model. Given the number of regression paths tested and the associated increase in Type I
error, some significant findings may have been attributable to chance.
Despite these limitations, the percentile bootstrap mediation technique was
preferable to alternative tests of mediation because it is robust to violations of
assumptions about the distribution of the data (e.g., normality). Violations of such
assumptions can lead to misrepresentations of the data, particularly when a technique
based on least squares is utilized (Wilcox, 2005). Identification of non-normal
distributions is not straightforward, as graphical illustrations cannot always distinguish
between normal and non-normal distributions (Wilcox, 2003). Furthermore, commonly
used transformations do not necessarily result in symmetry and may not address outliers
(Wilcox, 2005). The assumption of homoscedasticity also poses problems. If
homoscedasticity is incorrectly assumed, the standard error used to calculate confidence

76

intervals is inaccurate (Wilcox, 2003), yielding misleading results. While there are
methods for testing the assumption of homoscedasticity, it is unclear whether these tests
have sufficient statistical power to detect heteroscedasticity (Wilcox, 2003). The
assumption of a linear model can be problematic in the presence of heteroscedasticity
because situations arise in which the slope of the regression line can change dramatically
depending on the range of X values utilized. Indeed, the quantile regression analyses
utilized provide evidence of this problem, as significant predictors of suicidality were
found only at high suicidality values (i.e., .75 quantile). Issues such as these are non-
trivial, but can be effectively addressed through the use of robust statistical methods such
as the percentile bootstrap method.
Alternative statistical techniques based on latent variable analysis (e.g., SEM) could
have been employed to test the hypothesized theoretical model. However, as explained
earlier, this method is susceptible to problems when assumptions of elliptical
distributions are violated because it relies on non-robust correlation matrices (Wilcox,
2003). As an alternative to SEM, a robust principal components analysis was used to
create a joint caregiver discrepancy score and a depression composite score. However,
this technique could not be utilized to collapse other conceptually similar variables (e.g.,
family conflict and family cohesion), as these variables were generally correlated to the
same degree with conceptually different variables. Three suicidality indices (i.e., past
suicidal ideation, future suicidal ideation, and future suicidal attempts) were the
exception. However, a composite suicidality score was not calculated because the SBQ
measure provided an overall suicidality score that combined items from these scales.

77

An additional limitation is that data was obtained primarily via youth self-report, and
thus we cannot be certain that caregiver values or family interaction patterns were
reported accurately. Although some parent reports were obtained (n = 70), few
significant associations were found, which may be attributable to low sample size and
insufficient power (1-β = .550, Faul & Erdfelder, 2007). While youth add an important
perspective when assessing child mental health problems, particularly in the case of
internalizing disorders (Loeber, Green, & Lahey, 1990), youth perspectives are more
likely to reflect their own values than those of their parents (Whitbeck & Gecas, 1988),
suggesting that supplementing youth reports with data from multiple informants may be
beneficial.
Identifying effective strategies for analyzing suicidality data presented an additional
challenge. The limited variance in suicidality responses prevented correlation analyses
that dealt effectively with outliers (i.e., the R program did not permit the calculation of a
correlation when outliers were removed because the interquartile range was reduced to
zero). Therefore, extreme values were not removed when testing associations between
suicidality and other key variables. While attention must be given to outlying values in
this situation, one must be careful not to rely too heavily on one or two extreme points
when determining the presence (or strength) of an association. This issue raises an
important statistical consideration regarding the extent to which the effect of outliers
should be permitted. Outliers are known to have a notable impact on the magnitude of a
correlation (Wilcox, 2003), and can both mask a significant association or create a
significant association where none exists. Making statements about the nature of an
association, or even whether a true association is present, is not straightforward,

78

particularly when dealing with data that violates conventional assumptions of normality
and homoscedasticity. For this reason, multiple methods (e.g., plots of the regression
surface, quantile regression, and explanatory power analyses) were utilized to understand
the nuanced associations between suicide and other key variables in the study.
An additional limitation of this study relates to the scales used to assess gender role
beliefs and acculturation. The gender role scale items appear to focus more on the
domineering and patriarchal aspects of machismo (e.g., “A wife should never disagree
with her husband in public” and “Wives should respect the man’s position as head of the
household”) rather than positive qualities such as bravery, strength, and loyalty.
Similarly, scale items tend to highlight the submissive and dependent aspects of
marianismo (“A woman shouldn’t be able to go to the same places or to do the same
things as men” and “On average, women are less able to make money than men”) over
positive strengths such as fostering relationships and caring for the family. The biases of
the Machismo Scale and AWS may inadvertently simplify our understanding of
machismo and marianismo, and may have failed to evaluate the more culturally-nuanced
attitudes of gender role beliefs among Latinos (Torres, Solberg, & Carlstrom, 2002).
Similarly, using only the assimilation subscale of the AHIMSA prevented evaluation
of more complex acculturation experiences. While the term assimilation is frequently
used interchangeably with the term acculturation, it captures only one aspect of the
acculturation process: the experience whereby an individual adopts all aspects of the
mainstream society while simultaneously losing all connections to the native culture
(Berry & Kim, 1988). However, acculturation is more broadly defined and captures
alternative experiences, such as those of individuals who (1) maintain ties to the native

79

culture and adopt no new values from the mainstream culture, (2) share aspects of both
the native and mainstream cultures, and (3) hold no aspects of either culture (Berry &
Kim, 1988; Kaplan & Marks, 1990).
These four modes of acculturation do not fall along the same continuum; thus the
assimilation scale alone cannot provide a complete understanding of an individual’s
acculturation experience. Indeed, individuals who are truly assimilated and those who
are bicultural may both be classified as highly assimilated, though they have undergone
qualitatively different acculturation processes. The null findings for acculturation
discrepancy may be due to an overemphasis on assimilation discrepancy. Including
discrepancy in other modes of acculturation (e.g., biculturalism) may have yielded
significant results.
Directions for Future Research
An important empirical question is whether the results found in this study are truly
specific to Latino youth, or whether they may be generalizable to youth of other
backgrounds. It is reasonable to hypothesize that the cultural discrepancy process
proposed in this study might apply to youth from other immigrant families. Indeed, a
review of studies on acculturation suggest that the immigration process may cause stress
for individuals of many backgrounds and may negatively impact both physical and
mental health (Berry, Kim, Minde, & Mok, 1987). The cultural discrepancy process
described in this study is expected to occur in families that are immigrating to a new
country and undergoing the acculturation process. However, discrepancy in the domains
of familism, machismo, and marianismo are expected to be specific to Latino populations
given arguments that these beliefs are characteristic of Latino families (Baca Zinn, 1994;

80

Comas-Díaz, 1987; Cortes, 1995; Falicov, 1998; Grebler, Moore, & Guzmán, 1970;
Marín & Marín, 1991; Rodriguez & Kosloski, 1998; Sabogal et al., 1987; Senour, 1977;
Soto, 1983; Steidel & Contreras, 2003). As the present study did not evaluate the
cultural process model with non-Latinos, future research should investigate not only
whether the broader cultural discrepancy model applies to other ethnic groups (which is
predicted), but also whether discrepancy in family- and gender-based beliefs are relevant
to these groups (which is not expected).
A related question is whether these processes might vary among Latino subgroups
(e.g., Mexican-American, Central-American, Cuban-American). Findings from this
study found marginal differences between Mexican-American and Central-American
youth in measures of acculturation discrepancy. However, these findings revealed no
consistent pattern and no conclusions could be drawn about how these groups might
differ. Research often groups Latinos together, ignoring intra-group differences in beliefs
and values (Ungemack & Guarnaccia, 1998) that may impact psychosocial functioning.
As one example, tendencies to react angrily to suicidal behavior (Reynolds, Kalish, &
Farberow, 1975) and to associate suicidal behavior with shame or disgrace (Oquendo et
al., 2004), may explain why Mexican-Americans are at comparatively lower risk for
suicidal ideation and attempts compared to those of other Latino backgrounds (Oquendo
et al., 2001; Ungemack & Guarnaccia, 1998).
If Latino subgroups differ in their beliefs about suicidality, they may also differ in
family- or gender-based beliefs, which may exert different effects on youth psychosocial
functioning. In other words, cultural discrepancy processes might unfold in different
ways depending on the Latino subgroup being studied. Continued research is needed to

81

elucidate differential causes and consequences of cultural mechanisms in Latino youth
from different backgrounds.
An additional area for future research to explore is whether cultural discrepancy in
beliefs may increase risk for psychopathology more broadly. Research shows that family
conflict is associated with externalizing behavior, aggression, and conduct problems in
youth (David, Steele, Forehand, & Armistead, 1996; Jaycox & Repetti, 1993; Weidner,
Hutt, Conner, & Mendell, 1992). To the extent that cultural discrepancy in immigrant
families contributes to conflictual interactions among family members, we may expect to
see an increased risk for maladaptive externalizing behaviors among Latino adolescents
(Szapocznik et al., 1989). This alternative view should be empirically tested in
subsequent studies to determine whether the cultural process model proposed is specific
to internalizing disorders or generalizes to other indices of psychosocial functioning.
Alternative perspectives on the direction of effect between cultural discrepancy and
depression/suicidality should also be considered. The theoretical model proposed in this
study presumes linear paths of influence from cultural discrepancy to family dysfunction
to fatalism to psychopathology. It is also possible that cultural discrepancy, family
dysfunction, and fatalism are consequences of youth psychopathology. Longitudinal
research suggests that individuals with unipolar depression often generate stressful life
experiences, which frequently take the form of interpersonal problems (Hammen, 2005).
Accordingly, for depressed adolescent Latinos, the pessimism, irritability, and anhedonia
characteristic of clinical depression may contribute to interpersonal problems and
conflictual interactions with parents (Daley et al., 1997; Hammen, 2005). As a result,
depressed adolescents may begin to devalue the views of their caregiver, thereby

82

intensifying the existing values discrepancy. The cross-sectional nature of the present
study prevented a test of this alternative hypothesis. Hence, future studies should
consider longitudinal methods to discern the sequencing of depression and suicidality in
relation to hypothesized causal mechanisms.
Treatment Implications
This line of research carries treatment implications for emotionally distressed,
adolescent Latinos. Szapocznik and colleagues have developed family-based treatments
that target maladaptive family interactions resulting from intercultural conflict, which
have shown some success at reducing externalizing and drug-related problems for
adolescent Latinos (Szapocznik et al., 1986; Szapocznik et al., 1989). The findings from
the present study suggest that appropriate targets for intervention may be the discrepancy
in family- and gender-based expectations as well as resulting pessimistic cognitions,
which may both increase risk for depression/suicidality.
Whether interventions targeting family dysfunction should focus on reducing family
conflict or increasing family cohesion is a nontrivial issue. In a study by Herman,
Ostrander, and Tucker (2007), low family cohesion was uniquely associated with
depression among African-American adolescents, whereas high family conflict was
uniquely associated with depression among European-American adolescents. Herman et
al.’s findings suggest that the nature of the family dysfunction is important in predicting
adolescent psychopathology and may differ by culture or ethnicity. While these issues
have not been studied in Latino populations, they raise an important consideration with
regard to appropriate clinical intervention. It may be that reducing family conflict and

83

promoting family cohesiveness have differential impacts on diminishing risk for
depression and suicidal behavior in Latinos.

84

References

Afifi, A.A. & Elashoff, R.M. (1966). Missing observations in multivariate statistics.
Part 1. Revies of the literature. Journal of the American Statistical Association, 61,
595-604.

Aisenberg, E., Trickett, P.K., Mennen, F.E., Saltzman, W., & Zayas, L.H. (2007).
Maternal depression and adolescent behavior problems: An examination of
mediation among immigrant Latino mothers and their adolescent children
exposed to community violence. Journal of Interpersonal Violence, 22,
1227-1249.

Asarnow, J. R. & Carlson, G. (1988). Suicide attempts in preadolescent child
psychiatry inpatients. Suicide and Life-Threatening Behavior, 18, 129-136.

Asarnow, J. R., Carlson, G. A. & Guthrie, D. (1987). Coping strategies, self-
perceptions, hopelessness, and perceived family environments in depressed and
suicidal children. Journal of Consulting and Clinical Psychology, 55, 361-366.

Baca Zinn, M. (1994). Adaptation and continuity in Mexican-origin families. In R.L.
Taylor (Ed.), Minority families in the United States: A multicultural perspective (pp.
64-94). Englewood Cliffs, NJ: Prentice Hall.

Baptiste, D. A., Jr. (1993). Immigrant families, adolescents and acculturation:
Insights for therapists. Marriage and Family Review, 19, 341-363.

Barber, J.G. & Delfalbro, P. (2000). Predictors of adolescent adjustment: Parent-peer
relationships and parent-child conflict. Child and Adolescent Social Work Journal,
17, 275-288.

Baron, R.M. & Kenny, D.A. (1986). The moderator-mediator variable distinction in
social psychological research: Conceptual, strategic, and statistical considerations.
Journal of Personality and Social Psychology, 51, 1173-1182.

Berry, J. W. & Kim, U. (1988). Acculturation and mental health. In P. R. Dasen, J.
W. Berry, & N. Sartorius (Eds.), Health and cross-cultural psychology (pp. 207-
236).Thousand Oaks: Sage Publications.

Berry, J.W., Kim, U., Minde, T., & Mok, D. (1987). Comparative studies of
acculturative stress. International Migration, 21, 491-511.

Birmaher, B., Ryan, N.D., Williamson, D.E., Brent, D.A., Kaufman, R.E., Dahl,
R.E., et al., (1996). Childhood and adolescent depression: A review of the past 10
years (part 1). Journal of the American Academy of Child and Adolescent Psychiatry,
35, 1427-1440.

85

Blazer, D.G., Kessler, R.C., McGonagle, K.A., & Swartz, M.S. (1994). The
prevalence and distribution of major depression in a national community sample:
The national comorbidity survey. American Journal of Psychiatry, 151, 979-986.

Bouma, E.M., Ormel, J., Verhulst, F.C., & Oldehinkel, A.J. (2008). Stressful life
events and depressive problems in early adolescent boys and girls: The influence of
parental depression, temperament, and family environment. Journal of Affective
Disorders, 105, 185-193.

Burgess, E.W., Locke, H.J., & Thomes, M.M. (1963). The family: From institution
to companionship (3
rd
edition). New York: American Book Company.

Burton, E., Stice, E., & Seeley, J.R. (2004). A prospective test of the stress-buffering
model of depression in adolescent girls: No support once again. Journal of
Counseling and Clinical Psychology, 72, 689-697.

Canino, G. (1982). Transactional family patterns: A preliminary exploration of
Puerto Rican female adolescents. In R.E. Zambrana (Ed.), Work, family, and health:
Latina women in transition (pp. 27-36). New York: Hispanic Research Center,
Fordham University.

Carlton-Ford, S., Paikoff, R. L., & Brooks-Gunn, J. (1991). Methodological issues in
the study of divergent views of the family. In R. L. Paikoff (Ed.), New directions for
child development: Shared views in the family during adolescence (pp. 87-102). San
Francisco: Jossey-Bass.

Centers for Disease Control and Prevention (2006). CDC Surveillance Summaries,
June 9, 2006. MMWR;55 (No. SS-5).

Céspedes, Y.M. & Huey, S.J. (2008). Depression in Latino adolescents: A cultural
discrepancy perspective. Cultural Diversity and Ethnic Minority Psychology, 14,
168-172.

Chabrol, H. & Moron, P. (1988). Depressive disorders in 100 adolescents who
attempted suicide. American Journal of Psychiatry, 145, 379.

Chioqueta A.P. & Stiles, T.C. (2007). The relationship between psychological
buffers, hopelessness, and suicidal ideation: Identification of protective factors.
Crisis, 28, 67-73.

Christiansen, T. F. (1997). The acculturation gap: A study of relations among
acculturation, acculturative stress, and coping responses for Latino adolescents and
parents. Unpublished doctoral dissertation, University of Iowa.

Cole, D. (1989). Psychopatholgy of adolescent suicide: Hopelessness, coping
beliefs, and depression. Journal of Abnormal Psychology, 9, 248-255.

86

Cole, D.A. & McPherson, A.E. (1993). Relation of family subsystems to adolescent
depression: Implementing a new family assessment strategy. Journal of Family
Psychology, 7, 119-133.

Comas-Díaz, L. (1987). Cross-cultural mental health. In L. Comas-Díaz & E.
Griffith (Eds.), Clinical guidelines in cross-cultural mental health (pp. 337-361).
New York: Wiley.

Compas, B.E. (1997). Depression in children and adolescents. In E.J. Mash & L.G.
Terdal (Eds.), Assessment of childhood disorders (3
rd
edition) (pp 197-229). New
York: Guilford Press.

Constantine, M.G. (2006). Perceived family conflict, parental attachment, and
depression in African-American female adolescents. Cultural Diversity and Ethnic
Minority Psychology, 12, 697-709.

Corder, B. F., Shorr, W., & Corder, R. F. (1974). A study of social and psychological
of adolescent suicide attempters in an urban, disadvantaged area. Adolescence, 9, 1-
5.

Cortes, D.E. (1995). Variations in familism in two generations of Puerto Ricans.
Hispanic Journal of Behavioral Sciences, 17, 249-255.

Crane, D.R., Ngai, S.W., Larson, J.H., & Hafen, M. (2005). The influence of family
functioning and parent-adolescent acculturation on North American Chinese
adolescent outcomes. Family Relations, 54, 400-410.

Crockett, L.J., Randall, B.A., Shen, Y., Russell, S.T., &Driscoll, A.K. (2005).
Measurement equivalence of the Center for Epidemiological Studies Depresión Scale
for Latino and Anglo adolescents: A nacional study. Journal of Consulting and
Clinical Psychology, 73, 47-58.

Cuéllar, I., Arnold, B., & González, G. (1995). Cognitive referents of acculturation:
Assessment of cultural constructs in Mexican-Americans. Journal of Community
Psychology, 23, 339-356.

Cumsille, P.E. & Epstein, N. (1994). Family cohesion, family adaptability, social
support, and adolescent depressive symptoms in outpatient clinic families. Journal of
Family Psychology, 8, 202-214.

Daley, S. E., Hammen, C., Burge, D., Davila, J., Paley, B., Lindberg, N., &
Herzberg, D. (1997). Predictors of the generation of episodic stress: A
longitudinal study of late adolescent women. Journal of Abnormal Psychology,
106, 251-259.

87

David, C., Steele, R., Forehand, R., & Armistead, L. (1996). The role of family
conflict and marital conflict in adolescent functioning. Journal of Family Violence,
11, 81-91.

Demaray, M. K. & Malecki, C. K. (2002). The relationship between perceived social
support and maladjustment for students at risk. Psychology in the Schools, 39, 305-
316.

De Wilde, E. J., Kienhorst, I. C., Diekstra, R. F., & Wolters, W. H., (1993). The
specificity of psychological characteristics of adolescent suicide attempters. Journal
of the American Academy of Child and Adolescent Psychiatry, 32, 51-59.

Domino, G., Fragoso, A., & Moreno, H. (1991). Cross-cultural investigations of the
imagery of cancer in Mexican nationals. Hispanic Journal of Behavioral Sciences,
13, 422-435.

DuBois, D. L., Felner, R. D., Brand, S., Adan, A. M., & Evans, E. G. (1992). A
prospective study of life stress, social support, and adaptation in early adolescence.
Child Development, 63, 542-557.

Elliott, K. A. G. (2001). The relationship between acculturation, family functioning,
and school performance of Mexican American adolescents. Unpublished doctoral
dissertation, University of California, Santa Barbara.

Emslie, G. J., Weinberg, W. A., Rush, A. J., Adams, R. M., & Rintelmann, J. W.
(1990). Depressive symptoms by self-report in adolescence: Phase 1 of the
development of a questionnaire by self-report. Journal of Child Neurology, 5, 114-
121.

Erison, E. (1968). Identity: Youth and Crisis. New York: W.W. Norton.

Falicov, C.J. (1998). Latino families in therapy. New York: Guildford Press.

Farris, B.E. & Glenn, N.D. (1976). Fatalism and familism among Anglos and
Mexican Americans in San Antonio. Sociology & Social Research, 60, 393-402.

Faul, F., Erdfelder, E., Lang, A., & Buchner, A. (2007). G*Power 3: A flexible
statistical power analysis program for the social, behavioral, and biomedical
sciences. Behavior Research Methods, 39, 175-191.

Fendrich, M., Warner, V., & Weissman, M.M. (1990). Family risk factors, parental
depression, and psychopathology in offspring. Developmental Psychology, 26, 40-
50.

Flores, G. (2000). Culture and the patient-physician relationship: Achieving cultural
competency in health care. Journal of Pediatrics, 136, 14-23.

88

Fortuna, L.R., Perez, D.J., Canino, G., Sribney, W., & Alegria, M. (2007).
Prevalence and correlates of lifetime suicidal ideation and suicide attempts among
Latino subgroups in the United States. Journal of Clinical Psychiatry, 68, 572-581.

Garcia-Preto, N. (1998). Latinas in the United States: Bridging two worlds. In M.
McGoldrick (Ed.), Re-visioning family therapy: Race, culture, and gender in clinical
practice (pp 330-344). New York: Guilford Press.

George, L.K. (1986). Caregiver burden: Conflict between norms of reciprocity and
solidarity. In K.A. Pillemer & R.S. Wolf (Eds.), Elder abuse: Conflict in the family
(pp 67-92). Greenwood, CT: Auburn House.

Gil, A.G., Wagner, E.F., & Vega, W.A. (2000). Acculturation, familism, and alcohol
use among Latino adolescent males: Longitudinal relations. Journal of Community
Psychology, 28, 443-458.

Grebler, L., Moore, J. W., & Guzmán, R.C. (1970). The Mexican American people.
New York: The Free Press.

Hammen, C. (2005). Stress and depression. Annual Review of Clinical Psychology,
1, 293-319

Hammen, C. & Rudolph, K.D. (1996). Childhood depression. In E.J. Mash & R.A.
Barkley (Eds.), Child psychopathology (pp 153-195). New York: Guildford Press.

Hampel, F.R., Ronchetti, E.M., Rousseeuw, P.J., & Stahel, W.A. (1986). Robust
Statistics: The Approach Based on Influence Functions. New York: Wiley.

Hankin, B.L., Mermelstein, R., & Roesch, L. (2007). Sex differences in adolescent
depression: Stress exposure and reactivity models. Child Development, 78, 279-295.

Harwood, R.L., Schoelmerich, A., Venturea-Cook, E., & Schulze, P.A. (1996).
Culture and class influences on Anglo and Puerto Rican mothers' beliefs regarding
long-term socialization goals and child behavior. Child Development, 67, 2446-
2461.

Herman, K.C., Ostrander, R., & Tucker, C.M. (2007). Do family environments and
negative cognitions of adolescents with depressive symptoms vary by ethnic group?
Journal of Family Psychology 21, 325-330.

Hines, P.M., Garcia-Preto, N., McGoldrick, M., & Rhea, A. (1992). Intergenerational
relationships across cultures. Families in Society. Special Issue: Multicultural
practice, 73, 323-338.

89

Hollis, C.P. (1996). Depression, family environment, and adolescent suicidal
behavior. Journal of the American Academy of Child and Adolescent Psychiatry, 35,
622-630.

Holmbeck, H. N. (1997). Toward terminological, conceptual, and statistical clarity in
the study of mediators and moderators: Examples from the child-clinical and
pediatric psychology literatures. Journal of Consulting and Clinical Psychology, 65,
599-610.

Hops, H., Lewinsohn, P.M., Andrews, J.A., & Roberts, R. (1990). Psychosocial
correlates of depressive symptomatology among high school students. Journal of
Clinical Child Psychology, 19, 211-220.

Hovell, M., Sipan, C., Blumberg, E., Atkins, C., Hofstetter, C. R., & Kreitner, S.
(1994). Family influences on Latino and Anglo adolescents' sexual behavior.
Journal of Marriage and the Family, 56, 973-986.

Hovey, J.D. (2000). Acculturative stress, depression, and suicidal ideation in
Mexican immigrants. Cultural Diversity and Ethnic Minority Psychology, 6, 134-
151.

Hovey, J. D. & King, C. A. (1996). Acculturative stress, depression, and suicidal
ideation among immigrant and second-generation Latino adolescents. Journal of the
American Academy of Child and Adolescent Psychiatry, 35, 1183-1192.

Hovey, J.D. & King, C.A. (1997). Suicidality among acculturating Mexican-
Americans: Current knowledge and directions for research. Suicide and Life-
Threatening Behavior. Special Issue: Suicide: Individual, Cultural, International
Perspectives, 27, 92-103.

Huber, P.J. (1981). Robust Statistics. New York: Wiley.

Jamieson, P.E., & Romer, D. (2008). Unrealistic fatalism in U.S. youth ages 14 to
22: Prevalence and characteristics. Journal of Adolescent Health, 42, 154-160.

Jaycox, L.H. & Repetti, R.L. (1993). Conflict in families and the psychological
adjustment of preadolescent children. Journal of Family Psychology, 7, 344-355.

Joiner, T.E., Perez, M., Wagner, K.D., Berenson, A., & Marquina, G.S. (2001). On
fatalism, pessimism, and depressive symptoms among Mexican-American and other
adolescents attending an obstetrics-gynecology clinic. Behaviour Research and
Therapy, 39, 887-896.

Kaplan, M. S. & Marks, G. (1990). Adverse effects of acculturation: psychological
distress among Mexican-American young adults. Social Science & Medicine, 31,
1313-1319.

90

Kaslow, N,J., Rehm, L.P., & Siegel, A.W. (1984). Social-cognitive and cognitive
correlates of depression in children. Journal of Abnormal Child Psychology, 12, 605-
620.

Kranau, E. J., Green, V., & Valencia-Weber, G. (1982). Acculturation and the
Hispanic woman: attitudes toward women, sex-role attribution, sex-role behavior,
and demographics. Hispanic Journal of Behavioral Sciences, 4, 21-40.

Lau, A.S., Yeh, M., Wood, P.A., McCabe, K.M., Garland, A.F., & Hough, R.L.
(2005). The acculturation gap-distress hypothesis among high-risk Mexican-
American families. Journal of Family Psychology, 19, 367-375.

Leaper, C. & Valin, D. (1996). Predictors of Mexican-American mothers’ and
fathers’ attitudes toward gender equality. Hispanic Journal of Behavioral Sciences,
18, 343-355.

Lester, D. & Anderson, D. (1992). Depression and suicidal ideation in African-
American and Hispanic-American high school students. Psychological Reports, 71,
618.

Lewinsohn, P. M., Rohde, P., & Seeley, J. R. (1993). Psychosocial characteristics
of adolescents with a history of suicide attempt. Journal of the American Academy of
Child and Adolescent Psychiatry, 32, 60-68.

Lewinsohn, P.W., Rohde, P., Seeley, J.R. (1994). Psychosocial risk factors for future
adolescent suicide attempts. Journal of Consulting and Clinical Psychology, 62, 297-
305.

Lewinsohn, P. M., Rohde, P. & Seeley, J. R. (1996). Adolescent suicidal ideation
and attempts: prevalence, risk factors, an clinical implications. Clinical Psychology:
Science & Practice, 3, 25-46.

Linehan, M. (1996). Suicidal Behaviors Questionnaire (SBQ). Unpublished
instrument, University of Washington, Seattle, WA.

Liu, R.X. (2005). Parent-youth closeness and youth’s suicidal ideation: The
moderating effects of gender, stages of adolescence, and race or ethnicity. Youth &
Society, 37, 145-175.

Locke, T.F. & Newcomb, M.D. (2005). Psychosocial predictors and correlates of
suicidality in teenage Latino males. Hispanic Journal of Behavioral Sciences, 27,
319-336.

Loeber, R., Green, S. M., & Lahey, B. B. (1990). Mental health professionals'
perceptions of the utility of children, mothers, and teachers as informants on
childhood psychopathology. Journal of Clinical Psychology, 19, 136-143.

91

MacKinnon, D.P., Lockwood, C.M., Hoffman, J.M., West, S.G., & Sheets, V.
(2002). A comparison of methods to test mediation and other intervening variable
effects. Psychological Methods, 7, 83-104.

Madsen, W. (1973). Mexican Americans of South Texas (2
nd
Ed.). New York: Holt,
Rinehart, & Winston.

Maker, A.H., Shah, P.V., & Agha, Z. (2005). Child physical abuse: Prevalence,
characteristics, predictors, and beliefs about parent-child violence in South Asian,
Middle Eastern, East Asian, and Latina women in the United States. Journal of
Interpersonal Violence, 20, 1406-1428.

Mann, J.J., Waternaux, C., Haas, G.L., & Malone, K.M. (1999). Toward a clinical
model of suicidal behavior in psychiatric patients. American Journal of Psychiatry,
156, 181-189.

Marcus, R.F. & Betzer, P.D. (1996). Attachment and antisocial behavior in early
adolescence. Journal of Early Adolescence, 16, 229-248.

Marin, G., & Marin, B. (1991). Research with Hispanic populations. Newbury Park,
CA: Sage.

Maronna, R. (2005). Principal components and orthogonal regression based on
robust scales. Technometrics, 47, 264-273.

McLaughlin, K.A., Hilt, L.M. & Nolen-Hoeksema, S. (2007). Racial/ethnic
differences in internalizing and externalizing symptoms in adolescents. Journal of
Abnormal Child Psychology, 35, 801-816.

Miller, M. L., Chiles, J. A., & Barnes, V. E. (1982). Suicide attempters within a
delinquent population. Journal of Consulting and Clinical Psychology, 50, 491-498.

Mirowsky, J. & Ross, C.E. (1984). Mexican culture and its emotional contradictions.
Journal of Health and Social Behavior, 25, 2-13.

Moos, R. & Moos, B. (1981). Manual for the Family Environment Scale. Palo Alto,
CA: Consulting Psychologists Press.

National Center for Health Statistics (2000). Health, United States, 2000 with
Adolescent Health Chartbook. Hyattsville, MD.

Neff, J.A. & Hoppe, S.K. (1993). Race/ethnicity, acculturation, and psychological
distress: Fatalism and religiosity as cultural resources. Journal of Community
Psychology, 3-20.

92

Negron, R., Piacentini, J., Graae, F., Davies, M., & Shaffer, D. (1997). Microanalysis
of adolescent suicide attempters and ideators during the acute suicidal episode.
Journal of the American Academy of Child and Adolescent Psychiatry, 36, 1512-
1519.

Nieto, C. (1997). The Chicana and the women’s rights movement. In A.M. Garcia
(Ed.), Chicana feminist thought: The basic historical writings (pp. 206-211). New
York: Routledge.

NietoGomez, A. (1997). La Chicana – Legacy of suffering and self-denial. In A.M.
Garcia (Ed.), Chicana feminist thought: The basic historical writings (pp. 48-50).
New York: Routledge.

Nolen-Hoeksema, S., & Girgus, J. S. (1994). The emergence of gender differences in
depression during adolescence. Psychological Bulletin, 115, 424-443.

Norris, A.E., Ford, K., & Bova, C.A. (1996). Psychometrics of a brief acculturation
scale for Hispanics in a probability sample of urban Hispanic adolescents and young
adults. Hispanic Journal of Behavioral Sciences, 18, 29-38.

Okagaki, L., & Sternberg, R. J. (1993). Parent beliefs and children's school
performance. Child Development, 64, 36-56.

Olson, D. H., Portner, J., & Lavee, Y. (1985). FACES-III. St. Paul, MN: Family
Social Science, University of Minnesota; 1985.

Oquendo, M.A., Ellis, S.P., Greenwald, S., Malone, K.M., Weissman, M.M., &
Mann, J.J. (2001). Ethnic and sex differences in rates of suicide completion relative
to major depression in the United States. American Journal of Psychiatry, 158, 1652-
1658.

Osman, A., Barrios, F.X., Gutierrez, P.M., Wrangham, J.J., Kopper, B.A., Truelove,
R.S., & Linden, S. (2002). The positive and negative suicide ideation (PANSI)
inventory: Psychometric evaluation with adolescent psychiatric inpatient samples.
Journal of Personality Assessment, 79, 512-530.

Osman, A., Gutierrez, P.M., Kopper, B.A., Barrios, F.X., & Chiros, C.E. (1998). The
positive and negative suicide ideation inventory: Development and validation.
Psychological Reports, 82, 783-793.

Osman, A., Kopper, B.A., Barrios, F.X., Osman, J., Besett, T., & Linehan, M.
(1996). The brief reasons for living inventory for adolescents (BRFL-A). Journal of
Abnormal Child Psychology, 24, 433-443.

Parker, S. & Kleiner, R. (1966). Mental illness in the urban Negro community. New
York: Free Press.

93

Pearlin, L.I., Menaghan, E.G., Lieberman, M.A., & Mullan, J.T. (1981). The stress
process. Journal of Health and Social Behavior, 22, 337-356.

Pfeffer, C.R. (1997). Childhood suicidal behavior. Suicide, 20, 551-562.

Radloff, L.S. (1977). The CES-D scale: A self-report depression scale for research in
the general population. Applied Psychological Measurement, 1, 385-401.

Radloff, L.S. (1991). The use of the Center for Epidemiologic Studies Depression
scale in adolescents and young adults. Journal of youth and adolescence. Special
Issue: The emergence of depressive symptoms during adolescence, 20, 149-166.

Raffaelli, M. & Ontai, L.L. (2004). Gender socialization in Latino/a families: Results
from two retrospective studies. Sex Roles, 50, 287-299.

Reyes-Pulliza, J.C. (1990). Prevalence of suicide ideation and attempts in a sample
of adults 17-64 years old in Puerto Rico in 1984. Master’s Thesis, University of
Puerto Rico, San Juan, Puerto Rico.

Reynolds, C.R. (1995). Critical measurement issues in learning disabilities. Journal
of Special Education, 18, 451-476.

Reynolds, D.K., Kalish, R.A., & Farberow, N.L. (1975). A cross-ethnic study of
suicide attitude and expectation in the United States. In N.LO. Farberow (Ed.)
Suicide in different cultures. Baltimore: University Park Press.

Reynolds, W. M. (2002). Reynolds Adolescent Depression Scale-2: Professional
Manual. Odessa, Fl: Psychological Assessment Resources.

Roberts, R. E. & Chen, Y. (1995). Depressive symptoms and suicidal ideation
among Mexican-origin and Anglo adolescents. Journal of the American Academy of
Child and Adolescent Psychiatry, 34, 81-90.

Roberts, R. E., Roberts, C. R., & Chen, Y. R. (1997). Ethnocultural differences in
prevalence of adolescent depression. American Journal of Community Psychology,
25, 95-110.

Roberts, R. E., Roberts, C. R., & Chen, I.G. (2000). Fatalism and risk of adolescent
depression. Psychiatry, 63, 239-252.

Roberts, R. E. & Sobhan, M. (1992). Symptoms of depression in adolescence: a
comparison of Anglo, African, and Hispanic Americans. Journal of Youth and
Adolescence, 21, 639-651.

94

Rodriguez, J.M. & Kosloski, K. (1998). The impact of acculturation on attitudinal
familism in a community of Puerto Rican Americans. Hispanic Journal of
Behavioral Sciences, 20, 375-390.

Rodriguez, N., Mira, C. B., Paez, N. D., & Myers, H. F. (2007). Exploring the
complexities of familism and acculturation: Central constructs for people of Mexican
origin. American Journal of Community Psychology, 39, 61-77.

Rogler, L., Cortes, D., & Malgady, R. (1991). Acculturation and mental health status
among Hispanics: Convergence and new directions for research. American
Psychologist, 46, 585-597.

Ross, C.E., Mirowsky, J., & Cockerham, W.C. (1983). Social class, Mexican culture,
and fatalism: Their effects on psychological distress. American Journal of
Community Psychology, 11, 383-399.

Rotter, J.B. (1966). Generalized expectancies for internal vs. external control of
reinforcement. Psychological Monographs, 80, 1-28.

Rubenstein, J. L., Heeren, T., Housman, D., Rubin, C., & Stechler, G. (1989).
Suicidal behavior in “normal” adolescents: Risk and protective factors. American
Journal of Orthopsychiatry, 59, 59-71.

Rudolph, K. D., & Hammen, C. (1999). Age and gender as determinants of stress
exposure, generation, and reactions in youngsters: A transactional perspective.
Child Development, 70, 660-677.

Rudolph, K. D., Hammen, C., & Burge, D. (1994). Interpersonal functioning and
depressive symptoms in childhood: Addressing the issues of specificity and
comorbidity. Journal of Abnormal Child Psychology, 22, 335-371.

Ruiz, R. A. (1977). The delivery of mental health and social change services for
Chicanos: Analysis and recommendations. In J. L. Martinez (Ed.), Chicano
psychology (pp. 233-248). New York: Academic Press.

Rutter, P.A. & Behrendt, A. E. (2004). Adolescent suicide risk: Four psychosocial
factors. Adolescence, 39, 295-302.

Sabogal, F., Marín, G., Otero-Sabogal, R., Marín, B., & Perez-Stable, E.J. (1987).
Hispanic familism and acculturation: What changes and what doesn’t? Hispanic
Journal of Behavioral Sciences, 9, 397-412.

Senour, M. (1977). Psychology of the Chicana. In J. L. Martinez (Ed.), Chicano
psychology (pp. 329-340). New York: Academic Press.

95

Shaffer, D., Gould, M.S., Fisher, P., Trautman, P., Moreau, D., Kleinman, M., &
Flory, M. (1996). Psychiatric diagnosis in child and adolescent suicide. Archives of
General Psychiatry, 53, 339-348.

Sheeber, L., Hops, H., Alpert, A., Davis, B., & Andrews, J. (1997). Family support
and conflict: Prospective relations to adolescent depression. Journal of Abnormal
Child Psychology, 25, 333-344.

Simmons, J., Cooper, M.J., Drinkwater, J., & Stewart, A. (2006). Cognitive
schemata in depressed adolescent girls and their mothers. Behavioral and Cognitive
Psychotherapy, 34, 219-232.

Simon, T.R., Swann, A.C., Powell, K.E., Potter, L.B., Kresnow, M., & O’Carroll,
P.W. (2001). Characteristics of impulsive suicide attempts and attempters. Suicide
and Life Threatening Behavior, 32, 49-59.

Sorenson, S.B. & Golding, J.M. (1988). Prevalence of suicide attempts in a Mexican-
American population: Prevention implications of immigration and cultural issues.
Suicide and Life-Threatening Behaviors, 18, 322-333.

Soto, E. (1983). Sex-role traditionalism and assertiveness in Puerto Rican women
living in the United States. Journal of Community Psychology, 11, 346-354.

Southall, D. & Roberts, J.E. (2002). Attributional style and self-esteem in
vulnerability to adolescent depressive symptoms following life stress: A 14-week
prospective study. Cognitive Therapy and Research 26, 563-579.

Spence, J. T., Helmreich, R., & Stapp, J. (1973). A short version of the Attitudes
Toward Women scale (AWS). Bulletin of the Psychonomic Society, 2, 219-220.

Spirito, A., Brown, L., Overholser, J., & Fritz, G. (1989). Attempted suicide in
adolescence: A review and critique of the literature. Clinical Psychology Review, 9,
335-363.

Staudte, R.G. & Sheather, S.J. (1990). Robust Estimation and Testing. New York:
Wiley.

Steidel, A.G.L., & Contreras, J.M. (2003). A new familism scale for use with Latino
populations. Hispanic Journal of Behavioral Sciences, 25, 312-330.

Sue, D. W. & Sue, D. (1990). Counseling the culturally different: Theory and
practice (2
nd
ed.). Oxford: John Wiley & Sons.

96

Swanson, J.W., Linskey, A.O., Quintero-Salinas, R., Pumariega, A.J., & Holzer,
C.E. (1992). A binational school survey of depressive symptoms, drug use, and
suicidal ideation. Journal of American Academy of Child Adolescent Psychiatry, 31,
669-678.

Szapocznik, J., Rio, A., Perez-Vidal, A., Kurtines, W., Hervis, O., & Santisteban, D.
(1986). Bicultural effectiveness training (BET): An experimental test of an
intervention modality for families experiencing intergenerational/intercultural
conflict. Hispanic Journal of Behavioral Sciences, 8, 303-330.

Szapocznik, J., Santisteban, D., Rio, A., Perez-Vidal, A., Santisteban, D., &
Kurtines, W. M. (1989). Family effectiveness training: an intervention to prevent
drug abuse and problem behaviors in Hispanic adolescents. Hispanic Journal of
Behavioral Sciences, 11, 4-27.

Tobin, J.J. (2000). Adolescent suicide attempters in the year following
hospitalization: Does the parent-child relationship predict symptom relapse and
recurrent suicidality? Dissertation Abstracts International: Section B: Sciences &
Engineering, 61, 2225.

Torres, J. B., Solberg, S. H., & Carlstrom, A. H. (2002). The myth of sameness
among Latino men and their machismo. American Journal of Orthopsychiatry,
72, 163-181.

Turner, J. E. & Cole, D. A. (1994). Development differences in cognitive diatheses
for child depression. Journal of Abnormal Child Psychology, 22, 15-32.

Twenge, J.M. & Nolen-Hoeksema, S. (2002). Age, gender, race, socioeconomic
status, and birth cohort differences on the children’s depression inventory: A meta-
analysis. Journal of Abnormal Psychology, 111, 578-588.

Ungemack, J.A., Guarnaccia, P.J. (1998). Suicidal ideation and suicide attempts
among Mexican-Americans, Puerto Ricans, and Cuban-Americans. Transcultural
Psychiatry, 35, 307-327.

Unger, J.B., Gallaher, P., Shakib, S., Ritt-Olson, A., Palmer, P.H., & Johnson, C.A.
(2002). The AHIMSA acculturation scale: A new measure of acculturation for
adolescents in a multicultural society. Journal of Early Adolescence, 22, 225-251.

Varela, R. E., Vernberg, E. M., Sanchez-Sosa, J. J., Riveros, A., Mitchell, M., &
Mashunkashey, J. (2004). Parenting style of Mexican, Mexican American, and
Caucasian-Non-Hispanic families: Social context and cultural influences. Journal of
Family Psychology, 18, 651-657.

97

Vega, W.A., Kolody, B., Aguilar-Gaxiola, S., Alderete, E., Catalano, R., & Caraveo-
Anduaga, J. (1998). Lifetime prevalence of DSM-III-R psychiatric disorders among
urban and rural Mexican-Americans in California. Archives of General Psychiatry,
55, 771-778.

Wagner, B.M. (1997). Family risk factors for child and adolescent suicidal behavior.
Psychological Bulletin, 121, 246-298.

Wagner, B.M., Silverman, M.A. & Martin, C.E. (2003). Family factors in youth
suicidal behaviors. American Behavioral Scientist, 46, 1171-1191.

Weidner, G., Hutt, J., Conner, S.L., & Mendell, N.R. (1992). Family stress and
coronary risk in children. Psychosomatic Medicine, 54, 471-479.

Weinberg, W. A. & Emslie, G. J. (1987). Depression and suicide in adolescents.
International Pediatrics, 2, 154-159.

Weisz, J.R. (1992). Primary-Secondary Control Scale. Unpublished measure,
University of California, Los Angeles.

Weisz, J.R., McCabe, M.A., & Dennig, M.D. (1994). Primary and secondary control
among children undergoing medical procedures: Adjustment as a function of coping
style. Journal of Consulting and Clinical Psychology, 62, 324-332.

Wheaton, B. (1980). The sociogenesis of psychological disorder: An attributional
theory. Journal of Health and Social Behavior, 21, 100-124.

Whitbeck, L.B. & Gecas, V. (1988). Value attributions and value transmission
between parents and children. Journal of Marriage and the Family, 50, 829-840.

Wilcox, R.R. (1996). Statistics for the Social Sciences. San Diego: Academic Press.

Wilcox, R.R. (2003). Applying contemporary statistical techniques (1
st
edition). San
Diego: Academic Press.

Wilcox, R.R. (2005). Introduction to robust estimation and hypothesis testing (2
nd

Edition). San Diego: Academic Press.

Yuan, K. & Bentler, P.M. (1998). Robust mean and covariance structure analysis.
British Journal of Mathematical and Statistical Psychology, 51, 63-88.

Zayas, L.H. & Dyche, L.A. (1995). Suicide attempts in Puerto Rican adolescent
females: A sociocultural perspective and family treatment approach. In J.K.
Zimmerman & G.M. Asnis (Eds.), Treatment approaches with suicidal adolescents
(pp. 203-218). New York: John Wiley & Sons, Inc.

98

Zayas, L.H., Fortuna, L.R., Lester, R.J., & Cabassa, L.J. (2005). Why do so many
Latina teens attempt suicide? A conceptual model for research. American Journal of
Orthopsychiatry, 75, 275-287.

Zayas, L.H., Kaplan, C., Turner, S., Romano, K., & Gonzalez-Ramos, G. (2000).
Understanding suicide attempts by adolescent Hispanic females. Social Work, 45,
53-63.

Abstract (if available)

Abstract Adolescent Latinos report high levels of depression and suicidality compared to youth from other cultural backgrounds, with Latinas reporting the highest rates. Despite this well-documented ethnic disparity, few studies have investigated why Latinos are at disproportionately high risk. Cultural process models argue that intercultural discrepancy between Latino youth and their immigrant caregivers contributes to poor mental health outcomes through its mediating effects on family functioning. This study provided an empirical test of this model, with a primary focus on youth depression and suicidality. This study also evaluated the contributing role of fatalism, a variable excluded from prior cultural process models. Self-report measures assessing acculturation, gender role beliefs, familism, family functioning, fatalism, depression, and suicidality were administered to 395 Latino students in a Los Angeles high school. Results showed that youth-caregiver discrepancy in familism was significantly associated with depression, with this effect partially mediated by family conflict and cohesion. Discrepancy in familism was also significantly associated with suicidal ideation, with this effect partially mediated by family conflict. Results also showed that family conflict was significantly associated with depression, suicidal ideation, and negative thoughts about suicide, with these effects partially mediated by fatalism. Moderator analyses suggested that the relationship between discrepancy in gender role beliefs and depression may be more robust for females than for males. Treatment implications and directions for future research are discussed.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Sociocultural stress, coping and substance use among Hispanic/Latino adolescents

PDF

Depression and suicidality in Latino adolescents: A study of acculturation and gender role beliefs

PDF

Perceived discrimination and Latino youth adjustment: examining the influence of relinquished control and immigration status

PDF

Psychosocial and behavioral ractors associated with emotional eating in adolescents

PDF

Family conflict, negative mood, and adolescents' daily problems in school

PDF

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

PDF

Drinking alcohol to improve mood partially mediates the relation between major depression and alcohol dependence

PDF

Social network influences on depressive symptoms among Chinese adolescents

PDF

The effects of familism and cultural justification on the mental and physical health of family caregivers

PDF

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

PDF

Acculturation and Latino immigrants ' health attitudes and behaviors: a bi-dimensional acculturation model

PDF

Investigating discrimination and depression within a couple context

PDF

Community violence exposure and children's subsequent rejection within the school peer group: the mediating roles of emotion dysregulation and depressive symptoms

PDF

Psychosocial adjustment of Latina cervical cancer patients

PDF

Opportunity beliefs and behavioral outcomes in Latinx youth: an exploration of Ogbu’s cultural-ecological model

PDF

Emotion regulation as a mechanism linking parents’ marital aggression to adolescent behavioral problems: a longitudinal analysis

PDF

The successful African American Parenting (SAAP) Program: an evaluation of a culturally sensitive parenting intervention

PDF

Spouse aggression, depression, and physical health: a multivariate longitudinal study of midlife couples

PDF

Cumulative risk as a moderator of multisystemic therapy effects for juvenile offenders

Asset Metadata

Creator Céspedes, Yolanda Maria (author)

Core Title Evaluating a cultural process model of depression and suicidality in Latino adolescents

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Publication Date 06/27/2010

Defense Date 05/14/2008

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

Tag Adolescent,Depression,OAI-PMH Harvest,suicidality

Language English

Advisor Huey, Stanley J., Jr. (committee chair), Gatz, Margaret (committee member), Margolin, Gayla (committee member), Unger, Jennifer B. (committee member), Wilcox, Rand R. (committee member)

Creator Email cespedes@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1298

Unique identifier UC1103580

Identifier etd-Cespedes-20080627 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-80336 (legacy record id),usctheses-m1298 (legacy record id)

Legacy Identifier etd-Cespedes-20080627.pdf

Dmrecord 80336

Document Type Dissertation

Rights Céspedes, Yolanda Maria

Type texts

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

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu