Neighborhood Contexts, Fathers, and Mexican American Young Adolescents’ Internalizing Symptoms

Abstract

The family stress model posits that contextual stressors, such as neighborhood danger, negatively influence youth adjustment, including internalizing symptoms, via disruptions in parenting and family processes. The current study examined a culturally and contextually modified family stress model in a diverse sample of Mexican origin fathers and their children (N = 463) from the Southwestern U.S. Results supported the hypothesized negative influence of neighborhood danger on youth internalizing symptoms via disruptions in family cohesion. Paternal warmth did not play a role in linking contextual stress to outcomes. The role of harsh parenting was highly nuanced. Results suggest that both culture and context have the potential to moderate putative family stress model associations for specific parenting behaviors and further our understanding of the ways that culture and context may operate in models of family stress and youth outcomes.

Research, mostly focused on European American or low-income African American mothers and their children, has long established that the quality and quantity of family processes influence children’s mental health and that the stressfulness of the context in which families reside impacts the quality of those processes (Conger, Conger, & Martin, 2010; McLoyd, 1990). The influence that living in a stressful context has on fathers’ contributions to family processes and that those processes have on Mexican American youth are largely overlooked, despite the fact that Mexican Americans are a large population with disparate rates of adolescent mental health problems, especially internalizing spectrum disorders (Delva et al., 2005).

Explanations for such disparities often focus on the population’s disproportionate exposure to contextual risk factors, such as poverty and poor quality neighborhoods (Jargowsky & Bane, 1990). But, much of the work on the effects of neighborhood quality on youth has focused on single-parent, female-headed families, perhaps because of their prevalence among the low-income European American and African American families that have been the focus of most neighborhood research (e.g., Sampson, Raudenbush, & Earls, 1997). In contrast, two-parent households are prevalent among Mexican American families, including those living in poverty (Upchurch, Aneshensel, Mudgal, & McNeely, 2001) and fathers are notably influential (Cabrera & Garcia-Coll, 2004). Even when studies of Latinos or Mexican Americans have included two-parent families, fathers’ roles in childrearing generally have not been examined (Cabrera, Shannon, West, & Brooks-Gunn, 2006); a few notable exceptions showed these fathers were highly involved (Coltrane, Parke, & Adams, 2004; Parke et al., 2004; Plunkett & Bámaca-Gómez, 2003).

Guided by the family stress model, Conger and colleagues have repeatedly demonstrated that family (e.g., relationship conflict, family arguments, interparental conflict and withdrawal) and parenting (e.g., management, warmth, hostility, inconsistent discipline ) processes play a role in linking parental stress to child outcomes, including internalizing spectrum problems (Conger, Patterson, & Ge, 1995; Conger et al., 2002; Conger, et al., 2010). Though mediation was not always tested directly, the body of work suggests that both family processes and parenting processes act simultaneously to mediate the association between parental stress and child internalizing problems (see Conger, et al., 2010 for a review). The same model has been used to describe associations between neighborhood pressures (e.g., living in dangerous neighborhoods), parental functioning, parenting and family processes, and youth outcomes (Gonzales et al., 2011; Kotchick, Dorsey, & Heller, 2005; White, Roosa, Weaver, & Nair, 2009), but only a few studies employing family stress perspectives have tested it with Mexican Americans, or applied it to Mexican American fathers (Behnke et al., 2008; Parke et al., 2004). The model posits that stressors lead to disruptions in family and parenting processes (e.g., higher levels of negative processes and lower levels of positive processes), in turn placing youth at risk for behavioral health problems.

Bioecological perspectives suggest that the influence stressors have on proximal processes and that these processes have on outcomes may vary with cultural and contextual circumstances (Bronfenbrenner & Morris, 1998), painting a more nuanced picture of the ways in which contextual stress influences child outcomes than is typically conveyed by family stress model research. In the current study we examined how fathers’ perceptions of neighborhood danger were related to key parenting/family processes and, in turn, how those processes influenced Mexican origin youth internalizing symptoms. Because the current study successfully sampled (a) Mexican origin families ranging from immigrants to fifth generation and (b) diverse neighborhoods, we were able to address noted gaps in the family stress literature (Conger et al., 2010) by examining two sources of variation in the model: cultural values and neighborhood disadvantage. The former was hypothesized to moderate the associations between fathers’ perceptions of neighborhood danger and family processes; the latter the associations between these processes and youth internalizing symptoms.

Neighborhood Danger, Fathers’ Parenting Behaviors, and Youth Internalizing Symptoms

Neighborhood danger is a chronic contextual stressor (Attar, Guerra, & Tolan, 1994; McLoyd, 1990). Research has shown that parents’ perceptions of their neighborhoods may be as, or more important than objective assessments of neighborhoods in determining the manner in which they respond, parentally, to neighborhood threats (O’Neil, Parke, & McDowell, 2001). Though some researchers have questioned the value and meaning of examining the link between parents’ perceptions of neighborhoods and child outcomes (Burton & Jarrett, 2000), the family stress model places parents’ perceptions as central in the transmission of risks to their children, via disruptions in parenting (Conger et al., 1999). Because neighborhood perceptions have been shown to vary systematically as a function of individual characteristics (Roosa, White, Zeiders, & Tein, 2009), fathers’ perceptions of neighborhood danger may be expected to directly influence family/parenting processes, but only indirectly relate to youth internalizing symptoms.

The family stress model emphasizes parenting and family processes as mediators in the association between contextual stressors and adolescent outcomes. Three parenting/family processes were examined in the current study because of their centrality to (a) family stress models, (b) neighborhood research, (c) cultural variation, and (d) internalizing spectrum problems among youth. First, research, some of which has included Mexican Americans, generally has shown that lower levels of warmth and supportive parenting play a role in linking contextual stressors and internalizing problems (see Conger et al., 2010). For instance, Conger and colleagues have shown that family economic stress among European Americans and African Americans is negatively related to parental warmth, which is, in turn, negatively related to internalizing symptoms (Conger et al., 1995; Conger et al., 2002). More recent work with Mexican origin fathers also has shown a negative association between neighborhood danger and paternal warmth (White et al., 2009). Culturally, warmth in the parent – child relationship may be especially important as a means of maintaining harmonious interpersonal relations, a goal of traditional Mexican American families (Halgunseth, Ispa, & Rudy, 2006).

Second, family cohesion represents an important family construct that may be especially influenced by fathers’ perceptions of neighborhood danger because the challenges of living in poor quality, dangerous neighborhoods limit support resources (Bowen, Bowen, & Ware, 2002; Furstenberg et al., 1993; McLoyd, 1990), including the amount of support individuals in families can provide (Deng et al., 2006; Roosa et al., 2005). Further, family cohesion is consistently related to youth internalizing problems (Herman, Ostrander, & Tucker, 2007; Lucia & Breslau, 2006) and may be especially relevant to a Latino subgroup like Mexican Americans (Behnke et al., 2008), who tend to have a more collectivist orientation than European American or African American families (Oyserman, Coon, & Kemmelmeier, 2002)

Third, harsh parenting is relevant as a mediator of the association between perceived neighborhood danger and youth internalizing symptoms because harsh and coercive forms of control have been found to positively relate to both low quality neighborhoods (Hill & Herman-Stahl, 2002) and internalizing symptoms (Conger et al., 2002). Though several studies have failed to identify control-related parenting strategies as mediators of neighborhood influences on child outcomes (Caughy, Nettles, & O’Campo, 2008; Colder, Mott, Levy, & Flay, 2000; Peeples & Loeber, 1994), these either used an underpowered test of mediation or parenting measures lacking evidence of reliability and validity, and none focused on Mexican Americans. Findings from traditional family stress model research have consistently shown that higher stress is associated with more reliance on harsher forms of parental control, and, in turn, more adjustment problems (Conger et al., 2002). Several studies have found that Latino and Mexican American parents evince more harshness than European American parents (Bulcroft, Carmody, & Bulcroft, 1996; Hill, Bush, & Roosa, 2003), a difference that may have its origin in cultural values (Halgunseth et al., 2006), the meaning attributed to control behaviors (Deater-Deckard, Dodge, & Sorbring, 2005), or high rates of exposure to disadvantaged developmental contexts that may heighten parental reliance on control-related behaviors as a means of protecting children from perceived dangers (Kelley, Power, & Wimbush, 1992).

Cultural and Contextual Moderation in the Family Stress Model

The ways in which Mexican American fathers respond to perceived neighborhood circumstances may be moderated by their cultural values (Coltrane et al., 2004; Halgunseth et al., 2006). Familism, which has been shown to positively influence fathering (Coltrane et al., 2004), is a multidimensional value particularly important to Mexican origin families that refers to the salience of family obligations and the inclination to rely on family as a primary source of support (Knight et al., 2010). Support is an important resource that can compensate for the detrimental effects of stress on parenting (McLoyd, 1990), but forms of support specific to Mexican Americans, like familism, have not been examined in this framework.

In neighborhood research, the people and/or services available in neighborhoods are characterized as a source of support (Elliott et al., 1996), but their availability is often quite low in dangerous contexts (Bowen et al., 2002; Furstenberg et al., 1993; McLoyd, 1990). Indeed, multiple support constructs have been found to be protective in the association between stressors and parenting behaviors in past family stress model research (Leinonen, Solantaus, & Punamäki, 2003). From a family stress perspective, highly familistic fathers in dangerous neighborhoods may be protected from putative deficits in parenting (e.g., lower warmth and family cohesion; Coltrane et al., 2004) because they are not looking outside of the family, to the neighborhood, for support. Though family stress model research generally conceives of higher harsh parenting as a deficit, we recognize that harsh parenting may be a special case because parents living in unsafe areas may intentionally employ more harsh parenting strategies to safeguard their children from potential threats in the neighborhood (Furstenberg et al., 1993; Letiecq & Koblinsky, 2004). In the context of this literature, highly familistic fathers may use harsher parenting because of their stronger sense of obligations to their families and their need, bolstered by family support, to protect their families from neighborhoods they perceive to be dangerous.

Neighborhood disadvantage may moderate the association between specific parenting behaviors and youth outcomes (see Kotchick & Forehand, 2002 for a review). Early work showed racial differences in the effects of certain control-related strategies on youth externalizing outcomes, such that harsher forms of control were only detrimental to European American, not African American youth (Deater-Deckard, Dodge, Bates & Pettit, 1996). Deater-Deckard and colleagues (2005) later demonstrated the influence of a given parenting strategy could vary by context, perhaps because (a) the demands of the developmental niche influence the strategy’s effects on youth and/or (b) there is variation in the cultural meaning attached to different parenting strategies. Prior work has found that heightened control may be protective for externalizing problems in risky neighborhoods (Beyers, Bates, Pettit & Dodge, 2003). Dearing’s (2004) research showed the relations between restrictive parenting and parental warmth with child outcomes varied across neighborhoods and ethnicities. For example, warmth was protective for depression in low quality neighborhoods only, and restrictive parenting had different impacts depending on neighborhood, race (African vs. European American), and outcome examined. We aim to continue extending the examination of (a) the contextual variability of the relation between parenting and internalizing symptoms and (b) the contextual variability of putative family stress associations by simultaneously examining whether paternal warmth, family cohesion, and paternal harshness are equally suited to diverse neighborhood contexts.

In sum, the current study tests a culturally and contextually modified family stress model in a diverse sample of Mexican origin fathers and their children. We first hypothesize (H 1.1 – 1.3) that fathers’ perceived neighborhood danger will be associated with higher youth internalizing symptoms via disruptions in parenting and family processes (1.1. paternal warmth; 1.2. family cohesion; 1.3. paternal harshness). We next hypothesize (H 2.1 – 2.3) that familism values will be protective in the association between perceived danger and parenting/family processes [i.e., moderate the association between perceived danger and (2.1) paternal warmth, (2.2) family cohesion, and (2.3) paternal harshness]. We recognize the distinct culturally based possibility that more familistic fathers may view harsh parenting as adaptive in dangerous neighborhoods (see Kotchick & Forehand, 2002 for a review). Within this cultural framework, higher familism values may be associated with a positive association between perceived danger and harsh parenting. We also examine the potential that neighborhood disadvantage will render a given process more or less beneficial (H 3.1 – 3.3, for paternal warmth, family cohesion, and paternal harshness, respectively). We expect that paternal warmth and family cohesion may be more beneficial in disadvantaged contexts. It is difficult to predict whether the results for paternal harshness will be more consistent with research among African American or European American families. Finally, we explored gender differences in our model because differences in rates of internalizing symptoms appear starting in early adolescence (Nolen-Hoeksema, 2001), but evidence that gender moderates the association between context and process or process and outcome is mixed (Deng et al., 2006; Simons, Johnson, Beaman, Conger, & Whitbeck, 1996).

Method

Data for this study are from a study of the roles of culture and context in the lives of Mexican American families (Roosa et al., 2008). Participants were 750 Mexican American students in 5^th grade and their families living in a large metropolitan area of the the Southwestern U.S. Eligible families met these criteria: (a) they had a 5^th grader attending a sampled school; (b) mother and youth agreed to participate; (c) the participating mother was the youth’s biological mother, lived with the youth, and self-identified as Mexican or Mexican American; (d) the youth’s biological father was of Mexican origin; (e) the youth was not severely learning disabled; and (f) no stepfather or mother’s boyfriend was living with the youth. The sample was diverse in terms of social class, neighborhoods, and generation status, and was similar to the census description of this population on education and family income (Roosa et al., 2008). Although participation was optional for fathers, 467 (81.9%) fathers from the 570 two-parent families in the study also participated; 463 father – youth dyads had complete data for the current study. Those families in which fathers participated were similar to those two-parent families in which father did not participate on income, child nativity, father nativity, child gender, internalizing symptoms, and child reports on paternal warmth and paternal harshness. Father-participating families came from less disadvantaged neighborhoods (M_father = .00, M_nofather = 0.80 [F (1, 576) = 8.07, p < .01]). In the final sample 70.5% of the youth, 26.0% of the mothers, and 20.1% of the fathers were born in the U.S. Most fathers (76.6%) completed the interview in Spanish; only 17.2% of youth chose to do so. Average annual income was $35,001 – $40,000. Fathers, on average, completed 10.1 (SD = 3.94) years of schooling. Mean age of youth and fathers was 10.4 (SD = 0.55) and 38.1 (SD = 6.27) years, respectively.

The complete research procedures are described elsewhere (Roosa et al., 2008). Here we summarize key features of these procedures, noting that all study materials were available in English and Spanish. Using a combination of random and purposive sampling, the research team identified communities served by 47 public, religious, and charter schools from throughout the metropolitan area that represented the economic, cultural, and social diversity of the community. Recruitment materials were sent home with all children in the 5^th grade in these schools. Computer Assisted Personal Interviews (about 2 hours) were completed (from fall, 2004 to spring, 2006) with 750 families, 73% of those eligible. Participants were paid $45 each.

Measures

Perceptions of neighborhood danger

Fathers reported their perceptions of neighborhood danger using a 3-item subscale of the Neighborhood Quality Evaluation Scale (Roosa et al., 2005), a measure with demonstrated evidence of cross language equivalence (Kim, Nair, Knight, Roosa, & Updegraff, 2009). Fathers indicated their levels of agreement ranging from (1) not true at all to (5) very true on items such as “It is safe in your neighborhood” (reverse coded). Higher scores reflected a higher sense of danger. In the current study α = .88.

Parenting processes

Fathers and children reported on fathers’ parenting using the Children’s Report of Parental Behavior Inventory (CRPBI), a measure with demonstrated evidence of cross language equivalence (Nair, White, Knight, & Roosa, 2009). Only the acceptance and harsh parenting subscales of the CRPBI were used in the current study; both had a Likert response scale ranging from (1) almost never to (5) almost always. The 8-item acceptance subscale assessed warmth within the parent – child relationship; the 8-item harsh parenting subscale assessed control attempts with negative affect. Children were asked how often their fathers performed the behaviors (e.g., “Your father screamed at you when you did something wrong”); fathers reported on their own behaviors. Reliability for acceptance (α_father = .74, α_child = .87) harsh parenting (α_father = .69, α_child = .76) were acceptable.

Family processes

Fathers reported on the strength of emotional bonds among family members using the cohesion subscale of the Family Adaptability and Cohesion Evaluation Scales (Olson, Portner, & Bell, 1982), a measure with demonstrated evidence of cross language measurement equivalence (Knight & Hill, 1998). The subscale contained 16 items assessing how often specific family behaviors (e.g., “Your family gets together in the same room”) occurred: (1) almost never or never to (5) almost always (α = .81).

Familism

Two dimensions of familism values (i.e., family support and emotional closeness, obligations to family) were measured using fathers’ responses to the familism subscale of the Mexican American Cultural Values Scale (MACVS, Knight et al., 2010). Because the subscale scores were highly correlated (r = .57, p < .0001) they were combined into a single score of familism. Participants were asked to rate how much they agree or disagree with each item (e.g., “It is important for family members to show their love and affection to one another”): (1) not at all to (5) completely (α = .70).

Neighborhood disadvantage

All neighborhood indicators were captured from the 2000 Census. For each neighborhood, data of the percent of families below the poverty level, the percent of the male population 16 years and over in the labor force who were unemployed, and the percent of the population 25 and over who had not graduated from high school in each census tract were used as indicators of disadvantage (c.f., Deng et al., 2006; Sampson et al., 1997).

Youth internalizing symptoms

We used the computerized version of the Diagnostic Interview Schedule for Children (DISC-IV) to assess youth internalizing symptomatology (Shaffer, Fisher, Lucas, Dulcan, & Schwab-Stone, 2000). Symptom counts for anxiety and mood disorders were summed to represent internalizing symptoms, consistent with prior work (e.g., Garland et al., 2001; Ge, Brody, Conger, & Simons, 2006). Mother and youth were administered the DISC independently and their reports were combined to represent the average report of symptoms across reporters (Shaffer et al., 2000).

Analytic Strategy

The culturally and contextually adapted family stress model was estimated in Mplus (Muthén & Muthén, 1998) using the COMPLEX option; standard errors of path coefficients were adjusted to account for neighborhood clustering. Because only four families had incomplete data for the current study, list wise deletion was used. All variables were standardized (i.e., converted to z-scores) based on the means and standard deviations at the individual family level. Interaction terms were created by calculating the product of the two (standardized) variables of interest and using the product as a manifest variable (Tein, Sandler, MacKinnon, & Wolchik, 2004). Follow up analyses for interactions were conducted according to Aiken and West (1991). Mediation effects were tested using the product of coefficients method with the multivariate delta method of deriving the standard error (Sobel, 1982). Standardized path coefficients are presented for models, which can be interpreted as the number of standard deviations change in the outcome for a 1 standard deviation change in the predictor. Good (acceptable) model fit is reflected by a nonsignificant chi-square test, comparative fit index (CFI) greater than .95 (.90), root mean square error of approximation (RMSEA) less than .05 (.08), and standardized root mean square residual (SRMR) less than .05 (.08) (Kline, 2005).

Both father and youth reports were available for paternal warmth and harshness. Research shows only low to moderate levels of parent – child agreement on parenting measures, suggesting that the preferred analytic strategy is to run separate analyses for each reporter (Tein, Roosa, & Michaels, 1994). In preliminary analyses, multigroup models with child gender as the grouping variable were examined. Two multigroup models were compared: a fully fixed model in which all model paths were constrained to be the same for boys and girls and a fully free model in which all model paths were allowed to vary. Nested model tests comparing these two models indicated that the fully free model did not fit significantly better than the fully fixed model (χ² = 9.95 (16), NS for father model, χ² = 7.02 (16), NS for child model). This suggests that the paths in the model did not differ significantly across child gender. Thus, all models presented in the current paper are for the entire sample of boys and girls. Because of well documented associations between adolescent internalizing symptoms and both family income and youth gender (Anderson & Mayes, 2010; Nolen-Hoeksema, 2001), we controlled for both in all model tests.

Results

For correlations among the study variables and descriptive statistics, see Table 1. Intraclass correlations (ICC) were used to quantify the degree of resemblance among families within neighborhoods. The highest ICCs were for fathers’ perceptions of neighborhood danger: 17% of the variance was associated with the level 2 unit (τ₀₀ = 0.15, p < .05). All other individual-level variables had ICCs of 7% or less.

Table 1.

Correlation matrix, means, and standard deviations for study variables(N = 463).

	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.
1. Neighborhood Disadvantage	1	0.30**	0.01	0.02	−0.02	−0.04	−0.04	−0.01	−0.02	−0.30***
2. Neighborhood Danger		1	−0.21***	−0.08	0.04	0.03	−0.23***	0.08	−0.11*	−0.28***
3. Warmth (P)			1	0.03	−0.07	−0.05	0.37**	−0.07	0.27***	0.05
4. Warmth (C)				1	−0.01	−0.13**	0.12**	−0.09*	0.01	0.06
5. Harsh Parenting (P)					1	0.15**	−0.10*	0.06	0.02	−0.01
6. Harsh Parenting (C)						1	−0.03	0.13**	−0.02	−0.01
7. Cohesion							1	−0.14**	0.19***	0.13**
8. Internalizing								1	0.02	−0.05
9. Familism									1	−0.10*
10. Income										1
M	0.00	2.41	4.20	4.34	1.94	1.91	3.95	15.85	4.39	7.74
SD	2.52	0.94	0.54	0.72	0.58	0.74	0.51	9.10	0.40	4.61

Note.

^* p < .05;

^**= p < .01;

^***= p < .001;

Test of Hypothesized Culturally and Contextually Adapted Family Stress Model

Father-report model

The fit of the hypothesized father-report model was adequate (χ² (24) = 55.66, p < .001, CFI = 0.71; RMSEA = 0.05; SRMR = 0.04), but improved to good (χ² (23) = 21.74, NS, CFI = 1.00; RMSEA = 0.00; SRMR = 0.03) when fathers’ reports of warmth were permitted to correlate with family cohesion. Figure 1 shows the results for this model. Girls demonstrated higher internalizing symptom means than boys. Hypotheses 1.1 – 1.3 received partial support. Fathers’ perceptions of neighborhood danger were associated with less paternal warmth, but the latter was unrelated to internalizing symptoms (1.1). Family cohesion (1.2) mediated the association between perceived danger and internalizing symptoms (z = 0.03, p < .05). Paternal harshness did not mediate the association between perceived danger and internalizing symptoms (1.3). Fathers’ familism values were positively related to paternal warmth and family cohesion, and negatively related to internalizing symptoms via family cohesion (z = − 0.02, p < .05).

Figure 1.

Father report (child report) process model of paternal perceptions of danger, family/parenting processes, and youth internalizing symptoms (N = 463).

Note. Paths significant (p < .05) in at least one model are represented by solid lines; paths marginal (.05 < p < .10) in at least one model are represented by a long dash; nonsignificant paths in both models are dashed.

Hypotheses regarding cultural (2.1 – 2.3) and contextual (3.1 – 3.3) moderation of the family stress model received mixed support. Familism values did not moderate the association between perceived danger and paternal warmth (2.1), or perceived danger and family cohesion (2.2). Perceived danger interacted with fathers’ familism values to influence paternal harshness (marginal, p < .10), which provided some evidence in support of Hypothesis 1.3. Neighborhood disadvantage did not moderate the association between the parenting behaviors and internalizing symptoms (Hypotheses 3.1 – 3.3).

Due to strong contemporary interest in cultural moderation of the family stress model (Conger et al., 2010) and cultural precursors of paternal parenting (Behnke, Taylor, & Parra-Cardona, 2008), we performed simple slopes analysis of the marginal familism interaction. The pattern of the interaction was such that there was a marginally positive association between perceived danger and paternal harshness at 1 standard deviation above the mean on familism values (z = 0.11, p = .07), and there was no relation at 1 standard deviation below the familism mean (z = −0.03, NS).

Child-report model

The fit of the hypothesized child-report model (Figure 1) was good (χ² (24) = 28.76, NS, CFI = 0.92; RMSEA = 0.02; SRMR = 0.03). Girls demonstrated higher internalizing symptom means than boys. Hypotheses 1.1 – 1.3 received partial support. Children’s reports on paternal warmth did not mediate the association between perceived danger and internalizing symptoms (1.1). Family cohesion (1.2) mediated the association between perceived danger and internalizing symptoms (z = 0.03, p < .05). Fathers’ perceived danger was not associated with paternal harshness, but the latter was positively related to internalizing symptoms (1.3). Fathers’ familism values were positively related to family cohesion and negatively related to internalizing symptoms via family cohesion (z = − 0.02, p < .05).

Hypotheses regarding cultural (2.1 – 2.3) and contextual (3.1 – 3.3) moderation of the family stress model received mixed support. Though there was no main effect of perceived danger to paternal warmth, the former did interact (p < .05) with fathers’ familism values to influence paternal warmth (2.1). Familism did not moderate the relation between perceived danger and family cohesion (2.2), or paternal harshness (2.3). Neighborhood disadvantage did not moderate the relation between paternal warmth (3.1), or family cohesion (3.2) and internalizing symptoms; it moderated the association between paternal harshness and internalizing symptoms, which provided support to Hypothesis 3.3.

The pattern of the interaction between fathers’ perceived danger and paternal warmth was such that there was a negative association at 1 standard deviation above the mean on familism values (z = −0.15, p < .05), and there was no association at 1 standard deviation below the mean on familism values (z = 0.01, NS). The pattern of the interaction between paternal harshness and objective neighborhood disadvantage was such that there was no association at 1 standard deviation below the mean on neighborhood disadvantage (z = −0.02, NS) and a positive association at 1 standard deviation above the mean (z = 0.27, p < .001).

Discussion

This study applied the family stress model to increase our understanding of how perceptions of neighborhood danger may influence Mexican American fathers’ behaviors and, ultimately, their early adolescents’ internalizing symptoms. Based on the interactive nature of development (Bronfenbrenner & Morris, 1998), the current study examined cultural and contextual variation in the family stress model. The use of an ethnic homogenous research design, a very diverse, multigenerational sample, and the inclusion of cultural values offers a strong test of the hypothesized model’s applicability to Mexican Americans (Garcia Coll et al., 1996) and guides the development of further research questions regarding the ways that culture may influence development (Quintana et al., 2006). The focus on Mexican American fathers is a substantial contribution because two-parent families are common among Mexican origin families living in the U.S. (Upchurch et al., 2001), Mexican American fathers are highly involved in parenting (Coltrane et al., 2004), and fathers are largely absent from the literature on neighborhood effects.

Guided by the family stress model, we hypothesized that parenting and family processes would mediate the association between fathers’ perceived neighborhood danger and youth internalizing symptoms. Our study hypotheses were partially supported. Family cohesion mediated the association between fathers’ perceptions that their neighborhoods were dangerous places to live and youth internalizing symptoms. Higher perceived neighborhood danger was associated with lower levels of family cohesion and, in turn, higher levels of internalizing symptoms for youth. This finding held across genders, as well as diverse levels of fathers’ familism values, and neighborhood disadvantage. The mediation findings are strengthened because (a) perceived danger and internalizing symptoms were assessed from different reporters (minimizing the impact of shared method variance), and (b) we controlled for family income (ruling out within neighborhood income homogeneity as an explanation for our findings).

Neighborhood theory suggests that low quality neighborhoods undermine mutual trust and closeness among neighbors (Sampson et al., 1997). Consistent with other studies (Deng et al., 2006; Kotchick et al., 2005), our results suggest that dangerous neighborhoods undermine parallel processes within families. Family stress perspectives suggest Mexican origin families’ cohesion is negatively influenced by the stress of living in dangerous neighborhoods, and this, in turn, is associated with higher levels of internalizing symptom among youth. Our findings represents an extension of the traditional family stress model to Mexican American fathers and suggests that family-wide processes, like cohesion, may be susceptible to the stress that fathers experience as part of living in dangerous neighborhoods. Findings parallel a large body of research on mothers’ parenting in other ethnic groups (see Conger et al., 2010 for a review). The lack of moderation in the cohesion model may reflect the importance of cohesion for Mexican origin families across diverse cultural and neighborhood contexts (Deng et al., 2006).

Contrary to our study hypothesis, harsh parenting did not mediate the association between perceived danger and youth internalizing symptoms. Rather, it played a far more nuanced role in the model, adding to the developing understanding of harsh parenting in Latino families (Halgunseth et al., 2006). For fathers’ report on harsh parenting, the association with neighborhood danger was moderated by fathers’ familism values, consistent with our cultural moderation hypothesis. The association between children’s reports on harsh parenting and internalizing symptoms was moderated by neighborhood disadvantage, providing support for our contextual moderation hypothesis.

Familism values moderated the association between perceived neighborhood danger and fathers’ report on harsh parenting. We found a positive (trend level) association between fathers’ perceptions of neighborhood danger and paternal harshness in the context of high familism values, and no association between fathers’ perceptions of neighborhood danger and paternal harshness in the context of low familism values. Findings confirm that cultural values are a source of variability in how parents respond to perceived contextual threats, such as neighborhood danger (Halgunseth et al., 2006). The family stress model interpretation is that only those fathers high on familism experienced the putative parenting deficits (i.e., higher paternal harshness) associated with the stress of living in a dangerous neighborhood. That interpretation suggests strong familism values place families at risk.

An alternate interpretation recognizes the distinct cultural possibility that more familistic fathers intentionally use harsher parenting in light of perceived danger. Parents living in unsafe areas may use more authoritarian parenting strategies to safeguard their children from potential threats (Furstenberg et al., 1993). Fathers with high familism values place a greater emphasis in their families on maintaining close relationships, emotional support, and tangible care giving (Knight et al., 2010). In response to threats, these fathers engage in higher degrees of harsh parenting, perhaps trying to safeguard their children from the dangers of the neighborhood. Consistent with conceptual work on factors that shape parenting practices (Kotchick & Forehand, 2002), familism values influenced the ways in which these fathers respond to neighborhood danger. Highly familistic fathers may use harsher forms of parenting because they have a stronger sense of obligation to their families, and the necessary family support to try to protect their families from neighborhoods they perceive to be dangerous. More work is needed to examine the competing explanations offered by the family stress and cultural perspectives.

The association between youths’ reports on paternal harsh parenting and internalizing symptoms was moderated by the degree of disadvantage in the neighborhood. These findings build on research suggesting that the consequences of harsh parenting may vary in different contexts (Deater-Deckard et al., 2005), but also highlight the complexity of generalizing these association across diverse populations (Simons et al., 2002). We found a positive association between harsh parenting and internalizing symptoms when youth lived in highly disadvantaged neighborhoods and no association when youth lived in more advantaged ones. Our findings do not replicate prior work with African Americans, showing that parenting behaviors characterized as restrictive were more protective in high-risk neighborhoods (Dearing, 2004). Instead, our results are more consistent with an additive stress model, replicating prior work with European American children, which found that more restrictive parenting behaviors in non-Latino White families were associated with higher risk of child mental health problems (Dearing, 2004).

It is important to note that the current study focused on a very specific kind of control among fathers, namely punitive control in the form of harsh parenting, whereas Dearing (2004) studied parental control more generally among mothers. Similarly, most prior research has focused on the influence that mothers’ control behaviors have on youth; much less attention has been given to fathers’ control behaviors. It also is important to remember that ours was a very conservative test of this association; we tested the potential mediating effects of three family process variables simultaneously, providing a much more stringent test of each association. More research is needed to determine if our findings reflect differences between African American families and Mexican American families in disadvantaged contexts, or if our results reflect differences in the influence of paternal versus maternal harsh parenting.

Given the association between youth’s perceived paternal harshness and internalizing symptoms, traditional fathers’ choice to use harsh parenting in response to neighborhood threats appears questionable. Understanding the goals of parenting is important (Crick & Dodge, 1994), as goals are powerful organizers of behavior (Hastings & Grusec, 1998). Further, Halgunseth et al. (2006) suggest that understanding parenting goals may be especially important for research with Latinos because culture-specific, and therefore context-specific goals are at the root of many parenting decisions. For example, faced with perceived danger, these more familistic fathers may actively engage in harsher parenting in an attempt to protect their children from immediate physical harm. Doings so, nevertheless, may come at the expense of their children’s psychological health, which is not as immediately affected nor readily observed. More research is needed to examine these relations, especially considering that links from perceived danger to paternal harshness and from the latter to youth outcomes did not replicate across reporter.

Paternal warmth did not mediate the association between fathers’ perceived danger and adolescent internalizing symptoms. For both reporters, this finding held across boys and girls, and levels of neighborhood disadvantage. Fathers’ reports on their own warmth were negatively related to perceived danger, but this finding was only replicated by children’s reports when fathers had high familism values. Neither fathers’ nor children’s reports on paternal warmth were related to youth internalizing symptoms. Though some studies have found that warmth behaviors mediated associations between stressors and youth outcomes (e.g., Conger et al., 1995; Conger et al., 2002), a majority of previous examinations have focused on the influence that stressors have on parenting latent constructs operationalized by either diverse behaviors (e.g., warmth, monitoring, consistent discipline), or diverse reporters (e.g., parent, child, observer). The former approach assumes the influence of stressors is constant across behaviors; the latter approach eliminates much of the variance in the parenting behaviors assessed (Tein et al., 1994).

We looked at parenting behaviors separately because recent work with Latino fathers has demonstrated that parenting behaviors may be differentially influenced by the parenting context (Cabrera et al., 2006). Indeed, Cabrera et al.’s (2006) findings are further supported by the current work, wherein perceived neighborhood danger was negatively related to family cohesion (and, to some degree, paternal warmth) in a diverse sample of Mexican origin families, and positively related to harsh parenting behaviors, depending on fathers’ levels of familism values. Further, from a cultural perspective, it is interesting that the family-level process, cohesion, appeared to be more important to the degree of internalizing symptoms among Mexican American adolescents than a parallel process at the dyad level, paternal warmth. This finding may have its origins in the more collectivist family orientation present in many Mexican origin families (Oyserman et al., 2002) and the emphasis on believing the family is more important than the individual.

Fathers’ familism values moderated the association between perceived danger and youth reports on paternal warmth, such that only fathers with high familism values exhibited lower levels of warmth in response to perceived danger. This finding did not replicate when fathers’ own reports on warmth were used. Interpretation of this finding from the perspective of a traditional family stress model would suggest that only those fathers high on familism experience the putative parenting deficits (i.e., lower levels of paternal warmth) associated with the stress of living in a dangerous neighborhood. This interpretation again, would situate familism values as a risk factor. From a cultural perspective, and in light of the findings on paternal harshness, it may be that more familistic Mexican American fathers are pooling their parenting resources to demonstrate what they perceive as contextually relevant parenting behaviors, namely harsh parenting. They themselves apparently do not perceive the resultant deficits in paternal warmth, but their children do. Most importantly, this result must be interpreted in the context of the other parenting processes that were evaluated simultaneously; when fathers perceived their neighborhoods as dangerous, family cohesion was lower in general but paternal harshness was higher and paternal warmth lower only when fathers were highly familistic. This pattern of responses could be explained by familistic fathers choosing protection of their children from the dangers of the neighborhood as their immediate goal, with a reduction in the warmth of their interactions with their children as a side effect. Much more work is needed in this area and, once again, understanding Mexican American fathers’ parenting goals may make an important contribution to the literature (Halgunseth et al., 2006).

It is important to view the current study in light of its limitations. First, the current study was cross-sectional; longitudinal studies examining the family stress model among Mexican origin families are needed to understand the role of culture and context in these models and to properly test mediation. It also is important to note that our test of the family stress model was limited in its focus on parenting and did not test the full chain of family stress processes by including aspects of parental psychological well-being and marital conflict as additional mediators in the causal chain. Some of this work has already been done (Parke et al., 2004; White et al., 2009), but future investigations of a more complete model are warranted. Further, the fathers in this study were only from two-parent families. Fathers are likely influential in other family types, but those family types were excluded from the current study. Finally, it should be noted that multiple neighborhood theories and models suggest that parenting/family processes are only one set of potential mediators (Leventhal & Brooks-Gunn, 2000; Roosa, Jones, Tein, & Cree, 2003) and some theoretical treatments exclude parenting entirely (especially during late adolescence; Ingoldsby & Shaw, 2002). We chose to focus on parenting in the current study because of its centrality to the family stress model and because parenting and family processes are an important area of research for understanding developmental processes among Mexican American youth (Chao & Otsuki-Clutter, 2011).

The current study represents an ethnic homogenous investigation of the associations between Mexican American fathers’ perceived neighborhood danger, fathers’ parenting/family processes, and young adolescents’ internalizing symptoms. No neighborhood effects literature has focused specifically on fathers and very few family stress or neighborhood studies have focused on Mexican Americans. Following the advice of Conger et al. (2010) on the importance of examining moderation in the family stress model, we examined cultural and contextual variability by introducing familism values as a moderator of the association between perceived neighborhood danger and parent/family proximal processes, and neighborhood disadvantage as a moderator of the association between parent/family process and young adolescent outcomes. The attention given to the ways that culture and context moderate putative family stress model associations further our understanding of the role of cultural and social factors in family stress processes (Conger et al.). Our results suggest the degree to which specific cultural values and living conditions moderate putative family stress model associations among other diverse groups is an important area for future research. Future applications of the family stress model with minority populations should attempt to examine the model longitudinally, continuing to explore cultural and contextual adaptations that may make the model more appropriate for the targeted group and improve our understanding of differences in response to stressful conditions across populations. Finally, more work is needed to understand Mexican American fathers’ parenting goals. Such work could illuminate whether behaviors traditionally viewed as deficits in the family stress model are viewed by Mexican American fathers’ as adaptations to dangerous contexts.