Article Summary

Maternal Familismo and Early Childhood Functioning in Mexican and Dominican Immigrant Families

Esther J. Calzada, Keng-Yen Huang, Heliana Linares-Torres, S. Diana Singh, and Laurie Brotman

New York University School of Medicine

A large theoretical and empirical literature documents the central role of familismo (i.e., a strong emphasis on family) in the functioning of Latino youth. Few studies, however, have examined its association with early childhood functioning. The present study explored the potential risk and protective effects of maternal familismo on the adaptive and mental health functioning of 4- to 5-year-old Latino children. A sample of 205 Mexican and 147 Dominican immigrant families was recruited from New York City. Mothers reported on their level of familismo and acculturative status. Mothers and teachers rated child adaptive behavior and internalizing and externalizing problems. Findings suggest that maternal familismo is not uniformly associated with positive or negative early developmental outcomes but that its effects are moderated by child gender, family poverty, and cultural (e.g., maternal ethnic and U.S. American identity) characteristics. In addition, different mechanisms were identified for each ethnic group. Familismo was associated both positively (for boys) and negatively (for poor children) with adaptive behavior in the Mexican American sample. In the Dominican American sample, familismo showed a wide range of positive, albeit moderated, effects. Preven- tion efforts that help parents critically evaluate the impact of familismo on family processes, and preserve those manifestations of familismo that are protective, may best promote Latino child well-being.

Keywords: adaptive behavior, early childhood, familismo, mental health functioning

As a traditional, collectivistic cultural group, the Latino population is believed to adhere deeply to the value of familismo (Arditti, 2006), which refers to mutual support and obligation between family members (Baca-Zinn & Wells, 2000; Calzada, Tamis-LeMonda, & Yoshikawa, 2012; Harrison, Wilson, Pine, Chan, & Buriel, 1990; Keefe, 1984). One definition (of several offered in the literature) suggests that familismo comprises four core tenets: (a) belief that family comes before the individual, (b) familial inter- connectedness, (c) belief in family reciprocity,

and (d) belief in familial honor (Lugo Steidel, & Contreras, 2003), as well as its behavioral man- ifestations include financial support, shared liv- ing, shared daily activities, shared child rearing, and support for immigration (Calzada, Huang, & Brotman, 2012). With its reliance on the family unit for instrumental and emotional sup- port and family as referent, familismo has been shown to impact child development across sev- eral domains of well-being. Still, little is known about the association between familismo and child functioning during early childhood, a pe- riod during which key developmental compe- tencies take shape. The present study examined maternal familismo in relation to children’s functioning in a sample of Latino immigrant families of 4- and 5-year-old children.

Familismo and Child Developmental Competencies

Traditionally, research on familismo with La- tino immigrant youth has examined its protec-

This article was published Online First June 30, 2014. Esther J. Calzada, Keng-Yen Huang, Heliana Linares-

Torres, S. Diana Singh, and Laurie Brotman, Child Study Center, New York University School of Medicine.

Esther J. Calzada is now at the School of Social Work, University of Texas, Austin

Correspondence concerning this article should be addressed to Esther J. Calzada, The University of Texas School of Social Work, 1925 San Jacinto Boulevard, D3500, Austin, TX 78712-0358. E-mail: esther.calzada@austin.utexas.edu

Journal of Latina/o Psychology © 2014 American Psychological Association 2014, Vol. 2, No. 3, 156–171 2168-1678/14/$12.00 http://dx.doi.org/10.1037/lat0000021

156

mailto:esther.calzada@austin.utexas.edu
http://dx.doi.org/10.1037/lat0000021

tive effects because of the multiple benefits typically associated with the availability of a large family unit comprising nuclear and ex- tended family members. For example, familismo may enhance the development of adaptive behavior among Latino children via modeling as their parents attend to the instru- mental and emotional needs of extended family members. In addition, Latino children raised in highly familistic homes may be more likely to have responsibilities that benefit the family unit, such as assisting with childcare of younger fam- ily members (Calderón-Tena, Knight, & Carlo, 2011). As children are taught to put family needs before their own, they may be more likely to develop sensitivity for others’ needs.

Similarly, there is some evidence that familismo may be protective in the development of school functioning. In homes where familismo is valued, youth have higher class attendance and apply greater academic effort as they are motivated to do well in school for the sake of the family (Esparza & Sánchez, 2008; LaRoche & Shriberg, 2004). Parental monitor- ing and parent involvement in education asso- ciated with familismo have been found to pro- mote academic achievement (Loukas & Prelow, 2004; Niemeyer, Wong, & Westerhaus, 2009), and the social capital afforded by family net- works has been linked to higher academic per- formance (Esparza & Sanchez, 2008) and more informed educational decisions (Valdez & Ro- driguez, 2002). At the same time, however, family obligations may interfere with academic success as they put a toll on children’s time and energy that can lead to school absences, school dropout and lower rates of college enrollment (Desmond & Lopez-Turley, 2009; LaRoche & Shriberg, 2004; Suárez-Orozco & Suárez- Orozco, 2001; Velez, 1989). Based on a study with a national sample of immigrant high school students, Portes (1999) concluded that it is “moderate attachment to one’s family” (p. 502, emphasis added) that promotes academic achievement.

Other studies corroborate the potential risk caused by very high levels of familismo for youth mental health (Fuligni, Tseng, & Lam, 1999; Marsiglia, Kulis, Parsai, Villar, & Garcia, 2009). Most notably, Zayas and colleagues showed that suicidal Latina adolescents ex- plained their suicidal ideation in terms of sac- rificing for their families (Nolle, Gulbas, Kuhl-

berg, & Zayas, 2012). Yet others find that the family routines and cohesion that characterize familismo protect adolescent girls against inter- nalizing problems (Loukas & Prelow, 2004; O’Donnell, O’Donnell, Wardlaw, & Stueve, 2004), and that family conflict and a lack of family support (i.e., a lack of familismo) place youth at risk for internalizing problems (Coat- sworth et al., 2002).

Familismo is expected to reduce risk for ex- ternalizing behaviors, including substance use, by ensuring a strong attachment to family and fostering conventional ties that dictate higher levels of parental monitoring and give parents more influence over their adolescent’s behavior in ways that discourage Latino youth from en- gaging in a variety of problem behaviors (Brooks, Stuewig, & LeCroy, 1998; Keefe, Pa- dilla, & Carlos, 1978; Loukas & Prelow, 2004). Recent evidence further suggests that familismo may be protective against externalizing prob- lems via its association with the use of positive parenting practices (Calzada, Huang, & Brot- man, 2012; Santisteban, Coatsworth, Briones, Kurtines, & Szapocznik, 2012). But whereas some studies suggest that familismo is indeed protective against substance use (Gil, Wagner, & Vega, 2000; Horton & Gil, 2008) and behav- ior problems (Gamble & Modry-Mandell, 2008; Germán, Gonzales, & Dumka, 2009; Marsiglia, Parsai, & Kulis, 2009), including violent behav- ior (Estrada-Martínez, Padilla, Caldwell, & Schultz, 2011), others find no significant asso- ciation (Pabon, 1998; Shih, Miles, Tucker, Zhou, & D’Amico, 2010, 2012; Venegas, Coo- per, Naylor, Hanson, & Blow, 2012). A study with Mexican American youth found that high familismo may even increase adolescent girls’ risk for externalizing problems when mani- fested in the context of perceived discrimination (Delgado, Updegraff, Roosa, & Umaña-Taylor, 2011).

Evidence for Moderation

In light of the inconsistencies in the literature, Calzada, Tamis-LeMonda, and Yoshikawa (2012) posit that familismo manifests along a continuum in which costs and benefits coexist and may confer risk or protection depending on the ecological and developmental context of the child. For example, familismo may be protec- tive for children of mothers who are employed

157FAMILISMO AND EARLY CHILDHOOD FUNCTIONING

outside the home because it ensures the avail- ability of additional caregivers, but it may have no effect on children whose mothers are not employed. In representing a set of interrelated risk and protective factors, familismo may also exert unique effects on different domains of Latino youth functioning. Children of employed mothers may be protected against externalizing problems as their adult family members closely monitor their behavior, but for these same chil- dren, familismo may have no impact on inter- nalizing functioning. Evidence for moderation by demographic characteristics comes from a study of island and mainland Puerto Rican youth which found that familismo was protec- tive against antisocial behavior for girls regard- less of age (i.e., 5–13 years), but for boys only when they were younger (i.e., 5–9 years old; Morcillo et al., 2011).

A growing literature also shows how cultural context interacts with familismo to influence child functioning. For example, parent–adoles- cent conflict is more predictive of internalizing problems for adolescents with higher Latino cultural involvement, presumably because the dictates of familismo exacerbate the effects of conflict within the family (Smokowski & Ba- callao, 2007). For adolescents with higher in- volvement in U.S. American culture, and for bicultural adolescents, risk for internalizing and externalizing symptoms may be lower (Loukas & Prelow, 2004; Gonzales et al., 2008; Schwartz, Zamboanga, & Hernandez Jarvis, 2007; Smokowski & Bacallao, 2007). In addi- tion, aspects of the home environment (e.g., parenting practices; Peña et al., 2011) and fam- ily relations (e.g., mother-daughter mutuality; Baumann, Kuhlberg, & Zayas, 2010) have been found to moderate associations between familismo and suicidality.

The Present Study

The present study builds on the extant liter- ature to examine the association between ma- ternal familismo and the functioning of 4- and 5-year-old Latino children. Almost all past stud- ies of familismo have been conducted with older youth; virtually nothing is known about mater- nal familismo and early child development. From a theoretical perspective, early childhood marks a critical juncture in development as early ecological factors tend to activate a cas-

cade toward later developmental outcomes (Dodge et al., 2009), making its study ideal for the identification of risk and protective factors that are expected to have long-term impact. We examined early childhood outcomes in three domains: adaptive behavior, externalizing prob- lems, and internalizing problems; and in two settings: home and school. We considered adap- tive behavior as the primary outcome because we expect that in young children, the influence of familismo on adaptive skills may be more immediate whereas its influence on mental health may emerge over time, as problems crys- talize (or are curtailed).

Consistent with the goal of examining both risk and protection, we conduct a more nuanced examination of familismo by considering con- textual characteristics that may moderate its as- sociation with Latino child functioning. We consider child gender and acculturative status, given past evidence (albeit with older children) of moderation (i.e., Morcillo et al., 2011; Smokowski & Bacallao, 2007), and poverty given theoretical arguments regarding its poten- tially moderating role (Calzada, Tamis-LeM- onda, & Yoshikawa, 2012). For the sake of parsimony, we limit moderation analyses to these three variables, recognizing the overlap with other potentially relevant factors (e.g., poverty status is highly correlated with moth- er’s employment and marital status and so the latter two were not included).

Importantly, we studied two specific ethnic groups because according to the integrative de- velopmental theory, ethnicity, as a social posi- tion variable, is associated with a host of social mechanisms that ultimately influence develop- mental outcomes (García-Coll et al., 1996). Consistent with this view, Calzada, Tamis- LeMonda, and Yoshikawa (2012) have argued that the manifestation of familismo, and its in- fluence on developmental trajectories, are con- text-dependent, a context defined in large part by a child’s ethnicity. Neither of these concep- tual models, however, addresses the question of how precisely to define ethnicity (i.e., as Latino–a pan-ethnic categorization, or as Mex- ican or Dominican American–a specific ethnic group categorization). In choosing to operation- alize ethnicity with more specificity (i.e., based on country of origin), we are, in essence, mak- ing it an independent variable in our study. Concerns over using race or ethnicity as an

158 CALZADA, HUANG, LINARES-TORRES, SINGH, AND BROTMAN

independent variable have been expressed by scholars (Helms et al., 2005) based on the prem- ise that race/ethnicity is simply a proxy for underlying concepts and as such, a meaningless variable for psychological research. In the pres- ent study, we too view ethnicity as a proxy, but recognize that the underlying factors and pro- cesses are not yet identified or understood by scholars, or were simply not measured in our study. In the absence of these data, we rely on ethnic categorizations.

We consider additional advantages of our approach to defining ethnicity with more preci- sion. When ethnicity is operationalized by collapsing individuals from various Spanish- speaking countries into one category, within- group differences along historical, social, and cultural dimensions are likely obfuscated, mak- ing it difficult to interpret study findings (Szapocznik, Prado, Burlew, Williams, & San- tisteban, 2007). Alternately, our categorization based on country of origin recognizes that in certain but unknown ways, a child’s ethnic group membership influences his or her devel- opment. Finally, sampling based on country of origin is consistent with the goal of exploring developmental mechanisms within specific eth- nic groups, rather than examining generaliza- tion of such mechanisms across groups, and thus was the approach used in the present study.

The first group we selected was Mexican Americans (MA) because the majority of U.S. Latinos (65%) come from Mexico, and though they have not historically resided in the North- east, MAs are poised to become the largest group in New York City (NYC; where the pres- ent study took place) by 2021. The second group was Dominican Americans (DA), who are expected to surpass Puerto Ricans to be- come the largest group, after Mexicans, in New York City (Bergad, 2011). Significant gaps ex- ist in the study of MA and DA families to date. DAs have been neglected in the literature, such that virtually nothing is known about family processes and child development in this group although they have had a long-standing pres- ence (i.e., more than 50 years) in the northeast- ern United States. MAs have been the focus of a relatively large and growing literature, but consistent with past population trends, studies have been primarily limited to samples from traditional receiving communities in the West- ern United States. Migration to other states,

such as New York, is shifting the demographic profile of the MA population, providing oppor- tunities to examine the generalizability of past study findings to MA families residing in new receiving communities.

Method

Participants

Participants were drawn from a prospective longitudinal study examining the early child- hood development of Mexican (MA) and Do- minican American (DA) children. In the larger, ongoing study, children who were enrolled in the first two years of recruitment (2010 and 2011; n � 412), who had complete data (n � 383; 93%) and who had immigrant mothers (n � 352; 92%) were included in the present study sample. Despite our interest in MA and DA mothers across generations, we excluded the 31 children of U.S.-born mothers because we did not have the power to analyze whether there were meaningful differences on study variables based on mother’s nativity. Mother– child dyads were recruited from prekindergar- ten and kindergarten classrooms from 24 public schools in NYC. All children were 4 or 5 years old at the time of enrollment. As shown in Table 1, MA (n � 205) and DA (n � 147) families differed on most demographic characteristics, reflecting the unique ecology of each group. For example, compared with DA mothers, MA mothers were more likely to be married to or living with the child’s father and to have larger families. MA mothers were also younger, less likely to have graduated from high school, and less likely to be working for pay. The average income was $19,685 (SD � 12,876) for MA families and $24,250 (SD � 21,101) for DA families. In both groups, household income had a wide range (up to $93,000 in the MA sample; up to $150,000 in the DA sample), but only 7% of MA mothers, and 12% of DA mothers, re- ported an income above $40,000. In considering poverty status, MA mothers and their children were more likely than DAs to be classified as poor.

Measures

Demographic characteristics. Mothers completed an extensive demographic form that

159FAMILISMO AND EARLY CHILDHOOD FUNCTIONING

included information on date of birth (mother and child), immigrant status, years of residence in the United States, employment status, marital status, household income (brought in by any person within the home who shared household expenses with the mother), and household com- position. To calculate poverty, we considered income relative to number of persons living in the home for whom the mother was financially responsible or with whom she was sharing household expenses. We used the federal pov- erty guidelines, which considers number of per- sons in the home, to categorize families as poor, defined as living 100% below the poverty line.

Maternal acculturative status. The Ab- breviated Multidimensional Acculturation Scale (AMAS; Zea, Asner-Self, Birman, & Buki, 2003) is a 42-item measure of acculturative status that was developed with Spanish- and English-speaking Latinos. Items are rated on a scale of 1 (strongly disagree) to 4 (strongly agree) and correspond to specific domains of cultural competence, language competence, and identity; higher scores indicate higher levels of competence or identity. All domains are mea- sured for both culture of origin (enculturation) and mainstream/ “U.S. American” culture (ac- culturation), allowing for an examination of ac- culturation/enculturation as a bidimensional construct. The AMAS subscales showed ade- quate internal consistencies with the present sample of MA and DA Spanish- and English- speaking mothers (alphas ranged from .77–.93).

Familismo. The Mexican American Cul- tural Values Scales (MACVS; Knight et al.,

2010) assesses several Mexican cultural values including familismo. The familismo scale has 16 items rated from 1 (not at all) to 5 (completely) that tap into the desirability to maintain close relationships (“emotional support”), the impor- tance of tangible caregiving (“obligation to family”), and the reliance on communal inter- personal reflection to define the self (“family as referent”). Though the MACVS was developed for MA families, familismo is a relevant con- struct for all Latinos including DA families (Calzada, Tamis-LeMonda, & Yoshikawa, 2012); confirmatory factor analyses showed reasonable model fit indices for both MA and DA mothers (RMSEA � .06 for both samples). The MACVS was developed in Spanish and English (Knight et al., 2010) and showed high internal consistency (.83–.98) in both lan- guages, and with both ethnic groups, in the present study.

Child functioning. The Behavior Assess- ment System for Children, Parent Rating Scale and Teacher Rating Scale (BASC PRS, BASC TRS, Kamphaus, Reynolds, & Hatcher, 1999) is a measure of child functioning for children be- tween the ages of 2 and 18 years. Composite scales include Adaptive Behavior, Externalizing Problems, and Internalizing Problems. The Adaptive Behavior composite includes scales of social skills and adaptability, the Externalizing Problems composite scale includes subscales of aggression and conduct problems, and the In- ternalizing Problems composite scale includes subscales of depression, anxiety and somatiza- tion. The BASC-P was standardized with a

Table 1 Demographic Characteristics for Mexican American and Dominican American Families

Characteristic

MA (n � 205) DA (n � 147)

Mean SD Mean SD t p

Child age 4.68 .59 4.68 .54 .05 .96 Maternal age 31.34 5.79 34.04 7.45 �3.81 �.001 Household size 5.59 1.81 4.46 1.37 6.35 �.001

% % �2 p

Child gender: male 49.27 48.30 0.03 .86 Child foreign-born 3.52 17.81 19.88 �.001 Two-parent home 88.29 62.59 32.51 �.001 Mother � high school education 61.95 18.37 66.08 �.001 Mother works for pay 24.39 62.59 51.89 �.001 Family lives in poverty 83.90 60.54 24.37 �.001

Note. MA � Mexican American; DA � Dominican American. Family poverty status based on federal guidelines, with consideration for number of persons living in the home.

160 CALZADA, HUANG, LINARES-TORRES, SINGH, AND BROTMAN

small sample of Spanish-speaking Latino fami- lies and showed adequate psychometric proper- ties (Kamphaus, Reynolds, & Hatcher, 1999). In the present study sample, internal consisten- cies for MA and DA children ranged from .81 to .87 for parent report, and .90 to.95 for teacher report.

Procedure

Families were recruited for a longitudinal study of Latino child development from prekin- dergarten and kindergarten classrooms in 24 partner public schools. Specifically, MA fami- lies were recruited from 13 elementary schools, and DA families from 11 different elementary schools in the city. Consistent with the aims of the larger study, three inclusion criteria were applied, and children of mothers who self- identified as MA or DA, who were newly en- rolled in the school, and who had no docu- mented developmental disorder were invited to participate in the project. At each school, bilin- gual research staff attended parent meetings and were present during daily school drop-off and pick-up to inform mothers of the study. The recruitment rate averaged 74% (53–94%) across the 24 schools; 16% of families identified as eligible could not be contacted during recruit- ment efforts and 10% refused to participate. Interested mothers were scheduled for an ap- pointment at their child’s school where they were met by research staff for an interview. Mothers were asked to indicate their language of preference for the study activities: the major- ity of mothers (99% of MA and 86% of DA) chose to be interviewed in Spanish. After ob- taining consent, interview questions and re- sponse choices were provided to mothers in written form (via a response booklet) and were also read aloud by research staff; mothers’ oral responses were recorded. Upon completing the 90-min interview, mothers were paid a small stipend for their participation. Teachers of par- ticipating children (99% of mothers consented to the collection of teacher report) were then asked to consent and complete a packet of ques- tionnaires on child functioning. Teachers re- ceived a stipend for classroom supplies and hands-on support in the classroom as incentive for participating in the project. One hundred forty-four teachers provided data for 95% (n � 194) of MA and 92% (n � 135) of DA partic-

ipant children. There were no differences on demographic characteristics or on study vari- ables between children who had teacher ratings and those who did not. All of the data used in the present study were collected at Time 1.

Approach to Analyses

To study the association between maternal familismo and child functioning, we modeled child outcomes as a function of familismo. We conducted analyses separately for each child outcome, and separately for MA and DA fam- ilies because, as described above, our primary interest was in examining child development within, not across, ethnic groups (i.e., ethnicity as a sampling strategy). In addition, this analytic approach is more appropriate when variances are not equivalent across groups, as was the case on several key variables (e.g., U.S. Amer- ican identity, ethnic identity, English and Span- ish language competence, familismo) in the present study.

To examine whether associations between familismo and child functioning were moder- ated by select contextual characteristics, two models were tested— one examined demo- graphic characteristics as moderators, and the other examined cultural characteristics as mod- erators. We included child gender and family poverty status as potential moderators repre- senting family demographics. We included ma- ternal ethnic identity, U.S. American identity, and English and Spanish language competence as potential moderators representing familial cultural context. We focused on maternal (rather than child) acculturative status because of the young age of participating children and because of a growing literature documenting the importance of parental acculturative status for Latino child development (Calzada, Huang, & Brotman, 2012; Gonzales, Knight, Morgan- Lopez, Sanz, & Sirolli, 2002). In this model, we did not control for demographic variables be- cause none were significantly correlated with the familismo scale.

In each set of analyses, we included familismo, a set of moderators (demographic or cultural), and a set of interaction terms between familismo and each potential moderator. A sig- nificant interaction effect would indicate that the association between familismo and a child outcome is different based on the level/status of

161FAMILISMO AND EARLY CHILDHOOD FUNCTIONING

the moderator. For variables with significant interaction effects, post hoc regression analyses were carried out to further understand whether there was an association and if so, how strong, between familismo and the child outcome for each subgroup defined by the moderating vari- able. A limitation of the post hoc analyses is the limited power to detect subgroup differences given the relatively small n of each subgroup, thus we focused on the pattern of associations in these interactions as well. Our approach is mod- eled after developmental studies with similar aims (Huang, Caughy, Miller, & Genevro, 2005; Lee, Halpern, Hertz-Picciotto, Martin, & Suchindran, 2006). Although all results are pre- sented in the text, only significant interactions on adaptive behavior, our primary outcome, are plotted in figures.

For behavior at school, some teachers pro- vided data on multiple students (the average number of students rated by each teacher was 2.33 [SD � 1.48], with less than 1% of teachers rating �5 students). To account for correlations between outcomes of children rated by the same teachers (Muthén & Satorra, 1995), we applied linear mixed effects models (using SAS PROC MIXED procedure) and included teacher as a random effect. Again, all analyses were con- ducted separately for each child outcome and for MA and DA families.

Results

Maternal Acculturative Status, Familismo, and Child Functioning

Descriptive statistics for all study variables are presented in Table 2. In general, DA moth- ers were more acculturated (i.e., had higher US American identity and English language com- petence) than MA mothers, though neither group was highly acculturated. MA mothers reported a significantly higher ethnic identity than DA mothers, though ethnic identity was high for both groups. DA mothers reported higher Spanish language competence than MA mothers, perhaps because of the items related to literacy. All mothers reported high levels of familismo, but familismo was significantly higher among DA than among MA mothers. According to ratings from both mothers and teachers, DA children had higher levels of adap- tive behavior (at home and school) than MA Ta

bl e