Hispanic Fathers and Risk for Maltreatment in Father-Involved Families of Young Children

Abstract

The Hispanic population is the fastest growing segment of U.S. population. However, risks for child maltreatment in the foreign-born and native-born Hispanic populations are largely understudied. To address this knowledge gap, we explore the association of sociodemographic factors, psychosocial parenting factors, and nativity status with Hispanic fathers’ aggression toward their young children (3 to 5 years). Using the Fragile Families and Child Wellbeing Study and the follow-up In-Home Longitudinal Study of Pre-School Aged Children, we examine data for 372 foreign-born (FB; n = 155) and native-born (NB; n = 217) Hispanic biological fathers residing in the home when the study target child was 3 years old. Results of analysis at the bivariate level show FB Hispanic fathers engage in fewer aggressive behaviors than NB Hispanic, White, or Black fathers. Time-lagged path models of Hispanic fathers show FB Hispanic fathers use less aggression than NB Hispanic fathers. Length of time in the United States was not associated with parenting aggression. Path models also examine paternal psychosocial factors such as alcohol use, depression, parenting stress, and involvement in caregiving, and control for the child’s aggressive behavior. Results suggest one reason Hispanic children do not face heightened risk for child welfare involvement, despite socioeconomic risks, is that FB Hispanic fathers use less aggression toward their young children. An implication of this finding is that socioeconomic and parenting behavior risks must be considered separately when practitioners are considering issues related to the representation of minority children in the child welfare system.

The overrepresentation of minority children in the child welfare system has been a longstanding concern in social work. However, most research has focused on the overrepresentation of African American children relative to White children. Only recently have studies begun to examine Asian and Hispanic families’ involvement in the child welfare system. Interestingly, despite the Hispanic population’s lower socioeconomic status as compared with non-Hispanic Whites (Morales, Lara, Kington, Valdez, & Escarce, 2002), there is some evidence that Hispanic children are slightly underrepresented in the child welfare system (Dettlaff & Johnson, 2011; Zhai & Gao, 2009). In fact, a recent study using national child welfare data indicated that even though Hispanic families were 3 times more likely to live below the federal poverty level than White families, Hispanic children were at no greater risk for child welfare involvement than non-Hispanic White children (Drake et al., 2011).

Such findings imply the presence of protective factors that contribute to lower rates of child welfare involvement among Hispanic families. Among the considerable array of potential protective factors, we chose to examine Hispanic fathers’ aggressive parenting behaviors and psychosocial risk factors that may contribute to both use of aggressive parenting and child welfare involvement. A large proportion of the research on parenting risks and risk for maltreatment has focused on the characteristics of mothers in father-absent homes; however, far less is known about paternal characteristics and behaviors associated with maltreatment risk in father-involved households. This knowledge gap is amplified when focusing on Hispanics families, which is an oversight that is troubling for two reasons. First, among families involved with child welfare, immigrant households (40.6%) were more than twice as likely as native-born households (18.6%) to have a biological father present in the home (Dettlaff, Earner, & Phillips, 2009). Second, fathers are overrepresented as perpetrators of child maltreatment, including severe physical abuse and neglect that resulted in child homicide (Fujiwara, Barber, Schaechter, & Hemenway, 2009; Radhakrishna, Bou-Saada, Hunter, Catellier, & Kotch, 2001; U.S. Department of Health and Human Services, 2010; Stiffman, Schnitzer, Adam, Kruse, & Ewigman, 2002).

Further, the Hispanic population now constitutes the largest minority group in the United States, accounting for more than 16% of the U.S. population. Between 2000 and 2010, the U.S. Hispanic population increased by 43%, which accounted for more than half of the total U.S. population growth during that period (Ennis, Rios-Vargas, & Albert, 2011). Almost two fifths (39%) of Hispanics were born outside of the United States, comprising 47% of the U.S. foreign-born population (U.S. Census Bureau, 2009). However, even though the U.S. Hispanic population is rapidly growing and a large portion of that population segment is foreign-born, risks for child maltreatment in the foreign-born (FB) and native-born (NB) Hispanic populations are largely understudied.

The current study used a community sample of father-involved Hispanic families to examine predictors of paternal physical and psychological aggression toward young children between the ages of 3 and 5 years. Physically aggressive behaviors ranged from shaking a child (which is considered maltreatment in many cases; Runyan, 2008), to spanking (which has been shown to increase risk for physical child abuse; Straus, Hamby, Finkelhor, Moore, & Runyan, 1998). Paternal psychological aggression was also considered, including behaviors such as threatening to send the child away or calling the child dumb or lazy. The importance of psychological aggression as a potential risk factor for child maltreatment was demonstrated in a study that found it was uncommon for to use psychological aggression toward young children (Lee, Kim, Taylor, & Perron, 2011). Indeed, the study also showed that whether a father used psychological aggression was a distinguishing factor that differentiated the most aggressive fathers from their peers, suggesting that psychological aggression—particularly when directed toward very young children (i.e., 3 years)—may indicate heightened risk for maltreatment (Lee, Kim, et al., 2011).

Risk and Protective Factors for Parenting Aggression and Child Maltreatment

Informed by the developmental ecological model (Belsky, 1993), we sought to compare paternal aggression among FB and NB Hispanic fathers by examining individual and family psychosocial and demographic variables previously shown to be associated with child maltreatment and parental aggression. The developmental ecological model proposes that risk for child maltreatment is influenced by individual-level characteristics of the parent and the child, family-level factors, and the community-level context (i.e., community characteristics, collective cultural norms). In the current study, this model was particularly informative because the framework holds that the etiology of maltreatment is influenced by the developmental context, the specific characteristics of the parent or child that heighten risk for maltreatment, and the immediate interactional context or family processes (Belsky, 1993). In addition, this model illuminates how the processes that increase risk for child maltreatment can be influenced by characteristics specific to (a) the parent (e.g., nativity status, alcohol use, depression, age); (b) the child (e.g., gender, behavior problems); and (c) the family (e.g., interpersonal aggression and violence, poverty). It is important to note that the current study did not include community-or neighborhood-level data, which prevented us from examining the influence of the broader context in the etiology of child maltreatment. Therefore, we focus on the developmental context and factors that may influence the parent–child interaction, such as paternal psychosocial factors, child behavior problems, and family characteristics.

Psychosocial factors

Research has linked suboptimal parenting and risk for child maltreatment to a variety of psychosocial risks, including alcohol abuse (Dube et al., 2001; Kotch, Browne, Dufort, Winsor, & Catellier, 1999; Lee, Kim, et al., 2011; Lee, Perron, Taylor, & Guterman, 2011; Widom & Hiller-Sturmhofel, 2001); depression (Bronte-Tinkew, Moore, Matthews, & Carrano, 2007; Paulson, Dauber, & Leiferman, 2006); and parenting stress (Kotch et al., 1999; Taylor, Guterman, Lee, & Rathouz, 2009; Windham et al., 2004). Several recent studies have suggested that paternal alcohol use (Lee, Kim, et al., 2011; Lee, Perron, et al., 2011) and paternal depression (Lee, Taylor, & Bellamy, 2011) may be particular risk factors for maltreatment in father-involved families of young children.

Protective factors

Epidemiological studies have indicated that among the U.S. Hispanic population, parental characteristics such as foreign birth and lower acculturation to U.S. culture serve as protective factors for some parenting risks. As compared with either U.S.-born Hispanics or the U.S. population as a whole, FB Hispanic parents and those who were less acculturated to U.S. norms engaged in less frequent use of drugs and alcohol (Allen et al., 2008; Gil, Wagner, & Vega, 2000; Vega, Alderete, Kolody, & Aguilar-Gaxiola, 1998) and had lower rates of depression (Escobar, Nervi, & Gara, 2000; Vega, Alderete, et al., 1998; Vega, Kolody, et al., 1998). These differences appear to extend to Hispanic families involved in the child welfare system. In a study of FB and NB Hispanics involved in the child welfare system, the FB Hispanic parents reported less drug use and lower levels of parenting stress even though the immigrant Hispanic parents had lower income than their NB counterparts (Dettlaff et al., 2009). In the current study, we assessed a comprehensive set of psychosocial variables based on evidence linking parental psychosocial functioning to risk for child maltreatment, and evidence suggesting variation in these risk factors as a function of nativity status.

Nativity status and parental aggression

In addition to the child welfare literature, community-based studies of parental characteristics have also provided support for nativity status or lower acculturation as factors that may be directly associated with lower risk for child maltreatment. Compared with U.S.-born mothers, immigrant mothers reported lower levels of physical aggression, psychological aggression, and neglect of their young children; however, partner violence was associated with greater risk for maltreatment among immigrant mothers (Taylor et al., 2009). In a study from the Early Head Start project that compared White, Black, and Hispanic mothers, researchers found that immigrant Hispanic mothers who were less acculturated to the host culture spanked their young children (i.e., ages 2 to 3 years) less frequently than other mothers (Berlin et al., 2009). Likewise, a study of Hispanic mothers indicated that FB mothers were significantly less likely than NB mothers to use physical aggression, including spanking, with their 5-year-old children (Altschul & Lee, 2011). It is possible that immigrants have behaviors, such as higher levels of religiosity and traditional maternal health behaviors (e.g., Kimbro, Lynch, & McLanahan, 2008), and have fewer risk factors (e.g., alcohol use, depression) that are protective and associated with lower levels of parent-to-child aggression.

Similarly, although fewer studies have examined fathers, community-based studies have suggested that Hispanic fathers residing in the United States may also use less aggression toward their children and spank less frequently than White or Black fathers (Lee, Guterman, & Lee, 2008; Lee, Perron, et al., 2011). However, unlike the research with mothers, the studies of fathers did not assess paternal nativity status or measures of acculturation that might serve as protective factors, nor did these studies examine how nativity status might relate to common risk factors for maltreatment such as alcohol, parenting stress, and positive father involvement.

Nativity Status and Other Measures of Acculturation

Commonly used indicators of acculturation include an individual’s nativity status and years since immigration spent in the host culture. Nativity status assesses differences that may be observed over a lifetime, acting as a proxy for the context of early socialization or selection for better health and positive characteristics among immigrants (Franzini, Ribble, & Keddie, 2001). In Hispanic samples, nativity has been associated with adoption of host cultural practices (Schwartz, Pantin, Sullivan, Prado, & Szapocznik, 2006) and attitudes about gender norms (Phinney & Flores, 2002), both of which are considered to be indicators of acculturation. In studies examining aggressive parenting among Hispanic mothers, acculturation has been measured through a combination of generation status (e.g., FB, U.S.-born to immigrant parents, and U.S.-born to parents also born in the United States) and language use (Berlin et al., 2009) or nativity status alone (Taylor et al., 2009). Another study of mothers used indicators of religiosity, traditional gender norms, and length of time in the United States in addition to nativity status (Altschul & Lee, 2011). Similarly, our study of fathers used nativity status as an indicator of fathers’ early socialization context, and the number of years the father had been in United States as an indicator of the influence of the host culture on behaviors that may change more rapidly.

In addition, we examined attitudinal and behavioral changes that are likely to take place as individuals acculturate to U.S. norms, and may relate to fathers’ aggressive parenting behaviors. Traditional gender norms reinforce the father’s role as breadwinner and primary disciplinarian for the family, and research has linked the endorsement of those gender norms to greater use of physical punishment (Ferrari, 2002). Religiosity is another proxy for acculturation (Knight et al., 2009). Recent immigrants tend to attend religious services frequently, which may enhance access to social support networks that promote resiliency and better health outcomes (Arredondo, Elder, Ayala, Campbell, & Baquero, 2005; Gallo, Penedo, Espinosa de los Monteros, & Arguelles, 2009). These support networks may be a protective factor for recent immigrants who would otherwise be alienated and isolated from aspects of the host culture. Membership in a religious community may influence parenting practices by providing a source of community support and cohesion that help mitigate parenting and acculturative stress.

Study Hypotheses

The evidence has suggested that factors associated with nativity status among Hispanic parents play a protective role regarding aggressive-parenting practices. However, the available studies have focused solely on mothers, which is problematic given that behaviors of both parents affect children in two-parent families. Our examination of whether foreign nativity served as a protective factor for paternal aggression in father-involved Hispanic families was guided by two primary research questions.

The first research question was descriptive: Do overall rates of parenting aggression toward young children differ among a diverse community-based urban sample of NB Hispanic fathers, FB Hispanic fathers, White fathers, and Black fathers? We focused on paternal aggression toward young children at ages 3 and 5 years because parental aggression, such as spanking, peaks when children are 3-years-old and correspond to sharp increases in children’s acting-out behaviors (Straus & Stewart, 1999). We hypothesized that FB Hispanic fathers would engage in less physical and psychological aggression toward young their children as compared with NB Hispanic fathers, White fathers, or African American fathers.

Our second research question focused solely on Hispanic fathers to examine the hypothesis that FB nativity is a protective factor for paternal psychological and physical aggression toward the child at age 3 and age 5 years. Path model analyses used time-lagged models to account for temporal ordering of predictors relative to outcomes. Our analyses included fathers’ psychosocial characteristics, aspects of the home environment, and child behavior problems measured at 3 years and earlier. Consistent with Belsky’s ecological model, we controlled for child behavior problems because children who are more aggressive may elicit more aggression from their parents (Black, Slep, & Heyman, 2001; Patterson, 1982). Models included religious attendance and endorsement of traditional gender norms to assess the influence of these variables in predicting Hispanic fathers’ parenting aggression. We also examined whether greater exposure to U.S. norms of parenting and discipline influenced paternal aggression by assessing whether time spent in the United States was associated with paternal aggression among FB Hispanic fathers only.

Method

Participants

This study used data obtained from the Fragile Families and Child Wellbeing Study (FFCWS) core interviews and the adjunct In-Home Longitudinal Study of Pre-School Aged Children. FFCWS is a birth-cohort study (N = 4,898) conducted in 20 U.S. cities with populations exceeding 200,000 persons. A key element of the FFCWS study design was the oversampling of nonmarital births. The nonmarried parents and their children are referred to as fragile families because nonmarital unions are at greater risk for relationship instability and poverty as compared with marital unions (Carlson, McLanahan, & England, 2004; Reichman, Teitler, Garfinkel, & McLanahan, 2001).

The baseline FFCWS core interviews were conducted with mothers in the hospital following the child’s birth. Most fathers were also interviewed in the hospital, although fathers had the option of conducting the interview over the telephone. These initial, baseline interviews were followed with separate interviews of both parents when the child was approximately 1-year, 3-years, and 5-years-old (i.e., four data waves). The FFCWS core interviews gathered a broad range of data on household socioeconomic factors, parental health, parental relationship, parenting behaviors of mothers and fathers, and child well-being, including child’s behavior at age 3 and 5 years.

Verbal and written informed consent was obtained from participants at each interview, and participants were compensated for their involvement in the FFCWS study. All respondents were informed of the interviewers’ obligation to report observations of child abuse. A detailed description of the FFCWS sampling strategy and interview protocol has been published elsewhere (Reichman et al., 2001).

As an adjunct to the FFCWS core interviews, a subset of mothers (n = 3,288) who completed the Wave 3 interview (i.e., when child was 3 years old) were invited to participate in the In-Home Longitudinal Study of Pre-School Aged Children, which collected additional data on parenting, child health, and development. The In-Home study gathered data using a survey of the primary caregiver (usually the mother) and in-home assessments at two time points: when the child was 3 years old (the 3-year In-Home study) and 5 years old (the 5-year In-Home study). Fathers were not eligible to participate in this additional study component; however, if the father was residing in the home, then the In-Home study also collected the mother’s report of the father’s aggressive behavior toward the target child.

To compare levels of physical and psychological aggression among Hispanic (n=360), White (n=407) and African American (n=515) residential fathers (Table 1), we used data provided by their female partners in the In-Home study. All subsequent analyses used data collected from only self-identified Hispanic fathers who resided with the FFCWS study target child’s mother when the child was 3 years old. Table 2 presents complete descriptive statistics on all study variables.

Table 1.

Cross-Race Comparison of Mean CTSPC Scores (N = 1,282)

CTSPC	Hispanic FB	Hispanic NB	White	Black	F (df=4)
	M (SD)	M (SD)	M (SD)	M (SD)
Wave 3	n = 146	n = 214	n = 407	n = 515
Physical	3.39a (7.77)	7.71b (12.50)	9.35b (14.31)	10.46b (15.08)	14.11***
Psychological	7.01a (11.69)	14.93b (17.11)	15.99b (17.14)	17.81b (17.63)	17.12***
Wave 4	n = 101	n = 110	n = 370	n = 437
Physical	3.34a (7.42)	6.23b (10.54)	7.26b (11.76)	8.44b (14.23)	6.22***
Psychological	8.86a (11.93)	16.56b (16.78)	17.26b (17.08)	17.87b (18.38)	7.86***

Note. CTSPC = Parent-Child Conflict Tactics Scale; FB = foreign-born; NB = native-born. Different subscripts indicate significant mean differences at p < .05 as indicated by the Dunnett’s T3 post hoc comparison when Levene’s test is violated and Scheffe’s post hoc comparison when equal variances can be assumed.

^***p < .001

Table 2.

Study Variables, by Paternal Nativity Status

Variable (Range)	All Hispanic Fathers	Foreign-born (FB)	Native-born (NB)	t(df)c or χ2(df)
	n= 372 (100%)	n= 155 (42%)	n= 217 (58%)
	% or M(SD)	% or M(SD)	% or M(SD)
CTSPC Aggression Toward Child
Physical aggression W4 (0 – 75)‡	4.85 (9.27)	3.34 (7.42)	6.23 (10.54)	t(207) = −3.16**
Psych aggression W4 (0 – 75)‡	12.88 (15.13)	8.86 (11.93)	16.56 (16.78)	t(209) = −3.85***
Physical aggression W3 (0 – 73)‡	5.96 (11.03)	3.39 (7.77)	7.71 (12.50)	t(347) = −4.58***
Psych aggression W3 (0 – 75)‡	11.72 (15.62)	7.01 (11.69)	14.93 (17.11)	t(358) = −5.52***
Acculturation Indicators
Years in U.S. (FB only) (1 – 45)	13.19 (7.38)	-- --	-- --
Religious attendance (1 – 5)‡	2.91 (1.34)	3.09 (1.33)	2.77 (1.33)	t(370) = 2.26*
Traditional gender norms (1 – 4)‡	2.46 (0.64)	2.71 (0.62)	2.29 (0.60)	t(370) = 6.62***
Sociodemographic Factors
Age at child’s birth (16–61)	27.46 (6.59)	29.89 (7.12)	25.73 (5.58)	t(280) = 6.07***
Relationship status				χ2(2) = 4.13
Married	57%	59%	56%
Cohabiting	40%	40%	40%
Not married or cohabiting	3%	1%	5%
Education				χ2(2) = 28.30***
Less than high school	48%	63%	37%
High school degree or GED	29%	17%	38%
Some college/tech. school	23%	20%	25%
Household income($0–230,000)	$37,875 (29,047)	$34,766 (23,579)	$40,015 (32,162)	t(334) = −1. 43
Psychosocial Factors
Parenting stress (1 – 4)‡	2.04 (0.72)	2.05 (0.85)	2.03 (0.61)	t(231) = 0.23
Involvement with child (1 – 7)‡	4.45 (1.07)	4.19 (1.07)	4.63 (1.03)	t(334) = −3.77***
Major depression (yes)	9%	5%	11%	χ2(1) = 3.64
Heavy alcohol use (yes)	30%	24%	34%	χ2(1) = 3.68
Father-to-mother IPAV (yes)	43%	51%	38%	χ2(1) = 5.93*
Child Characteristics
Gender (boy)	51%	50%	52%	χ2(1) = 0.14
Aggressive behavior W3 (0 – 2)‡	0.55 (0.34)	0.49 (0.34)	0.59 (0.33)	t(368) = −2.65**
Anxious/depressed W3 (0 – 2)‡	0.38 (0.33)	0.42 (0.31)	0.36 (0.35)	t(284) = 1.43
Withdrawn W3 (0 – 2)‡	0.30 (0.33)	0.33 (0.37)	0.27 (0.29)	t(227) = 1.54

Note: CTSPC = Parent-Child Conflict Tactics Scale; W3 = Wave 3, child 3 years of age; W4 = Wave 4, child 5 years of age; psych = psychological; IPAV = intimate partner aggression and violence.

^*p ≤ .05,

^**p ≤ .01,

^***p ≤ .001, two-tailed, denotes significant difference between FB Hispanic and NB Hispanic fathers.

^‡Higher scores indicate higher levels of the construct.

Approximately 27% of all children in the In-Home study subsample had Hispanic fathers (n = 886), of whom 46% (n = 411) were residential fathers at the time of 3-year In-Home interview that collected data on paternal aggression. An additional 39 fathers were excluded from our analyses because their nativity status was unavailable, yielding a final analytic sample of 372 residential Hispanic fathers.

The large percentage of fathers excluded based on nonresidential status is attributable to the FFCWS design, which oversampled nonmarital births at baseline (Reichman et al., 2001). By the Wave 3 data collection point, fewer than half of fathers were residing in the home with the child. Because the FFCWS included assessment of aggressive parenting only for residential fathers, we were unable to examine differences in aggressive parenting behaviors between residential and nonresidential fathers.

Of the Hispanic fathers in our analytic sample (N=372), 54% indicated Mexico as their country-of-origin, 17% reported Puerto Rico, 17% “other” Hispanic country, 3% Central America, 3% South America, 2% Cuba, and less than 1% indicated their country-of-origin as Spain or other European country; 3% of the sample did not indicate a country-of-origin. The interview survey was available in English and Spanish; 69% of the FB Hispanic fathers and 3% of the NB Hispanic fathers completed the interview in Spanish. Spanish language use was highly associated with nativity status (χ² (1) =186, p < .000) and was strongly correlated with both nativity status (Spearman’s rho = .710, p < .001) and length of time in the United States (Spearman’s rho = -.625, p < .001).

Measures

Most variables were based on paternal self-report; maternal report was used when paternal report was not available or when using paternal report was inappropriate. For example, we used mothers’ reports of interpersonal violence perpetrated by the father because other studies have suggested that women’s reports of intimate partner violence are more valid than the reports of interpersonal violence made by the women’s spouses or partners (Edleson & Brygger, 1986).

At baseline interview (Wave 1), time-invariant demographic characteristics included the following paternal characteristics: education level, age, nativity status, and years in the United States (for FB fathers). In addition, the baseline interview also measured religious attendance and traditional gender norms. The FFCWS core interview at Wave 3 (i.e., when the child was 3 years old) measured time-varying demographic variables (marital status, family income) and psychosocial variables (parental stress, depression, alcohol use, and father involvement). Mothers’ reports were collected on the following areas: child’s sex (Wave 1), father-to-mother intimate partner aggression and violence (Wave 3), child behavior problems (Wave 3), paternal use of physical and psychological aggression (Wave 3 and Wave 4). Table 1 provides descriptive statistics and bivariate comparisons between FB and NB Hispanic fathers.

Parent-Child Conflict Tactics Scale

The FFCWS used the Parent-Child Conflict Tactics Scale (CTSPC; Straus et al., 1998) to assess father-to-child aggression regardless of whether the child was physically injured. The CTSPC uses the mother’s report on five items to assess the father’s psychological aggression toward the child. These five items asked mothers to indicate how often during the past year the father exhibited aggressive behaviors toward the child such as shouting, yelling, screaming, cursing at child; threatening to spank or hit child without actually doing so; threatening to send child away or kick child out of the house; and calling the child names such as dumb or lazy. Physical aggression was measured with five items that asked the mother how many times in the past year the father had exhibited physically aggressive behaviors such as shaking the child; pinching or slapping the child on hand, arm, or leg; spanking the child’s bottom with a bare hand; and hitting child’s bottom with a hard object.

Incidence rates differed significantly across aggressive acts. Following the recommendations of the CTSPC authors (Straus et al., 1998), we used a count variable that approximated the total number of physically and psychologically aggressive acts perpetrated toward the child in the prior year. Response categories and their contributions to the count variable were as follows: one occurrence; two occurrences in past year; three to five events (counted as four occurrences); six to 10 events (counted as eight occurrences; 11 to 20 events (counted as 15 occurrences); and more than 20 events (counted as 25 occurrences) in the past year. In addition, reports categorized as “zero occurrences” included reports that the type physical aggression had occurred but had not occurred in the past year, as well as reports that this type of physical aggression had never occurred.

Paternal acculturation

The measures of paternal acculturation were based on the fathers’ self-report of nativity status (0 = U.S.-born, 1 = foreign born); length of time in the United States, that is, years since arrival in the United States for immigrants; frequency of attending religious services during the past year, with responses ranging from hardly ever (1) to once a week or more (5); and endorsement of traditional gender norms. The endorsement of traditional gender norms was measured as the mean of two questions: “The important decisions in the family should be made by the man of the house” and “It is much better for everyone if the man earns the main living and the woman takes care of the home and family,” with response options ranging from strongly disagree (1) to strongly agree (4). Cronbach’s alpha was .58.

Paternal psychosocial risk

All variables for paternal psychosocial risk were self-reported by fathers. The measure for parenting stress was adapted from the Parenting Stress Index Short Form (PSI-SF; Abidin, 1995). Respondents were asked to choose among four response options ranging from strongly agree (1) to strongly disagree (4), indicating the extent to which they agreed with four statements. For example, the items included statements such as, “Being a parent is harder than I thought it would be” and “I feel trapped by my responsibilities as a parent.” This scale had a Cronbach alpha of .65. Responses were reverse-scored and a mean score created.

Father involvement with the child was based on a mean score of the number of days per week (0 = never to 7 = every day) the father provided 13 types of child care, including singing songs or nursery rhymes with child, hugging or showing physical affection to child, telling child that he loves him or her, reading stories to child, assisting child with eating, and putting child to bed. This measure reflects both routine child-care activities and activities that indicate an emotional connection to the child. A score was created to indicate the average number of days per week the father said he was involved in those activities (α = .78).

Major depression was assessed using criteria in the Diagnostic and Statistical Manual of Mental Disorders-III-R (American Psychiatric Association, 1987). Based on these criteria, major depression is diagnosed when a person has experienced depressed mood most of the day, nearly every day, for a period of 2 weeks or longer within the past year, and in combination with four other physical or emotional symptoms (0 = no, 1 = yes).

Alcohol use was assessed based on daily consumption of alcohol during the past 12 months. Four or more drinks consumed in a single day was defined as heavy alcohol use (coded as 1) whereas zero to three drinks was defined as low-risk alcohol use (coded as 0). Although our measure is less stringent than the DSM-III-R alcohol dependence diagnosis, only 3.1% of the Hispanic fathers met the DSM criteria for alcohol dependence. The study measure of heavy alcohol use approximates the National Institute on Alcohol and Alcoholism’s (2005) definition of heavy drinking for men, which is five or more drinks in a single day.

Father-to-mother intimate partner aggression or violence

Our measure of father-to-mother intimate partner aggression or violence (IPAV) was based on mothers’ self-report of seven items. Four of the seven items indicated psychological aggression (e.g., “He tries to keep you from seeing or talking with your friends or family”; Lloyd, 1996; Weiss & Margolin, 1977), and three items indicated physical aggression (e.g., “He slaps or kicks you”; Straus, Hamby, Boney-McCoy, & Sugarman, 1996). A dichotomous variable (0 = no, 1 = yes) was used to indicate any instance of IPAV.

Child behavior problems

Child behavior problems were measured using the aggressive behavior, anxious/depressed, and withdrawn behaviors subscales of the Child Behavior Checklist/1 1/2 – 5 (CBCL/1 1/2-5; Achenbach & Rescorla, 2000), which was completed by mothers. Responses to all items were coded using a 3-point ordinal scale (0 = not true, 1 = somewhat or sometimes true, 2 = very true or often true). The aggressive behavior subscale included 19 items such as “is defiant” and “gets in many fights” (α= .88). The anxious/depressed subscale included eight items such as “clings to adults or is too dependent” and “feelings are easily hurt” (α= .62). The withdrawn subscale included eight items such as “acts too young for age” (α= .66). Child gender was indicated by (0 = girl, 1 = boy).

Paternal demographic factors

Fathers’ self-reported demographics included his age at child’s birth, education level (1 = less than high school, 2 = high-school degree or GED, 3 = some college/technical school or higher), relationship status with child’s mother (1 = married, 2 = cohabiting, 3 = not married or cohabiting), and report of total gross household income from all sources (i.e., combined gross income for all jobs, assistance programs, and all persons living in the household). If fathers refused to answer or indicated they did not know their household’s gross income, they were asked to estimate a range for their income using a set of categorical responses.

As noted above, fathers were selected into this study based on whether the mother indicated the man was a residential father at the time of her enrollment in the 3-year In-Home study (i.e., approximately when the child was a 3-year-old). However, the study’s relationship status covariate was based on each father’s self-report at the time he completed the Wave 3 core interview, which also corresponded to child’s age of 3 years. At the Wave 3 core interview with fathers, a few fathers indicated they were neither cohabiting nor married to the child’s mother, which conflicted with the mothers’ reports of residential father status. We included these fathers in our analyses because the discrepant reports could be explained by multiple factors. For example, the parents’ relationship status might have changed in the time between the mother’s In-Home interview and the father’s Wave 3 core interview. Further, the variable measuring relationship status was nonsignificant in our analyses reported here (e.g., bivariate results presented in Table 2 and multiple variable models presented in Table 3). In other analyses not reported here, omitting these few fathers for whom report of residential status conflicted with the mother’s report did not change the study results.

Table 3.

Path Models Results for Predictors of Hispanic Fathers’ Use of Aggression

Predictors of Aggression, Measured at W3 or Earlier	Physical Aggressiona				Psychological Aggressiona
	Model 1: Phys Aggression at W3		Model 2: Phys Aggression at W4		Model 3: Psych Aggression at W3		Model 4: Psych Aggression at W4
	B(SE)	β	B(SE)	β	B(SE)	B	B(SE)	β
Acculturation Indicators
Nativity (foreign-born)	−0.29(0.11)	−0.11**	−0.59(0.21)	−0.25**	−0.62(0.19)	−0.21****	−0.68(0.29)	−0.24*
Religious attendance	0.03(0.04)	0.03	0.06(0.09)	0.06	−0.01(0.05)	−0.01	−0.05(0.05)	−0.05
Traditional gender norms	0.18(0.12)	0.09	0.23(0.14)	0.13	0.15(0.14)	0.07	0.31(0.15)	0.14*
Psychosocial Factors
Parenting stress	0.05(0.07)	0.03	0.01(0.10)	0.01	0.05(0.08)	0.03	0.07(0.09)	0.04
Involvement with child	0.10(0.07)	0.09	0.01(0.08)	0.01	0.00(0.08)	0.00	−0.04(0.08)	−0.03
Major depression	0.06(0.08)	0.05	0.03(0.15)	0.02	0.07(0.09)	0.05	0.31(0.20)	0.23
Heavy alcohol use	0.16(0.11)	0.14	0.33(0.15)	0.31*	0.20(0.11)	0.15	0.19(0.20)	0.15
Father-to-mother IPAV	0.10(0.12)	0.08	0.11(0.08)	0.09	0.24(0.06)	0.17***	0.21(0.12)	0.16
Child Characteristics
Child gender (boy)	0.06(0.11)	0.03	0.23(0.16)	0.10	0.24(0.13)	0.08	0.41(0.15)	0.15**
Child aggressive behavior	0.99(0.10)	0.27***	0.15(0.22)	0.04	1.16(0.19)	0.27***	0.57(0.29)	0.14
Child anxious/depressed	−0.43(0.31)	−0.12	−0.07(0.26)	−0.02	0.01(0.25)	0.00	0.40(0.15)	0.10**
Child withdrawn	0.32(0.18)	0.09	0.23(0.30)	0.06	0.02(0.33)	0.01	−0.14(0.38)	−0.03
Sociodemographic Factors
Age at child’s birth	−0.03(0.02)	−0.18*	0.01(0.01)	0.03	−0.02(0.02)	−0.10	−0.02(0.01)	−0.08
Relationship status	0.00(0.09)	0.00	−0.01(0.12)	−0.01	0.03(0.07)	0.02	−0.09(0.12)	−0.07
Education level	0.05(0.06)	0.05	−0.10(0.08)	−0.10	−0.00(0.07)	−0.00	−0.15(0.08)	−0.12
Household incomeb	0.04(0.03)	0.07	0.02(0.03)	0.03	0.12(0.02)	0.17***	0.04(0.04)	0.06

Note. CTSPC = Parent-Child Conflict Tactics Scale; phys = physical; psych = psychological; IPAV = interpersonal violence and aggression; W3 = Wave 3, child 3 years of age; W4 = Wave 4, child 5 years of age. Model fit indexes were the same for all models: χ² (10) = 11.72, p = .30, CFI = .99, RMSEA = .02.

^aAll Wave 3 and Wave 4 aggression variables were natural log transformed for analyses.

^bHousehold income was square-root transformed for analyses.

^*p ≤ .05,

^**p ≤ .01,

^***p ≤ .001

Analysis Plan

Analysis of variance (ANOVA) was used to compare mean levels of physical and psychological aggression among FB Hispanic, NB Hispanic, White, and African American residential fathers. These descriptive comparisons were conducted in SPSS using listwise deletion. Subsequently, we used path analyses to assess the complex relationships between nativity status, all other paternal and child variables, and fathers’ aggression toward children. Unlike regression analyses, path models within a structural equation modeling framework account for covariance between independent variables. First, to account for the reciprocal nature of interactions between fathers and their children (Belsky, 1993), we examined the within-time associations between psychosocial risk factors, child behaviors, and paternal physical aggression (Model 1) and psychological aggression (Model 3) when the child was 3 years old. These models capture the within-time association between children’s aggressive behavior and fathers’ use of aggression. A second set of models used time-lagged data to account for the temporal ordering of predictors relative to outcomes. In these models, paternal and child factors measured at age 3 years, along with socioeconomic and demographic covariates from baseline, were used to predict fathers’ use of physical aggression (Model 2) and psychological aggression (Model 4) when children were 5 years old (Wave 4). Using the second set of models, we also assessed whether fathers’ psychosocial factors mediated effects of the three acculturation factors: nativity, religious attendance, and endorsement of traditional gender norms. We followed the approach proposed by Preacher and Hayes (2008) for assessing multiple mediators simultaneously to estimate indirect effects of acculturation indicators on paternal aggression. Total indirect effects and specific indirect effects through each psychosocial risk factor were calculated using the product-of-coefficients approach, such that a total indirect effect is the sum of specific indirect effect through each mediator.

All path model analyses were conducted in Mplus 6.1. The chi-square test, the comparative fit index (CFI), and the root mean square error of approximation (RMSEA) were used to evaluate fit between the hypothesized models and observed data, with cutoff values of .95 for CFI and .06 for RMSEA establishing good fit (Hu & Bentler, 1999). Across variables measured at baseline, data were missing in 0% of cases; whereas, for covariates assessed at Wave 3, data were missing in approximately 10% of cases, with the exception of child behavioral measures that had data missing in 23% of cases. The two outcome variables, physical and psychological aggression, had missing data in approximately 5% of cases at Wave 3 and 37% of cases at Wave 4. The Wave 4 outcome variables were assessed 5 years after the baseline assessment, and there was significant attrition in the sample. As explained in the User Guide for the 5-year In-Home study, attrition is primarily attributed to (a) budget constraints that resulted in some individuals not being selected to participate in the 5-year In-Home study; and (b) interview fatigue (Fragile Families and Child Wellbeing Study, 2009).

Other studies using FFCWS data have found that longitudinal samples for which data was available on all variables differed in important ways from the overall FFCWS sample. Specifically, a study of mothers participating in the FFCWS, found that Latina and immigrant mothers had higher levels of study attrition (Cooper, McLanahan, Meadows, & Brooks-Gunn, 2009). Thus, similar to the approach taken by Cooper and colleagues, our analyses used full information maximum likelihood (FIML) estimation in Mplus to consider all cases and patterns of missing data patterns with the aim of avoiding missing data bias and maximizing the sample size. FIML is a preferred method of model estimation with missing data (Allison, 2003) and estimating models with missing data is preferred over using listwise deletion when data do not appear to be missing completely at random (Allison, 2003; Graham, 2009). In addition, all models were estimated with complete case analysis (listwise deletion), yielding similar results to those reported here. However, the complete case sample was less than half the size of the full sample, and thus, had diminished statistical power to identify significant relationships; consequently, some relationships were marginally significant (p < .10) with complete cases.

Results

Descriptive Results

Table 1 presents a cross-race comparison of the CTSPC scores at Wave 3 and Wave 4 for FB Hispanic, NB Hispanic, White, and Black residential fathers in the FFCWS. FB Hispanic fathers had significantly less use of physical and psychological aggression than all other groups, whereas the NB Hispanic, White, and Black fathers did not significantly differ from each other. Table 2 presents descriptive statistics for the sample of Hispanic fathers used in further analyses; significant differences between FB and NB fathers are highlighted. FB fathers endorsed traditional gender norms more highly, had higher rates of attendance at religious services, and were older as compared with NB Hispanic fathers. NB fathers had higher levels of education, were more involved with their children, and had fewer incidences of IPAV as compared with FB Hispanic fathers.

Path Model Results

Physical aggression

Model 1 examined within-time associations of psychosocial and child factors to paternal physical aggression at Wave 3 (Table 3). This model accounted for 23% of the variance in father aggression toward the child at age 3 years. FB fathers used significantly less aggression than NB fathers. There was a strong association between child’s higher level of aggressive behavior and paternal physical aggression, and young paternal age increased risk for physical aggression. Model 2 assessed the effects of predictors (Wave 3 or prior) on paternal aggression toward the child at Wave 4, and accounted for 19% of the variance in paternal aggression toward the child. FB fathers used significantly less aggression toward their 5-year-old children than did NB fathers. Although child aggressive behavior had a strong within-time association with paternal aggression, in the time-lagged model, the effect of child aggressive behavior at Wave 3 was no longer significant. However, paternal heavy alcohol use at Wave 3 was associated with greater paternal physical aggression at Wave 4.

Psychological aggression

Model 3 examined within-time associations of paternal psychological aggression, paternal psychosocial factors, and child factors (Table 3). Fathers’ FB status was negatively associated with psychological aggression, whereas child aggressive behavior, father-to-mother IPAV, and income were positively associated with psychological aggression. The time-lagged model (Model 4) accounted for 31% of the variance in psychological aggression toward the child. FB nativity status was negatively associated with psychological aggression. Factors associated with higher levels of paternal psychological aggression included greater endorsement of traditional gender norms, the child’s anxious or depressed behavior, and the child being a boy. As hypothesized, all four models showed FB status was a significant protective factor for paternal physical and psychological aggression.

Years in the United States Among FB Fathers and Spanish Language Use

The study sample of FB fathers had lived in the United States an average of 13.19 years (SD = 7.38). We examined the hypothesis that among FB fathers greater time spent in the United States would relate to more aggressive parenting. Using only the subsample of FB Hispanic fathers (n = 155), we estimated the same path models described above, substituting “years in United States” for nativity status. Among FB fathers, years in the United States was not significantly associated with physical aggression (B = 0.01, SE = 0.03, β = 0.08, p = .64) or psychological aggression (B = 0.03, SE = 0.05, β = 0.17, p = .58).

In additional analyses, Spanish language use was included in the model alongside nativity status. As noted previously, Spanish language use was strongly correlated with both nativity status and length of time in the United States, and thus introduced multi-collinearity into the models. When Spanish language use was used in the model instead of nativity status, the variable for language provided no additional explanatory value. We elected to use nativity status in our models as a more direct indicator of the construct of interest.

Mediational Analyses

An advantage of path modeling is that this approach allows us to simultaneously estimate direct and mediated relationships across acculturation factors, psychosocial risk factors, and paternal parenting aggression. Results of our meditational analyses showed that, in general, paternal psychosocial factors did not play a significant role in mediating effects of nativity status on paternal aggression. However, we found that fathers’ endorsement of traditional gender roles had an indirect inverse relationship with paternal physical aggression (total indirect effect: B = −0.09, SE = 0.04, β = 0.05, p < .05), and that this relationship was largely mediated by fathers’ alcohol use (indirect effect through alcohol use: B = −0.11, SE = 0.05, β = 0.06, p < .05). Thus, fathers’ who more strongly endorsed traditional gender norms were less likely to engage in heavy alcohol use and, in turn, were less likely to use physical aggression with their 5-year-old children.

Discussion

Following from the developmental ecological model (Belsky, 1993), this study examined the ways in which fathers’ psychosocial characteristics, aspects of the home environment, and child behavior problems were associated with parenting aggression among residential Hispanic fathers. In addition, our analyses examined whether parenting aggression was associated with variables that may be of importance to Hispanic parents, including nativity status, length of time in the United States, adherence to traditional gender norms, and religiosity. Our analyses across four path models examining within-time and time-lagged predictors found a strong and consistent pattern showing that FB nativity status was associated with lower levels of paternal physical and psychological aggression. FB fathers had been in the United States an average of 13.19 years. However, a separate analysis of FB fathers showed that length of time in the United States did not significantly explain variance in parenting aggression. Given the findings for nativity status, these results suggest that for FB Hispanic fathers in this study, the use of parenting aggression was related to norms and values that do not change quickly, and may be closely linked to norms and beliefs associated with their culture-of-origin or with values and behaviors that are distinct among immigrants. Further, these results suggest the lower levels of parental aggression among Hispanic fathers found in prior studies (Lee et al., 2008; Lee, Perron, et al., 2011) may be attributed, at least in part, to lower levels of aggressive parenting used by FB Hispanic fathers in particular. It is important to emphasize neither this study nor others have provided evidence that NB Hispanic families are at greater risk for maltreatment than White or Black families (Dettlaff et al., 2009; Dettlaff & Johnson, 2011; Drake et al., 2011; Zhai & Gao, 2009). Our examination of mean paternal aggression among FB Hispanic, NB Hispanic, White, and Black fathers shows that aggression among NB fathers does not differ significantly from aggression among White and Black fathers.

A strength of the current study is that our path models controlled for child behavior problems, including child aggression. This step is critical because the developmental ecological model (Belsky, 1993) holds that young children who are more aggressive may elicit more harsh punishment from their parents (Black et al., 2001; Patterson, 1982). Parental FB nativity status remained protective even after accounting for the strong within-time association of child aggression and parenting aggression. Although child aggression had a strong within-time association to paternal physical and psychological aggression, these effects were not found in the time-lagged models. In other words, child aggression, measured at age 3 years, was not significantly associated with fathers’ subsequent use of aggression. Consistent with the developmental ecological model, this suggests a reciprocal association exists between children’s and fathers’ aggression within-time. Future research might use longitudinal assessment of these measures to examine the influence of fathers’ aggressive parenting on the development of child aggression over time.

We assessed the potential association of multiple psychosocial risks to Hispanic fathers’ parenting aggression. In the multivariate models, father involvement, parenting stress, and depression did not play any discernable role in explaining variance in paternal aggression. Heavy alcohol use was positively associated with physical aggression in the time-lagged model only, whereas IPAV at Wave 3 increased risk of psychological aggression at Wave 4. Bivariate results indicated that FB fathers had significantly higher levels of traditional gender norms and religious attendance. Although religious attendance was not associated with paternal aggression in any of the models, greater endorsement of traditional gender norms had a direct association with greater psychological aggression at Wave 4. This finding is consistent with a cross-sectional study that linked traditional gender norms with lower sensitivity to emotional maltreatment among fathers (Ferrari, 2002). However, we also found that greater endorsement of traditional gender norms had an indirect association with lower physical aggression mediated by lower alcohol use. Despite higher levels of traditional gender norms among FB fathers in this study (Table 1), FB Hispanic fathers were less likely to use aggression toward their young children than NB Hispanic fathers.

Hispanic Paradox and Paternal Aggression

The term Hispanic paradox refers to the phenomenon, documented primarily in the health behavior literature, whereby Hispanics living in the United States have better or similar health to that of non-Hispanic Whites, even though Hispanics’ tend to have risk factors for poor health outcomes such as lower incomes and less education. Studies of the Hispanic paradox have shown that FB or less acculturated Hispanics have lower rates of substance use (Bates & Teitler, 2008; Detjen, Nieto, Trentham-Dietz, Fleming, & Chasan-Taber, 2007; Kimbro, 2009; Page, 2007; Vega, Alderete, et al., 1998); lower rates of psychiatric disorders (Alegria et al., 2008; Alegria et al., 2007; Escobar, et al., 2000); and lower rates of engaging in risky sexual behaviors (Page, 2007). Several hypotheses have been proposed to explain the Hispanic paradox, including that immigrants to the United States bring a legacy of healthier customs from their country-of-origin than the health behaviors common among U.S.-born populations of similar socioeconomic status.

In the current study, we found nativity status may provide protective benefits related to fathers’ aggressive parenting behaviors. Our findings are consistent with those of Altschul and Lee’s (2011) study that examined Hispanic mothers’ use of aggressive parenting behaviors. These results suggest that the Hispanic paradox may extend to aggressive parenting behaviors. One possibility is that Hispanic immigrants come from a cultural context in which parental aggression toward young children (5 years or younger) is less common than in the United States. Similar to other health behaviors, the immigrant parents adhere to those “healthier” parenting norms following immigration whereas the parenting behaviors of more acculturated Hispanic parents become similar to those of the broader American population. Although limited, some cross-cultural research on parenting aggression has pointed to potential cultural differences in parenting styles. Moreover, this research has suggested that the parenting styles (authoritarian, authoritative, permissive, and neglectful) often used to describe U.S. parents may not apply to Latin American parents (García & Gracia, 2009; Guilamo-Ramos et al., 2007; Martínez & García, 2008), who have been described as more “protective” (Rodríguez, Donovick, & Crowley, 2009). Another study found that “indulgent” parents who were characterized by high levels of parental warmth and low levels of strictness, had better outcomes among youth in countries such as Spain and Brazil (García & Gracia, 2009; Martínez & García, 2008). However, these ideas are speculative. It is difficult to determine whether cross-cultural differences truly exist in parent-to-child aggression or other parenting behaviors that increase risk for child maltreatment because so few studies have compared the behaviors or parenting norms of Hispanic parents in the United States to the parenting norms of Mexican, Central American, or South American parents.

The social selection hypothesis offers an alternative explanation for the findings in the current study as well as the Hispanic paradox more generally (Palloni & Morenoff, 2001). This hypothesis argues that given the hardships and demands of migration, immigrants may be a healthier subpopulation than the general populations in either their countries-of-origin or the United States (Franzini et al., 2001; Markides & Coreil, 1986). When applied to parenting behaviors, social selection also offers a possible explanation of the lower rates of aggressive parenting among immigrants or less acculturated Hispanic parents. It is possible that parents who choose to migrate do so in part to improve outcomes for their children (e.g., Lopez, 2001), and thus, immigrant parents are, on average, relatively more invested in their children’s well-being and success than are parents in either their country-of-origin or U.S.-born parents.

Recent theorizing has emphasized that neighborhood characteristics, such as higher levels of collective efficacy and lower levels of social disorganization in immigrant communities, may explain some of the health advantages documented among immigrants in prior research (e.g., Cagney, Browning, & Wallace, 2007; Kimbro, 2009). One study indicated that neighborhoods with higher concentrations of immigrants also had lower levels of parent-to-child physical aggression (Molnar, Buka, Brennan, Holton, & Earls, 2003), which suggested that both individual-level and community-level measures of immigrant status may contribute to risk for parental aggression.

It is also possible that immigrant parents’ behaviors change when they arrive in the United States. For example, immigrant parents might use less aggression for fear of attracting the attention of child welfare authorities. However, the lack of a significant association between length of time in the United States and aggressive parenting suggests that, at least once parents are in the United States, parenting behaviors are related to norms and values that are slow to change.

These rival explanations present important hypotheses for future research. Specifically, future research may wish to examine (a) social selection as an explanation for the observed differences in aggressive parenting between FB and NB Hispanic parents; (b) cross-cultural differences in parenting style, and how such differences may relate to use of aggression toward children; (c) the presence of neighborhood characteristics, such as higher levels of collective efficacy and lower levels of social disorganization in immigrant neighborhoods, as an explanation for differences in parenting practices between FB and NB Hispanic parents, and (d) whether immigrant parents change their behaviors in response to a perceived threat of child welfare involvement.

Based on our finding that fathers who endorsed traditional gender norms had lower levels of alcohol use and were less likely to use aggression toward their young children, we suggest that future research should examine how fathers’ traditional gender norms may relate to parenting risk behaviors, such as alcohol and drug use, and subsequent family violence. In particular, our measure of traditional gender norms consisted of only two items and thus had low internal consistency. Researchers may wish to use a more robust measure to further disentangle these relationships.

Study Limitations

The results of this study must be interpreted in light of the study limitations. This study used a relatively small sample size, which had limited statistical power to detect modest relationships and did not allow for analyses by country-of-origin. Some scales, such as paternal traditional gender norms and parenting stress, consisted of fewer items than would be ideal and had low internal consistency, which may have limited our ability to detect mediation. All families were living in urban areas (Reichman et al., 2001), and therefore, the results should not be generalized to nonresidential Hispanic fathers or Hispanic fathers living in rural areas.

Study participants represent a unique subsample of involved, biological fathers, who were residing with their 3-year-old child. Numerous studies using FFCWS data have indicated that nonresidential and residential fathers differ in important ways. For example, unmarried fathers were more likely than married fathers to be members of a minority group, younger at the time of the child’s birth, and have lower levels of education (Carlson & McLanahan, 2010). In addition, unmarried fathers attended fewer religious services and had higher levels of depression. These differences are also likely to influence use of aggressive parenting. However, because of limitations of the available data, we were unable to examine differences between residential and nonresidential fathers. Indeed, comparing residential and nonresidential fathers on the range of behaviors assessed by the CTSPC may be difficult to do with any data set because the measure is predicated on the father having contact with the child. The nature of the parenting relationship is fundamentally different for nonresidential fathers, who spend less time with the child and have fewer opportunities to exercise parental discipline when compared to residential fathers.

Another limitation of the FFCWS study design made it necessary for our study to rely on maternal reports of fathers’ aggressive parenting of the child. Although researchers have suggested that women’s reports of aggression in the home may be more valid than men’s reports (Edleson & Brygger, 1986), we were unable to locate studies that specifically addressed similar issues with the CTSPC. Moreover, we were not able to examine the concordance of fathers’ and mothers’ reporting on the CTSPC because those measures were not collected from fathers as part of the FFCWS. However, findings from a study that used a different data set to examine parental concordance of CTSPC reports indicated high and statistically significant correlations between mothers’ reports of fathers’ aggression and fathers’ self-reported aggression (Lee, Lansford, & Pettit, 2011). Mothers and fathers agreed more than 90% of time on the most severe items measuring physical aggression. In general, although the parents agreed on the fathers’ aggressive actions, fathers’ self-reports indicated that they committed each item of aggression more frequently than what was reported by the mothers; such a pattern would bias the results of the current study toward null findings. Although the finding that mothers’ CTSPC reports of fathers’ behaviors are valid on a number of dimensions does not eliminate the possibility of biases that exist with all self-report measures, the available research suggests that mothers accurately estimate fathers’ use of aggression and that mothers’ reports of fathers’ behaviors are consistent with their partners’ reports of their behavior.

Conclusions

Although immigrant Hispanic families face socioeconomic disadvantage that is typically associated with greater risk for child maltreatment, these families are no more likely to be involved with the child welfare system than non-Hispanic White families (Drake et al., 2011). In fact, several studies suggest that Hispanic children are underrepresented among child welfare-involved families (Dettlaff et al., 2009; Dettlaff & Johnson, 2011; Zhai & Gao, 2009). The under representation of Hispanics in the child welfare system may be due to, at least in part, FB Hispanic fathers’ relatively lower levels of aggressive parenting behaviors that place children at risk for child abuse. Such findings have been replicated and extended in a study of Hispanic mothers (Altschul & Lee, 2011). The effects for nativity status were robust to sociodemographic and psychosocial control variables, and, notably, children’s levels of aggression.

An implication of this study is that when considering issues related to the representation of Hispanic children in the child welfare system, socioeconomic and parenting behavior risks must be considered separately. In other words, socioeconomic risks, such as lower income and limited education, should not be conflated with parenting risks, such as parenting stress, alcohol use, and depression. These results call for greater specificity in child welfare screening and assessment tools so that sociodemographic and parenting risk factors are assessed independently and considered separately. When constructing child welfare policies and training procedures, officials must not only consider race/ethnicity and nativity status differences in socioeconomic factors, but also consider how race/ethnicity and nativity status relate to specific parenting behaviors, such as use of parental aggression, that place children at risk for child maltreatment (Drake et al., 2011). Future research should examine whether the Hispanic paradox extends to the realm of parenting behaviors, and whether adapting to U.S. norms is linked to increased parenting aggression among Hispanic immigrant parents.