Introduction

Fathers are key socializers of gender in cultures both within and outside the United States, and there has been heightened interest in understanding father involvement in men from diverse backgrounds (Leaper and Friedman 2007). Nonetheless, the contextualized view of gender socialization in the United States has highlighted a number of limitations in the literature. The emphasis on fathers as “playmates” (e.g. Paquette 2004) has led to relative neglect of other everyday activities (e.g., caregiving) that may convey powerful messages about gender roles and behaviors. Moreover, most studies are limited to a single developmental period (e.g. Lindsey and Mize 2001), precluding assessment of changes to fathers’ gendered activities over time. Finally, there continues to be a paucity of knowledge about fathers’ role in children’s gender development in families from diverse ethnic/racial backgrounds, as highlighted in a special issue of this journal (Chuang and Tamis-LeMonda 2009).

In response, this research addresses three goals. The first is to document the types of daily activities that fathers engage in with their children and how those activities change over children’s first years of life. The second is to examine whether and how father-child activities differ by child gender. The third is to advance knowledge of gender socialization in ethnically and economically diverse U.S. populations. Our research is guided by an eco-cultural framework (e.g., Weisner 2002; Rogoff 1993), which: (1) views children and parents as co-participants in the daily routines of their families and communities, and (2) considers daily routines to be a principal context for parents’ transmission of messages about gender roles.

Although we focus on fathers in the United States, we document similarities and differences among men from different backgrounds, including recent immigrants. The U.S. has historically been characterized by ethnic and racial diversity, leading to varied views and practices among families from different cultural communities, including those concerning the fathering role (McFadden and Tamis-LeMonda in press). However, this work also has relevance beyond the U.S. given the changing demographics of countries throughout the world due to increased immigration. There is global need to better understand how cultural subgroups within a broader society navigate a variety of influences in socializing their young.

Gender Socialization within the United States: Focus on Daily Activities

The literature on parents’ role in children’s gender socialization in the U.S. is mixed. Here we review studies on this topic, although most such work is based on Caucasian, European American samples. A number of studies of European American families find that parents treat boys and girls differently: they decorate children’s rooms using gendered colors and materials (Rheingold and Cook 1975), dress their children in gender-typed clothing (Pomerleau et al. 1990), spend more time with children of their own gender (Harris and Morgan 1991; McHale et al. 1999), and are often children’s first models of gender (McHale et al. 2003). Through this constellation of behaviors, parents convey their attitudes about gender (Bussey and Bandura 1999), which in turn affect children’s gender knowledge and activities (Tenenbaum and Leaper 2002).

Others, however, find few differences in parents’ treatment of boys versus girls. An important review (Maccoby and Jacklin 1974) and a major meta-analysis (Lytton and Romney 1991) of studies based on largely European American samples both identified more similarities than differences in parents’ interactions with boys and girls. Parents of boys do not differ from those of girls in their emphasis on achievement, warmth and responsiveness, or encouragement of dependence or independence, as some examples (Lytton and Romney 1991). In fact, differences in parental behaviors with sons versus daughters were limited to the domain of play: parents support more masculine typed play for boys and more feminine typed play for girls.

One reason for the seeming lack of gender differences in parents’ direct interactions with children is due to a predominant research emphasis on how parents engage with their sons versus daughters (e.g., warmth, responsiveness) rather than what parents do with them (e.g., cooking, physical play). That is, parents may be equally “warm” towards their boys and girls, but do so in the context of different everyday activities. Eco-cultural theories, including those emphasizing “daily routines” (e.g., Weisner 2002) and “guided participation” (Rogoff 1993) provide frameworks for the idea that parents and children are social partners who engage in activities that both reflect and are shaped by cultural views and practices (including those regarding gender). These activities in turn shape children’s skills, preferences, and behaviors. As children observe and share in daily practices around eating, playing, bathing and sleeping (as examples), adults and other social partners convey important cultural messages (Rogoff 1993).

Accordingly, a first goal was to examine the types of daily activities that fathers engage in with their children. We focused on caregiving (e.g., feeding), physical play (e,g., chase games), literacy (e.g., book reading), and social visits (e.g., taking child to visit friends). Each of these activities serves unique and important functions. Caregiving is typically considered to be the purview of mothers, yet fathers are often highly involved in the care of their young children (Cabrera et al. 2000). During caregiving, parents must regulate both their own and their children’s emotional expressions (Dunn and Brown 1994). Physical play is especially important to inciting and regulating children’s arousal systems, and theoretical writings note that fathers frequently engage in this form of activity (Paquette 2004). Literacy activities are core to children’s language and literacy development, and provide many “teachable” moments (Raikes et al. 2006; Rodriguez et al. 2009). Although fathers engage less frequently in literacy activities than mothers, in one study of ethnically diverse fathers of predominantly White, Latino, and African American backgrounds, the frequency of father-child book reading related to children’s cognitive outcomes (Duursma et al. 2008). Finally, social activities are those that extend to networks beyond the family, providing a sense of social embeddedness and integration that is beneficial to well-being, buffers stressful events, and is associated with satisfaction with family functioning in European American families (Armstrong et al. 2005; Snowden et al. 1994).

However, little is known about fathers’ engagement in these daily activities with their children (see Yeung et al. 2001). One of the few studies on parenting and children’s everyday activities is reported by McHale and colleagues (1999). European American parents and their 9- to 10-year-old children, and younger sibling, all reported on their activities on a given day for 1 week. In families with both sons and daughters, in which fathers held more traditional attitudes, children were more likely to be involved in activities with their same gender parent. However, no information was provided on the specific activities that children shared with their parents in this study or a follow-up study (McHale et al. 2004).

Another line of research that indirectly addresses the topic of everyday activities is the literature on household division of labor. Compared to mothers, fathers of both European American and African American backgrounds spend less time with their children and the time they are with their children is predominantly spent in play (e.g., Hossain and Roopnarine 1994; Lamb 1997). Even European American fathers who take on the more non-traditional role of primary caregiver display more playfulness than mothers (Pruett and Litzenberger 1992). If fathers spend more time engaged in play and mothers spend more time engaged in caregiving, children are recipients of powerful messages about the roles of men and women in the family system (Smith Leavell and Tamis-LeMonda in press).

Fathers’ Activities and Child Gender

A second goal was to examine whether fathers’ activities differ by child gender. Gender socialization is a culturally embedded process. Children are influenced by multiple members of their families and social networks (e.g., McHale et al. 2003), and culture and ethnicity may dictate different roles within a family (Peplau et al. 1999). The role of the father is of particular interest, as gender role expectations and practices around child-rearing responsibilities may differ among different racial/ethnic groups (Sigel and McGillicuddy-De Lisi 2002). Fathers may communicate cultural-specific notions about gender behaviors and roles through the activities they engage in with their children.

However, despite recognition that mothers and fathers are core sources of information about gender (Leaper and Gleason 1996), the majority of studies on gender socialization in the U.S. focus on mothers’ attitudes and beliefs (see Lytton and Romney 1991). The limited research on fathers is problematic as fathers have been found to have stronger gender stereotyped attitudes and behaviors (e.g., Lamb 1977; Siegal 1987) and to be more likely to encourage gender-typed play than mothers (Leaper and Friedman 2007). Fathers may be particularly important to the gender development of boys. Boys may be more sensitive to, and adhere more strongly to their fathers’ gender attitudes than girls (Leaper 2002). In one study, European American fathers’ beliefs about their children’s abilities were more strongly associated with children’s sense of competence and values about sports than mothers’ beliefs (Fredricks and Eccles 2002).

There is some, albeit scarce, indication that fathers behave differently with sons than with daughters. For example, during physical science tasks European American fathers of sons use more explanations and scientific terms than fathers of daughters (Tenenbaum and Leaper 2003). Fathers also talk about different things with their sons than daughters. When discussing memories of past events, Latino fathers more frequently referenced action-based events (e.g., visits to the amusement park) with their sons but social events (e.g., going to a birthday party) with their daughters (Cristofaro and Tamis-LeMonda 2008). Fathers’ level of engagement in various activities also has been found to vary by child gender. Father-son dyads of European descent have been found to engage in more physical play (Lindsey and Mize 2001); and fathers from diverse ethnic backgrounds (White, Latino, African American) engage in less book reading and other forms of literacy than father-daughter dyads (Duursma et al. 2008). Together, these findings suggest that fathers provide unique and potentially very different messages and contexts for gender socialization.

Focus on Diverse Populations

A final goal was to contribute to the literature on gender socialization in ethnically and economically diverse populations within the United States. Most studies focus on White, middle-class populations, despite recognition that a parent’s ethnicity and/or social class provide important cultural contexts for gender development. Parents from different backgrounds, such as those of Mexican descent, have different views and practices around raising children, division of household labor (e.g. Pinto and Coltrane 2009), economic and educational resources, and gender norms that may come to be reflected in fathers’ daily activities with their boys and girls.

However, current knowledge about fathers from various racial/ethnic backgrounds does not provide clear predictions about their gendered behavior with their children. For example, Latinos in the United States may draw on roles and expectations from their own culture, while also learning and incorporating the values of fatherhood in America (Cabrera and Coll 2004). Latino families often endorse high masculinity in males (machismo) and self-sacrificing attitudes and family-oriented behaviors in females (marianismo or hembrismo) (Denner and Dunbar 2004), which are reflected in the division of household labor (Pinto and Coltrane 2009). For example, Mexican origin women in the United States do more housework compared to other ethnic groups and support unequal distributions of household labor (McLoyd et al. 2000). Additionally, Latina mothers living in the United States are more likely than fathers to encourage gender appropriate behavior with their daughters, whereas Latino fathers are more likely to encourage gender appropriate behaviors with their sons (Raffaelli and Ontai 2004).

However, traditional Latino values around “familismo” might also draw fathers into household participation and childcare. For example, Mexican immigrant fathers display higher prenatal involvement (e.g., obstetric visits) and subsequently greater participation in meal sharing and play with their toddlers compared to African American and Dominican fathers (Tamis-LeMonda et al. 2009). And, although both African American and Latino fathers appear to monitor their children’s activities more than White fathers, Latino fathers are found to interact more with their children than men of the other two groups (Toth and Xu 1999). Additionally, familismo might lead to greater reliance on extended family for emotional and social support in Latinos, particularly when compared to Whites (Mindel 1980), which might be seen in fathers’ participation in social visits with children. Both Mexican American and African American families have been found to make greater use of extended family support compared to Whites (Tienda and Angel 1982), challenging views of Latino fathers as detached and influenced by a machismo mentality (Cabrera and Coll 2004).

African American parents have been found to display relatively equitable childcare responsibilities, with mothers’ and fathers’ roles often overlapping (McAdoo 1988). This shared caregiving has been highlighted as a strength in African American families (Jarrett et al. 2002). When observed at home with their infants, African American fathers are found to be as involved, and sometimes more involved, than other ethnic groups in caregiving and social interactions, regardless of their socio-economic status (Roopnarine et al. 2005). Indeed, studies on African American fathers’ perception of fatherhood point to the importance of caregiving, over and above that of economic provider (Hamer 2001), and many African American men feel they should be there for their children physically and emotionally, as well as financially (Roopnarine 2004).

Finally, a family’s resources (i.e. education) and family structure (i.e. residency and marital status) may influence fathers’ activities with their children. Fathers with more education (even within low-income, ethnically diverse samples) provide more cognitive stimulation to their young child (Tamis-LeMonda et al. 2004), and marital status is associated with parental sensitivity in White, Black, and Latino families (Tamis-LeMonda et al. 2004). However, little is known about whether these measures relate to father’s engagement in daily activities. There is some suggestion that a family’s resources influence boys and girls in different ways. When families experience financial difficulties, sons are treated more favorably than daughters, at least in European American families (Eccles and Hoffman 1984). Similarly, when European American mothers experience a high level of work stress, demands for assistance with housework increases for girls but not boys (Crouter et al. 2001). Consequently, the daily activities seen in families with few resources may be especially gendered due to greater need for parents to include children in household responsibilities.

Current Study

We longitudinally examined fathers’ activities with their young children across a three-year period. Fathers were visited in their homes when children were 2 and 3 years, and preschool age, and asked about activities around caregiving (e.g., feeding child), physical play (e,g., run and chase games), literacy (e.g., book reading), and social visits (e.g., taking child to visit friends). These ages were chosen due to the important changes in gender development that occur over this period. Children exhibit a basic understanding of gender at around 24 months; by 3 years they develop an understanding that gender is a stable characteristic of self and others; and at around 4 to 5 years of age, they display a rigid adherence to and belief in appropriate gender behavior (Ruble et al. 2006). We hypothesized that:

  1. 1.

    Overall, fathers would engage more frequently in physical play compared to caregiving, literacy, and visiting activities (hypothesis 1a). The predominance of physical play was expected to be found at all ages (hypothesis 1b), aligning with a longstanding body of research indicating fathers’ role as playmates to children (see Paquette 2004).

  2. 2.

    Fathers of boys would engage in more physical play than fathers of girls (hypothesis 2a; see MacDonald and Parke 1986) whereas fathers of girls would engage in more literacy activities than fathers of boys (hypothesis 2b; see Duursma et al. 2008). These gendered differences in fathers’ activities were expected to emerge at the two later ages, being non-significant when children were 2 years of age and first becoming aware of their gender identity (hypothesis 2c; Martin and Ruble 2010; Ruble et al. 2006), but becoming significant at the later ages when children develop knowledge of gender stereotypes and are more rigid in their behaviors (Halim and Ruble 2010; Ruble et al. 2006).

  3. 3.

    Ethic/racial differences were expected to be seen in fathers’ activities with children. Due to differences seen in division of household labor, Latino immigrant men were expected to participate less in caregiving than African American and White men (hypothesis 3a; see McLoyd et al. 2000), whereas African American men were expected to participate in more caregiving than Latino and White men (hypothesis 3b; see Jarrett et al. 2002). Additionally, Latino and African American fathers were expected to engage in more visiting activities with their children, based on the literature which suggests that both groups make greater use of extend family for social support compared to White families (hypothesis 3c; Tienda and Angel 1982).

  4. 4.

    Latino men were expected to engage in more play with their children than the other two groups (hypothesis 4a; see Tamis-LeMonda et al. 2009), with highest levels of physical play being seen in Latino fathers of sons, in light of findings that Latino men endorse higher levels of traditional male gender roles compared to the other two groups (hypothesis 4b; Abreu et al. 2000).

  5. 5.

    Fathers’ resources (education and income) and family structure (marital status) were expected to predict higher levels of involvement in literacy activities (hypothesis 5a; see Tamis-LeMonda et al. 2004). Family structure (i.e. married or residency status) was expected to predict higher father involvement overall (5b; Tamis-LeMonda et al. 2004).

Method

Participants

Participants were fathers with children involved in the National Early Head Start Research and Evaluation Project. This project started in 1995 when 17 Early Head Start sites across the United States were selected to participate in a study of 3,001 families who applied for childcare or other public service programs for their newborns (Boller et al. 2006). At the 24-month assessment, 10 sites participated in gathering of data on fathers; mothers were asked to identify the biological father of the focus child, and father interviews and surveys were conducted when children were 2 years, 3 years and preschool age (around 50 months). Participating fathers were more likely to be resident, married to the focus child’s biological mother, and advantaged, compared to non-participants (Cabrera et al. 2004).

For purposes of the present study, biological fathers who self identified as White, African American or Latino and had participated in the interview at 2 years and at least one other time point (either 3 years or preschool age) were included (N = 426 fathers; 204 girls, 222 boys). The final sample was 50% White, 31% Latino, and 19% African American. The majority of the Latino group identified as first-generation Mexican Americans. At the time of the 24-month interview, the fathers ranged from 17 to 53 years of age (M = 28.55, SD = 6.62). Most fathers reported working full or part time at each of the three time periods, with monthly income averaging $1551.11 (SD = $992.74) at 2 years. The majority of children (58%) were first born.

Father residency and marital status differed by race/ethnicity, χ 2(2, N = 426) = 29.33, p < .001; and χ 2(2, N = 426) = 30.03, p < .001, respectively. African American fathers were less likely than White and Latino fathers to be married to or residing with the target child’s mother (p’s < .001). There was no significant difference in residency or marital status between White and Latino fathers. Fathers’ education differed among the three groups, χ 2(2, N = 426) = 100.39, p < .001; 59% of Latino fathers reported less than a high school diploma, compared to 10% of White fathers and 19% of African American fathers.

Procedures

Fathers and their children were seen in their homes when children were 2 years, 3 years, and preschool age. Fathers orally completed a survey with a trained researcher, which lasted approximately 1 h. Additionally, father-child dyads were videotaped participating in a variety of tasks. Focus here is on the father interview data. Fathers reported on their race/ethnicity, education, marital and residency status, as well as their children’s age and gender. Residency and marital status was coded as dichotomous, with fathers who were in residence with and/or married to the child’s mother at all time points coded as a resident father and/or married.

Father engagement in activities was assessed using a series of questions created for the National Early Head Start Research and Evaluation Project (Cabrera et al. 2004). Fathers were asked to rate how frequently they participated in 23 different activities with their children (e.g., sing songs, play chase, take child to visit relatives, give child a bath) on a four-point scale (rarely, a few time a month, a few times a week, everyday).

To reduce items for analyses and to create activity scales, an iterative process of factor analysis was conducted. An initial exploratory factor analysis using Principal Components Analysis on the 23 activity items resulted in 4 factors at each age. Items characterized by rare occurrence (take child to doctor, and stay home to care for sick child), items that double or triple loaded (play with toys for building things), and items that loaded on a separate scale (take child to a public place and go out to eat) were dropped, resulting in a final set of 18 items.

Based on the confirmatory analyses, four scales were created (Caregiving, Visiting, Literacy and Physical Play) as the primary dependent variables in all analyses at each age by averaging the items that loaded on each factor (see Table 1 for means by age and gender). The Caregiving scale consisted of six items: put child to bed, give child bath, help child get dressed, help with toilet or diapers, help brush teeth, and help prepare meals for child (α = .81–.90 for each group at 2 and 3 years, and α = .76–.81 for each group at preschool age). The Visiting Activities scale consisted of four items: have relatives visit, visit relatives, visit friends, and take child to play with other children (α = .70–.74 at 2 years, α = .68–.75 at 3 years, and α = .70–.80 at preschool age). The Physical Play scale consisted of four items: play outside, play chase, play ball games, and go for a walk (α = .65–.75 at 2 years, α = .73–.79 at 3 years, and α = .70–.83 at preschool age). Finally, the Literacy scale consisted of four items: sing songs, read stories, tell stories, and take child to religious services (α = .67–.70 at 2 years, α = .69–.80 at 3 years, and α = .58–.72 at preschool age). Although the literacy scale was lower for White fathers at preschool age (α = .58), it was retained due to higher reliabilities at other time points and in the other two groups at preschool age. When dropping “attending religious services”, the Cronbach’s alpha improved (overall α = .68 at 2 years, α = .75 at 3 years, and α = .67 at preschool age). However, because religious services are an important source of literacy for some cultural groups (see Ortiz 2004), the item was retained.

Table 1 Means and standard deviations for father engagement on activity scales by child age and gender
Full size table

Results

The goals of this study were to describe fathers’ participation in various activities with their young children by child age, gender, and father race/ethnicity (as well as interactions among these variables). An omnibus 2 (child gender) × 3(race/ethnicity) × 3 (age) × 4 (activity) Repeated Measures MANOVA was conducted to test all hypotheses, with activity type and child age serving as within subjects factors, and child gender and father race/ethnicity as between subjects factors. Significant main effects and interactions in the omnibus test were followed up by Repeated Measure ANOVAs for each activity type separately (see Table 2), and post-hoc t-tests using Tukey HSD correction (overall alpha .05). Post-hoc t-tests were collapsed across age, except where age was significant. Separate Repeated Measure ANCOVA’s explored the role of fathers’ education, marital status, and residency in father activities.

Table 2 Repeated measures ANOVA results for each activity type
Full size table

The full sample of 426 participants were represented in all analyses, with multiple imputation being used to estimate values for fathers missing data at one time point (SPSS Missing Values, Chicago, IL). Within the final sample, fathers who were missing data at one time point (N = 192) did not fit any clear pattern, and there were no significant differences among those missing from those not missing in terms of father race/ethnicity, child gender, or age. The imputation process relies on all available data for all individuals to estimate missing values until the differences in predicted values across iterations are negligible. There was enough information within the data set to accurately estimate missing data, making this form of imputation preferred to listwise deletion and mean substitution (Schafer and Graham 2002; Rubin 1987).

Activity and Age (Hypotheses 1a & 1b)

The hypothesis (1a) that fathers would engage more frequently in physical play activities relative to caregiving, literacy, and visiting activities was partially supported by a main effect for activity in the omnibus Repeated Measures MANOVA, F(3, 1260) = 174.06, p < .001, partial η2 = .29. Specifically, fathers engaged in more frequent physical play compared to literacy activities and visiting, but caregiving was also high (see Table 1). Contrary to expectations that physical play would predominate at all ages (hypothesis 1b), a significant interaction between activity type and age, F(6, 2520) = 65.77, p < .001, partial η2 = .14, indicated changes to activities across child age. Specifically, caregiving was the most prevalent father activity at 2 years; by 3 years, caregiving and physical play were equal in prevalence; and by preschool age, physical play predominated (see Table 1). A significant main effect for age, F(2, 840) = 65.53, p < .001, partial η2 = .12, revealed that fathers engagement in daily activities with their child declined over time.

Activity and Gender (Hypotheses 2a, 2b, & 2c)

The hypothesis that fathers’ activities would differ by child gender, was supported. The omnibus Repeated Measures MANOVA revealed an interaction between activity type and child gender, F(3, 1260) = 9.57, p < .001, partial η2 = .02, pointing to different activity patterns in fathers of girls versus fathers of boys. Specifically, fathers of sons participated more frequently in physical play compared to fathers of daughters, t(424) = 2.10, p < .05, as expected (hypothesis 2a). In contrast, fathers of daughters participated in literacy activities more frequently than fathers of sons, t(424) = 3.11, p < .01 (hypothesis 2b). The hypothesis that gender differences in fathers’ engagement would increase with age (hypothesis 2c) was not supported (i.e., the lack of a 3-way interaction between child age, gender, and activity type indicated that the gendered pattern of father activities was consistent across all ages).

Activity and Race/Ethnicity (Hypotheses 3a, 3b & 3c)

Fathers from the three race/ethnicity groups differed in the activities they engaged in with their children, as revealed in a significant interaction between race/ethnicity and activity type in the omnibus MANOVA, F(6, 1260) = 13.08, p < .001, partial η2 = .06 (see Table 3). The interaction was explained by differences in caregiving, visiting, and physical play (see Table 2). Latino fathers participated less frequently in caregiving compared to African American, t(211) = 2.13, p < .05, and White fathers, t(344) = 2.39, p < .05, supporting hypothesis 3a. African American fathers did not engage in more caregiving as had been hypothesized (3b), although this finding was qualified by a race × gender interaction described below.

Table 3 Means and standard deviations for frequency of father participation in four activity types by father race/ethnicity and child gender
Full size table

In support of hypothesis 3c, Latino & African American fathers engaged more frequently in visiting activities with their children compared to White fathers, t(344) = 4.33, p < .001 and t(344) = 3.72, p < .001 respectively. Moreover this race/ethnicity difference was driven by fathers of sons, with Latino and African American fathers of sons engaging in significantly higher level of visiting activities compared to White fathers of sons, t(151) = 4.81, p < .001 and t(178) = 3.71, p < .001, respectively. This pattern was not seen among fathers of daughters.

Race/Ethnicity and Gender Interactions (Hypotheses 4a & 4b)

The omnibus Repeated Measures MANOVA revealed a significant interaction between father race/ethnicity and child gender, F(2, 420) = 3.94, p < .05, partial η2 = .02. However counter to hypothesis 4a, that gender differences in activities would be greatest in Latino fathers, the interaction was primarily driven by a difference between African American fathers of sons versus daughters, with the exception of literacy. Post-hoc analyses indicated that African American fathers of sons reported higher levels of engagement in caregiving, t(78) = 2.13, p < .05, visiting, t(78) = 2.42, p < .05, and physical play, t(78) = 2.91, p < .01, compared to African American fathers of daughters (see Table 3). This pattern was not seen in White or Latino fathers. Furthermore, race/ethnicity differences emerged among fathers of sons only in the post hoc tests. Counter to hypothesis 4b, Latino fathers did not engage in more physical play with sons than African American fathers, although both groups did report higher levels of involvement in physical play compared to White fathers of sons, t(291) = 4.41, p < .001 and t(291) = 3.48, p < .001, respectively. In addition, African American fathers of sons reported spending significantly more time in caregiving compared to Latino fathers of sons, t(109) = 2.41, p < .05, and White fathers of sons, t(151) = 2.16, p < .05. Finally, the relatively high involvement of African American fathers of sons was also reflected in an omnibus main effect for father race/ethnicity, F(2, 420) =2.99, p ≤ .05. Overall, African American fathers reported higher levels of engagement with their children (M = 2.64, SD = .05) than White fathers (M = 2.51, SD = .03).

Father Education and Marital Status (Hypotheses 5a & 5b)

Fathers’ education was related to fathers’ participation in caregiving and literacy activities, as indicated by two separate 2 (Education level: Less than a high school diploma, high school diploma) by 3 (Age) Repeated Measures ANCOVAs with engagement in caregiving or literacy serving as the between-subjects variable. At all ages, fathers with less than a high school diploma participated less frequently in caregiving and literacy activities compared to fathers with a high school diploma (all p’s < .05), confirming our hypothesis (5a) that education would be associated with greater father engagement. In particular, the significant race/ethnicity effect for caregiving reported above, which indicated lower rates of caregiving in Latino fathers, attenuated to non-significance after controlling for father education, F(2, 419) = .44, p = .64. Thus, Latino fathers’ lower levels of participation in caregiving activities was explained by their lower levels of education compared to White and African American fathers.

Additionally, fathers’ marital status was associated with participation in literacy activities, in partial support of the hypothesis (5b) that marital status would be related to higher father engagement in all activities. Fathers who were continuously married to their child’s mother participated more frequently in literacy activities compared to fathers who were not continuously married to the mother at all three time points (all p’s < .001). However, marital status did not change any results by child gender or father race/ethnicity.

Discussion

In line with eco-cultural and guided participation frameworks (Rogoff 1993; Weisner 2002), we find that fathers’ everyday activities with their young children present a powerful context for gender socialization. Specifically, fathers’ participation in everyday activities, such as caregiving, visiting, literacy, and physical play across children’s preschool years, differ by child gender, father ethnicity, and/or the interaction between the two. This work moves beyond studies of gender socialization that principally focus on the behaviors, attitudes, and beliefs of White middle-class mothers at a single point in children’s development.

Differences by Child Gender

Fathers of girls and fathers of boys differed in the activities they engaged in with their children; moreover, these differences began to emerge when children were 2 years and were sustained through preschool. Fathers of sons engaged more frequently in physical play activities (e.g., ball play), which aligns with the predominant characterization of European American fathers as playmates and boys as play partners (Paquette 2004). How might engagement in physical play among fathers of boys influence boys’ developing understanding of gender? By the time children are 3 years old there is a well-documented gender difference in children’s activity levels, with boys consistently rated as higher in activity level compared to girls (Campbell and Eaton 1999). In addition, ethnographic studies of children’s behavior on playgrounds reveal that boys occupy large open spaces, while girls predominate in smaller fixed spaces closer to the school building (Thorne 1993). High rates of engagement in physical play by fathers’ of boys may reflect early socialization process that encourage “typical” male behavior.

Additionally, at all time points, fathers of daughters engaged more frequently in literacy activities than fathers of sons. The lower rates of literacy activities in fathers of sons may have implications for children’s perceptions of these activities. Others find that Latino parents engage in longer and more elaborate narratives with their daughters (Cristofaro and Tamis-LeMonda 2008); the current findings suggest that overall engagement in literacy practices might be a gendered activity for parents, specifically fathers. The conceptualization of reading/telling stories as gendered activities extends the boundaries of gender differences in parenting described in meta-analyses of largely European American families (e.g., Lytton and Romney 1991).

The early and consistent presence of gender differences in fathers’ daily activities with their young children was unexpected. We hypothesized that before children were highly gendered in their own understanding and beliefs (e.g., 3 years), fathers’ participation in routine activities would be gender neutral. Instead, fathers’ conscious or unconscious notions of gender appropriate behavior might influence the activities they engage in with their children before their children have a well-developed understanding of the stereotyped nature of those activities. In this way, fathers may channel their children toward gender-typed activities from an early age.

Children themselves might affect the activities that parents engage in with them for reasons that might be less directly linked to children’s understanding of gender stereotypes. For instance, girls might prefer more sedentary activities and show an interest in books whereas boys favor more active forms of engagement (Thorne 1993), which would then elicit different forms of activities in fathers. Additionally, girls tend to develop language skills earlier than boys, which may influence fathers’ participation in literacy activities (Huttenlocher et al. 1991). However, because these differences were found when children were as young as 2 years, it suggests that fathers’ beliefs and expectations may also play a role.

Differences by Father Race/Ethnicity

We posited differences in fathers’ activities by race/ethnicity, only some of which were supported. First, we expected Latino fathers to engage in the lowest levels of caregiving, and African American fathers the most. These hypotheses were based on gender divisions in household labor found in some studies of Latino families (Inclán and Herron 1998) and studies which find relatively equitable childcare responsibilities among African American parents (McAdoo 1988). Because a majority of the fathers in the Latino sample were first generation Mexican immigrants, we expected that their attitudes about gender roles might align with the Mexican culture more than that of the United States. As a result, they might favor traditional gender roles around household labor that would extend to caregiving (Valentine and Mosley 1998). However, although Latino fathers engaged in caregiving less often than White and African American fathers as hypothesized, this “ethnic” difference attenuated to non-significance when fathers’ education level was included in models. Latino men may become more egalitarian in childcare duties with increased education, or, Latino fathers with lower levels of education may work more hours at low-paying jobs, which prevents them from being as involved in caregiving activities as men working fewer hours. Further, there is evidence that while Latino men report more traditional gender attitudes, in practice they engage more frequently than their White counterparts in caregiving activities such as child care, cooking and cleaning (McLoyd et al. 2000). Patriarchal roles among Latino households in the United States may be shifting (Smith 2005), and more equitable division of household labor can be found in Latino households where woman earn more of the total household income (Pinto and Coltrane 2009). Additionally, more recent work has urged a re-conceptualization of machismo, which acknowledges a Latino man’s sense of family respect and responsibility which includes an involved role as a father (Tamis-LeMonda et al. 2009).

Counter to our hypothesis, African American fathers did not engage in more caregiving overall compared to the other two groups, although child gender did moderate this finding as discussed below. As hypothesized, both Latino and African American fathers engaged in higher levels of visiting activities compared to White fathers. White fathers actually participated least frequently in visiting and physical play activities with their children, and this pattern was particularly pronounced among White fathers of sons. White fathers’ lower participation in these activities might, to some extent, reflect cultural differences in values. Both African Americans and Latinos strongly identify with close family and neighborhood connections (Hovey and King 1996; Wilson 1986), and therefore visiting activities with their children might constitute an important aspect of their daily routines.

Interactions between Father Race/Ethnicity and Child Gender

Perhaps the most interesting findings involved the interactions between father race/ethnicity and child gender. African American fathers of sons reported higher engagement in caregiving, visiting, and physical play than African American fathers of daughters, a pattern not seen in the other two groups. Additionally, African American fathers of sons engaged in higher levels of caregiving compared to White and Latino fathers.

What might explain the relatively high involvement of African American fathers of sons in this sample? One possibility is selection bias: because fewer African American fathers agreed to study participation compared to White and Latino fathers, the African American fathers in this study might have been more involved, or at least report being more involved. However, this interpretation does not account for why African American fathers of sons report significantly higher levels of engagement compared to African American fathers of daughters.

Another possibility is that African American fathers of sons were highly involved with their boys as a form of resistance to U.S. stereotypes of absent Black fathers and troubled Black boys (e.g., Black et al. 1999). Their high involvement might be the outcome of efforts to protect their young male children against such stereotypes. Additionally, there may be potential differences among African American fathers in their attitudes and beliefs about certain forms of gender appropriate behavior. Despite holding egalitarian beliefs about gender roles related to the division of labor, African American mothers of boys have been found to be the least tolerant of their boys engaging in other-gender behaviors (Zosuls et al. 2009). Perhaps the development of a strong male identity is an important cultural goal that African American fathers hold for their sons and they are thus more involved in daily activities with them to encourage this identity. In addition, the high levels of engagement in physical play by both African American and Latino fathers compared to White fathers may be explained by different values surrounding the physical strength/prowess of boys. Valued characteristics of masculinity and athleticism may be encouraged at a young age through physical play (Hill 2005).

Resources and Family Structure

As hypothesized, fathers who were married to the mothers of their children across all three time points and fathers who had higher levels of education were more likely to engage in literacy activities with both their sons and their daughters. However, despite mean level differences in literacy activities by marital status and education, gender differences in literacy activities maintained after co-varying these variables.

Counter to expectations, fathers’ residency was unrelated to amount or types of activities fathers engaged in with their children. Selection bias may be one explanation for the lack of a difference by father residency. Non-resident fathers who participated in multiple waves of data collection were men who were very involved with their children. Moreover, the quality of the mother-father relationship may be more central to father involvement than residency per se (Cabrera et al. 2008). Alternatively, non-resident fathers might have interpreted questions about their activities with their children as relative to the days spent with them (e.g., reporting to “always” play with their children, when with them).

Limitations and Conclusions

This study has several limitations. Already noted is the strong selection bias that characterizes studies of father involvement. Fathers who agreed to participate and did so for at least two assessments are those who are likely to be more involved with their children and to have a positive relationship to the mothers of their children. As such, little can be said about the gendered ways that less involved fathers engage with their young children. Furthermore, the results are based on father self-report. Fathers may not accurately report on their involvement in activities with their children, although there is a strength to asking men about their involvement directly, rather than relying on mothers’ reports. Certainly, fathers are privy to what they do with their children, especially if some of the day is spent with children when mothers are not around. An investigation of mother’s involvement in these same activities would further an understanding of gender socialization in the broader family context.

Second, the measures of father involvement were limited to the activity domains assessed and items included in surveys. In terms of scale development the literacy activity scale revealed a low alpha for White fathers at preschool age. However, due to the reliability of this scale at all other time points and in all other groups we decided to retain the scale as well as the associated item “attending religious services”; participation in religious services is an important source of literacy for some cultural groups (see Ortiz 2004). In this regard, this work points to the challenges of scale development across different populations, as researchers must determine the extent to which item selection should be based on group-general or group-specific patterns.

Finally, many unmeasured variables co-vary with father race/ethnicity, and might shed further light on the differences found. It would be simplistic to conclude that differences in fathers’ activities were “caused” by race/ethnicity. As one example, the lower levels of education of Latino fathers accounted for their lower involvement in caregiving activities. Moreover, there is great heterogeneity within each of the racial/ethnic groups tested. For example, prior studies document differences in the activities of Mexican and Dominican fathers (Tamis-LeMonda et al. 2009) highlighting limitations of pan-ethnic classifications, and the need for further examination of father involvement in diverse U.S. groups.

In conclusion, the findings of this study provide evidence that counters the claim that parents do not differ in the ways they treat boys and girls beyond encouragement of gender typed play (Lytton and Romney 1991; Maccoby and Jacklin 1974). Fathers differed in their levels of involvement with boys and girls in all four activities, particularly when father race/ethnicity was considered. Additionally, fathers provide a unique gendered context for children’s developing understanding of gender that may be quite different from that provided by mothers. The lack of inclusion of fathers, let alone ethnically diverse fathers, in much of the gender socialization literature to this point has led to an incomplete picture of how parents influence children’s early gender development within a broader cultural context. By exploring the everyday routines of fathering within the context of race/ethnicity, our findings reveal the nuanced ways that men socialize gender in their very young children.