Low-Income Nonresident Father Involvement with their Toddlers: Variation by Fathers' Race and Ethnicity

Abstract

Using data from a racially and ethnically diverse sample of low-income mothers of two-year-old children participating in the Early Head Start Research and Evaluation Project (N = 883), fathers' education and employment, mother-father relationship, and mothers' relationships with kin in the household were examined to explain variation in nonresident father involvement across racial and ethnic groups. Nonresident White fathers were less involved with their children than African-American and Latino fathers. This difference was explained by the status of mother-father relationships. White nonresident fathers were less likely than minority nonresident fathers to maintain romantic relationships with their child's mother. Mothers in the White father group were also more likely to re-partner, which negatively related to biological fathers' involvement. These findings suggest that approaches to strengthen nonresident father involvement in children's lives need to consider how father ethnicity and mother-father relations are associated with differential patterns of father involvement.

Studies examining variation in father involvement by race and ethnicity have focused either on divorced fathers of older children or on resident fathers and find that nonresident minority fathers pay less child support, visit less, and are less engaged with their children than nonresident White fathers (Hofferth, 2003; King, Harris, & Heard, 2004). Fewer studies have focused on variations in nonresident father involvement among low-income, never-married men with young children, a group that has dramatically increased (Martin, Hamilton, & Sutton, 2005). A notable exception is a study of teen mothers that found minority fathers to be more involved with their noncustodial children than White fathers (Danzinger & Radin, 1990).

Resource theory stipulates that parents with greater resources (e.g., education and income) will invest more money and time in their children than those with fewer resources (Haveman & Wolfe, 1994). Thus, differential levels of resources among never married, low-income, nonresident fathers may contribute to variations in father involvement. In our society, fathers are expected to provide for their children; fathers with higher levels of education and stable employment are able to provide for their noncustodial children and consequently may be more involved with them than fathers who cannot fulfill this role (McLanahan, 2004). Because White families have, on average, greater economic resources, White children are more likely to receive child support, and may also see their fathers more often, than minority children whose fathers may be unemployed and have lower levels of education, even within a low-income sample (Huang, Mincy, & Garfinkel, 2005). But when fathers' financial contribution includes informal types of support, such as gifts or extra cash, minority nonresident fathers may contribute at levels close or equal to their White counterparts (Hofferth, 2003), suggesting that minority fathers may also visit children as frequently as White nonresident fathers.

We also draw from family systems theory that views relationships within a subsystem (e.g., marital) as sources of stress or support that can impact functioning within other subsystems (e.g., parental; Cox & Paley, 1997). Because the nature of family subsystems may vary by race and ethnicity, the quality of the relationship between nonresident fathers and mothers could differentiate levels of father involvement among racial and ethnic groups. Nonresident fathers visit their children more often when parents have positive relationships (Coley & Chase-Lansdale, 1999) and less often when relationships are hostile or unstable (Cutrona, Hessling, Bacon, & Russell, 1998). Minority unwed nonresident fathers are more likely to maintain romantic relationships with the child's mother than White fathers (Carlson, McLanahan, & Brooks-Gunn, 2005), which is linked to father involvement (Cabrera et al., 2004).

It is also possible that mother's relationships with a new romantic partner and extended family influence nonresident biological father involvement. Divorced fathers' visitation and child support payments are lower when mothers remarry than when mothers remain single (Hofferth, Pleck, Stueve, Bianchi, & Sayer, 2002). Overall, White mothers are more likely to remarry or cohabit with a new partner than minority mothers (Page & Stevens, 2005). Conversely, minority mothers live in intergenerational households more often than White mothers; grandmothers may see fathers as unreliable and prevent them from visiting (Sigle-Rushton & McLanahan, 2004).

These findings support the hypothesis that nonresident fathers' resources (i.e., education and employment), mother-father relationship status, and mothers' household structure will predict father involvement and explain differences by race and ethnicity. Using data from the Early Head Start Research and Evaluation Project (EHS) we investigated how nonresident father involvement varies by race and ethnicity and whether these differences are explained by fathers' resources and parents' relationships with one another and others in the mothers' household.

Method

Participants

Data were drawn from the EHS project (see Cabrera et al., 2004 for full description). From the main sample of 2,166 primary caregivers interviewed at 24 months, we restricted the sample to biological mothers (N = 2,092) who reported that the focus child's biological father was nonresident at 24 months (N = 1,024). Nonresidency was defined as the biological father not living with his child at all. The sample was restricted further to include mothers who reported mother-father relationship status (e.g., whether they were romantically involved or not) (N = 1,016), fathers who were White, African American or Latino (N = 995), and those with complete mother-report data on father involvement (N = 883). Participating families with nonresident fathers were more likely to be older, educated and employed than nonparticipating families with nonresident fathers (for complete selection bias of EHS study see Cabrera et al., 2004).

Measures

Fathers' race and ethnicity

Fathers' race and ethnicity were obtained from mother report at recruitment or father report during the 24- or 36-month father interview. Missing data (24% of respondents had neither mother nor father report of father race and ethnicity) were imputed using the Expectation-Maximization (EM) algorithm (McLachlan & Krishnan, 1997).

Mother and child characteristics

During recruitment, data on mothers' age at the child's birth, education (1 = high school degree/GED or higher), nativity status (1 = born in the U.S.), child gender (1 = male), and first born status (1 = first born) were collected from mothers. At 24 months, mothers reported on family income (natural log of family income was used in analysis).

Fathers' age and resources

Mothers were asked about fathers' age at the time of the child's birth, level of education (1= high school degree/GED or higher), and employment status (1 = working or in school) during the 24-month interview. Missing data values on fathers' age (8%), education (16%), and employment/schooling status (27%) were imputed using the EM algorithm (McLachlan & Krishnan, 1997). Income data are unavailable for nonresident fathers.

Household structure

Mothers' household was coded as Intergenerational (HH) if mother lived with her parents, grandparents, and/or aunts/uncles (1 = HH) and Coresident romantic if mother lived with a romantic partner (not the biological father; 1 = coresident romantic partner).

Parent relationship

Mothers' relationship status with nonresident biological fathers (e.g., romantic partners, friends, in no relationship, or separated/divorced/other) was recoded into two dummy variables: Boyfriend (1 = romantic partner; 0 = all other relationships) and Friend (1 = friends; 0 = all other); being in “no relationship/divorced/other” is the reference group. Parent relationship quality, measured as the amount of disagreement about the child between mothers and nonresident fathers, was based on 6 items adopted from the National Survey of Families and Households (NSFH) and adapted for use in the EHS (e.g., How much disagreement do you have about: How CHILD is raised? How father spends money on CHILD?) (Sweet & Bumpass, 1996). Items were rated from 0 (none) to 2 (a great deal) and averaged (α = .73).

Father involvement

Mothers were asked to assess 3 aspects of father involvement: accessibility, engagement and responsibility (Lamb, Pleck, Charnov, & Levine, 1987). One question about accessibility (e.g., how frequently fathers contacted their child during the past three months) was rated from 1 (never) to 5 (everyday or more). Four questions about engagement (i.e., how often fathers played with, ate, read to, and took a walk with the child in the past week) were rated from 1 (never) to 6 (several times a week) and averaged (M = 2.68; SD = 1.82; α = .95). Three questions about fathers' financial responsibility (i.e., how often fathers gave the child clothing, toys, or presents; paid for medicines or doctors; and, gave the mother extra money) were rated from 1 (never) to 3 (often) and averaged (M = 1.58; SD = .59; α = .74). Because the 8 indicators of father involvement were highly correlated (r = .68 to r = .86), they were standardized and averaged into a father involvement composite scale with high internal consistency (α = .93) and a normal distribution (skew = .41; kurtosis = 1.80).

Results

Descriptive Analyses

Table 1 presents descriptive statistics for all variables. White fathers were more likely to have completed high school than either African American or Latino fathers and were more likely to be working or in school than African American fathers. Mothers in the White father group were less likely to live with extended family and more likely to coreside with a romantic partner than mothers in both minority father groups. Minority mothers were more likely to describe nonresident fathers as their boyfriends (both groups) or friends (African Americans only) than White mothers. White fathers scored lower on the standardized measure of father involvement than fathers in both minority groups, although the difference between White and Latino fathers did not reach statistical significance in pairwise tests.

Table 1. Descriptive Statistics for Nonresident Fathers by Race and Ethnicity.

Variables	Total	White	African Amer.	Latino	F/χ2
	N = 883	n = 249 (55%)	n = 482 (28%)	n = 152 (17%)
Demographics/Father Resources
Mother > HS Ed	62%	74%a	63%b	43%c	40.00***
Mother's Age at Birth	21.5(5.8)	22.2a	20.7b	22.7a	9.63***
Mother HH Income	$12,299	14,296a	11,232b	12,410c	7.40***
Mother Born in U.S.	92%	98%a	98%a	63%b	206.04***
Child is Male	52%	49%a	52%a	56%a	1.85
Child is Firstborn	68%	64%a	71%a	66%a	3.37
Father's Age at Birth	24.2(6.2)	24.6ab	23.6b	25.1a	4.35*
Father > HS Ed	39%	48%a	39%b	26%c	18.69***
Father Work/School	77%	82%a	73%b	81%ab	10.46**
Household Structure
Intergenerational HH	25%	19%a	29%b	25%ab	8.28*
Resident Romantic Partner	11%	21%a	8%b	7%b	30.80***
Parental Relationship
Father is Boyfriend	15%	8%a	19%b	17%b	16.23***
Father is Friend	31%	22%a	38%b	26%a	21.44***
Parental Conflict	.51(.48)	.58a	.50a	.45a	2.75
Father Involvement (std scores)	.00(.82)	-.17a	.10b	-.02ab	9.19***

N =883 for Father Involvement; N = 662 for Parental Conflict.

^*p < .05;

^**p < .01;

^***p < .001.

Different letters indicate significant pairwise group differences at p < .05 in either bonferroni post-hoc adjustments or pairwise chi-square tests.

Multivariate Analyses

To test whether parents' resources and relationships accounted for the variation in father involvement by race and ethnicity, we conducted a series of Ordinary Least Squares regression models (see Table 2). In Step 1 we entered two indicator variables for African American and Latino father, with White father as the reference category, and controlled for family characteristics. In Step 2, we added fathers' resources (i.e., education, employment). In Step 3, we entered mothers' coresidence with extended family and romantic partner. In Step 4, we entered parents' relationship status. Parent conflict was not entered into models because it did not vary by race and ethnicity (see Table 1). Because there was little guidance from previous empirical findings we analyzed our data entering resource variables before relationship variables. However, results did not change when the order of entry was reversed.

Table 2. OLS Regression Models Predicting Nonresident Father Involvement (Accessibility, Engagement and Responsibility Scales Averaged).

	Step 1		Step 2		Step 3		Step 4
	B	SE	B	SE	B	SE	B	SE
African American	.27	.07***	.28	.07***	.21	.06**	-.01	.05
Latino	.28	.09**	.28	.09**	.21	.09*	.06	.08
White (reference)
Mother > HS Ed	.07	.06	.07	.06	.07	.06	.08	.05
Mother HH Income (ln)	.03	.03	.02	.03	.05	.03+	.07	.03**
Mother's Age at Birth	-.01	.00*	-.01	.00	-.02	.01*	-.01	.01
Mother Born U.S.	.26	.12*	.28	.12*	.29	.11*	.13	.09
Child is Male	.06	.05	.07	.06	.03	.05	.07	.04+
Child is Firstborn	-.06	.06	-.07	.06	-.06	.06	-.07	.05
Father's Age at Birth			-.00	.00	-.00	.01	-.00	.01
Father > HS Ed			-.03	.06	.02	.06	.03	.05
Father Work/School			.17	.07*	.15	.06*	.10	.05*
Intergen. HH					-.09	.07	-.12	.05*
Resident Partner					-.73	.09***	-.41	.07***
Father is Boyfriend							1.29	.06***
Father is Friend							.69	.05***
Constant	-.46	.36	-.54	.36	-.53	.35	-.99	.29**
Model F	4.17***		3.64***		8.66***		41.50***
Model R2	.04		.04		.12		.42
R2 Change			.00		.08***		.30***

N = 883.

⁺p < .10;

^*p < .05;

^**p < .01;

^***p < .001.

Step 1 reveals that the difference between White and African American fathers' involvement remained significant, and the difference between White and Latino fathers' increased, after controlling for family characteristics. Although White fathers were significantly more likely to have graduated from high school than minority fathers, and White and Latino fathers were more likely to be working at 24 months than African American fathers, the small R-square change for Step 2 indicates that fathers' resources explained little of the variation in involvement among the groups (although employment was positively associated with involvement, p < .05).

When mothers' coresidence with extended family and romantic partners were entered in Step 3, the coefficients for minority fathers were reduced by approximately 25%, significant reductions based on Wald post-hoc tests (at p < .01). Only presence of a resident romantic partner was negatively associated with father involvement (at p < .001), suggesting that higher rates of resident romantic partnerships among White mothers accounted for some of the variation in father involvement between White fathers and the other two groups, whereas higher rates of intergenerational living among minority mothers explained little of the difference.

When parents' relationship status variables (i.e., boyfriend and friend) were entered in Step 4, the coefficients for minority fathers were reduced to zero according to Wald post hoc tests (at p < .01). Although being the mothers' boyfriend or friend were positively associated with involvement, being the mothers' boyfriend was the strongest predictor, suggesting that minority fathers' higher likelihood of being romantically involved with the mother explained a large proportion of the variation in involvement between White and minority fathers.

Discussion

The present study yielded new findings on nonresident father involvement. Specifically, fathers' level of resources did not explain the variation in father involvement among White, African American and Latino fathers. Even after controlling for parents' education, age, and child gender, White fathers had lower levels of father involvement than minority fathers. This finding is noteworthy because White fathers were more likely to have resources such as a high school diploma and be employed than minority fathers in our sample, which resource theory suggests should make them more involved than fathers with fewer resources.

The difference in father involvement was explained, however, by unique patterns of relationships and household structure within the groups. Mothers in the White father group were more likely to report new romantic partners and less likely to report the biological father as her boyfriend or friend than mothers with minority partners. These results are consistent with past findings that low-income fathers are more likely to maintain high levels of engagement in children's lives when they are in a romantic relationship with child's mother (King et al., 2004) and less likely to be involved when mothers re-partner (Hofferth et al., 2002). In contrast to previous findings, we found that parent conflict (measured as parental disagreement about the child) did not vary by race and ethnicity. It is possible that we may have found different results if we had measured other aspects of the partner relationship. Overall, our findings support family system theory more strongly than resource theory when using these specific variables to explain differences in nonresident father involvement by race and ethnicity.

Our results advance the extant literature by extending these findings to unwed, nonresident fathers of toddlers and demonstrating that patterns of relationships between parents and household structures that are specific to each racial and ethnic group may account for differences in patterns of father involvement across racial and ethnic groups. Because minority nonresident fathers of toddlers are more likely to remain romantic partners or friends with their child's mother and minority mothers are less likely live with other romantic partners, fathers are more likely to be in regular contact with and potentially be engaged in positive nurturing activities (e.g., eating and playing) with their biological children than White fathers. The implications of these findings for child development need to be explored.

A strength of the present study is the use of a national dataset on low-income families. However, the study is limited by its select sample, which is poorer and younger than national norms, and under-represents rural families. Nonetheless, because unwed and non-coresident parents tend to be less advantaged than other parents (McLanahan, 2004), our findings highlight trends within the population most effected by rising rates of single parenthood. Another limitation is that because of the large number of missing data on fathers' reports of their own involvement we rely instead on maternal report. Mothers may under-report involvement or, if visitation occurs outside their home, may simply not know how often fathers engage in certain activities with children. However, even if mothers' reports are biased, we have no reason to suspect that mothers in one race or ethnic group would be more biased than another.

Our findings point to the need for more studies on racial and ethnic differences in father involvement over time, paying particular attention to the quality of relationships with partners and extended kin. It is possible that many of the romantic and friend relationships among parents will dissolve later in the children's lives, diminishing father involvement. On the other hand, it is also possible that the longer fathers are involved in their children's lives the more likely it is that they will feel invested and hence remain involved even when they are no longer in a relationship with their child's mother.