Introduction

A number of studies have demonstrated that single parenting—especially among mothers with limited access to social and financial support—is associated with parent stress, stemming in part from the single-handed negotiation of heavy parenting responsibilities (Brooks-Gunn and Duncan 1997; Ceballo and McLoyd 2003; Lamb 1997; McLoyd 1990). In addition, research has consistently shown that parents who experience economic stress display less nurturance and more harshness in their responses to their children (Lempers et al. 1989; McLoyd and Wilson 1991); that poverty diminishes the quality of parenting due to persistent daily stressors (McLoyd 1990; Lamb 1997); and that emotionally stressed mothers are more likely than others to rely on aversive, coercive discipline techniques (Conger et al. 1984; Crnic and Greenberg 1987; Jackson 2000; Jackson et al. 1998; Maccoby and Martin 1983; Straus et al. 1997).

In 2006, nearly 80% of births to black women under the age of 30 were to single mothers. For black children, more than half (51%) are in families headed by single-parent mothers (American Community Survey 2007; U.S. Census Bureau 2006), and children in these families have extraordinarily high rates of poverty (Duncan and Brooks-Gunn 1997; Huston et al. 1994). Differences in discipline practices may account for some of the variation in child behavior in poor and single-parent black families. While the effects of discipline on mainly white samples have received a great deal of research, the literature on within-group differences in discipline practices among single black mothers raising young children in poverty is sparse (see, for example, Harrison et al. 1990; Jackson et al. 1998; Lansford et al. 2005).

In a comparison of black and white children, Deater-Deckard et al. (1996) found that physical discipline practices in the first 5 years of life were associated with higher levels of teacher-reported behavior problems for white children in kindergarten through third grade, but they found no significant association between physical discipline and subsequent teacher-reported behaviors for black children. Others have reported similar race moderation effects (Gunnoe and Mariner 1997). Some believe that one explanation for these effects may be that physical discipline is more normative for black families than for white families (Deater-Deckard and Dodge 1997). Studies have shown, nevertheless, that negative parenting and harsh discipline—often associated with economic hardship, parent stress, and limited access to helpful social support—undermines black children’s socioemotional development (McLoyd 1995). Studies also show that mothers who get help and support from their partners are more effective parents (Chase-Lansdale et al. 1994) and that black preschoolers whose nonresident fathers maintain contact with them have fewer behavior problems in kindergarten (Jackson et al. 2009).

To better understand how individual differences in poor black children’s family relationships might account for some of the difference in child behavior, we examined the presence and impact of nonresident biological fathers on preschoolers’ development and early school adjustment as buffers of the potentially negative consequences of their (single) mothers’ parent stress and harsh discipline practices. The presence and effects of nonresident biological fathers are particularly relevant for black families, given the high rates of single parenthood (King and Heard 1999; Tamis-LeMonda and Cabrera 1999). Using two waves of data from a study of current and former welfare recipients, their preschool children, and the children’s kindergarten teachers, we investigated whether the presence of a nonresident biological father might moderate the relations among mothers’ parenting stress, harsh discipline practices, and preschoolers’ subsequent teacher-reported behavior problems in kindergarten.

Background

Studies comparing children raised in single-parent families to those raised in families with two biological parents consistently find that those raised in two-parent families with biological parents do better on educational achievement and adjustment in school (Carlson and Corcoran 2001; Hetherington and Clingempeel 1992; McLanahan and Sandefur 1994; Pryor and Rodgers 2001). A part of this difference may be due to single mothers’ lower educational attainment, less social support, fewer economic resources, and more stressful environments (Carlson and Corcoran 2001; Dunn et al. 1998; McLoyd 1990; O’Connor et al. 2001). However, children in mostly white, middle-class stepfamilies also do less well on a range of outcomes than their counterparts in two-parent families with biological parents (Amato and Gilbreth 1999; McLanahan and Sandefur 1994). This suggests that biological fathers are important for children’s well-being and development.

Despite the foregoing, however, some have concluded that absent biological fathers may not be all that important in black single-mother families, given the presence of father figures (or “social fathers”) who may be viable substitutes or replacements for absent and uninvolved biological fathers (Furstenberg 1995; King and Cherlin 2002). Since many black children have never lived with their biological father in the conventional sense, others have assumed that the implications of never having had a father present in the home are probably different from the implications of having had a biological father present who left, as in most white stepfamilies (Mott 1990). Stated differently, there is some suggestion in the literature that the salutary effects of being raised by two biological parents that seem to apply to children in middle-class white families may not apply to children in mostly poor and near-poor single-parent black families.

The present study tested this notion. In doing so, we expected that more frequent contact between nonresident fathers and their 3-year-old children at time 1 in poor and near-poor single-mother black families would moderate the potentially negative effects of mothers’ parenting stress and frequent spanking (or negative parenting) on child behavior problems in kindergarten a year to a year and a half later (at time 2). This is important because research demonstrates that children who perform well as they begin their school careers tend to continue to do so, while children who have poor starts tend to continue to do poorly in school (Alexander and Entwisle 1988; Ladd and Price 1987).

There are several ways in which nonresident fathers can have an influence on children’s well-being. They can maintain contact with their children and pay child support. Studies have shown that many nonresident fathers have infrequent contact with their children and fail to pay child support (Furstenberg and Harris 1992; Hawkins and Eggebeen 1991). While the quality of fathers’ contacts with their children is important (Amato and Gilbreth 1999), a nonresident father who visits frequently is likely to establish a much different (and probably better) relationship with the child than one who visits infrequently or not at all (Seltzer and Bianchi 1988). In addition, the payment of child support often is crucial for the economic well-being of poor and near-poor single mothers and their children (King 1994).

Still, it should be acknowledged that some researchers have found no relationship between father–child contact and child well-being (Amato and Gilbreth 1999; Furstenberg et al. 1987), although others have found beneficial effects when parents get along well together (King 1994). Much of this research has focused on the effects of father absence (due to divorce) in middle-class white families (Lamb 1997). For poor and near-poor black children in single-mother households, there is evidence that the frequency of nonresident fathers’ contact with their child and the child’s mother is associated with beneficial developmental outcomes for preschool children (Jackson et al. 2009a, b) and that such contacts may be beneficial for the children not necessarily due to the fathers’ greater involvement in child-rearing but in the greater emotional support father–child contacts provide to mothers (Cowan et al. 2009).

The present study was informed theoretically by Bronfenbrenner’s (1988) person-process-context model. As such, attention was directed to the impact of personal characteristics of family members, family processes, and particular external environments on the focal children’s behavioral development. This perspective assumes that processes operating in different ecological contexts are interrelated (Bronfenbrenner 1986). These interrelationships—in particular, proximal processes that occur between parents and children in the home environment—are considered key mechanisms by which child developmental potential is realized (Bronfenbrenner 1986, 1988).

We focused on relations between the preschool child and parental figures, including nonresident fathers (at time 1) and events that might influence the child’s progress in kindergarten (at time 2). More explicitly, we tested the relations among mothers’ parenting stress and spanking frequency, nonresident fathers’ contacts with the child, and the mothers’ satisfaction with the father–child relationship (our measure of relationship quality)—all at time 1 when the focal children were 3 years old—on teacher-reported child behavior problems in kindergarten at time 2. We controlled for the mother’s educational attainment and the child’s gender (personal characteristics of family members), because studies have shown that lower maternal educational attainment and family processes differentially influence boys and girls prior to adolescence (Baydar et al. 1993; Duncan and Brooks-Gunn 1997; Jackson et al. 2006). We controlled, as well, for the family’s income using a ratio of income-to-needs (an economic resource influenced by external environmental conditions).

In focusing on within-group differences among single black mothers with a preschooler who were receiving or who had received welfare benefits, the present study differs from previous research in several ways. First, most of the father-absent research has focused on middle-class white stepfamilies with adolescent children or compares black families to white families (Amato and Gilbreth 1999; Gershoff 2002). Second, previous research has given insufficient consideration to whether and how nonresident black fathers’ presence in the lives of their children might offset possible negative effects of poverty on child developmental outcomes (see, for example, McLoyd 1990; 1995). Studies have found that social support—in the present study in the form of nonresident fathers’ relationships with their children—has a beneficial effect on children’s development in poor families (Colletta and Lee 1983; Crnic and Greenberg 1987; Jackson et al. 1998), particularly families headed by low-income single mothers with low educational attainment (see, for example, McLoyd 1990, 1995; Pianta and Ball 1993). Finally, as stated earlier, the present study is a test—although preliminary, given the relatively small sample size—of the notion that absent biological fathers may not be all that important in poor and near-poor black families.

In the discussion that follows, nonresident fathers are referred to as fathers and single black mothers are referred to as mothers.

Method

Procedure

One hundred poor and near-poor mothers with a preschool child were interviewed in their homes between February and June 2004 at time 1 of this study. The mothers resided in communities in Pittsburgh that were composed predominantly of low-income black families. Recruited through the Allegheny County Assistance Office (the main welfare agency in Pittsburgh), the sample consisted of 134 randomly selected mothers—current and former welfare recipients—with a 3-year-old child, who also were single, black, and 18 years of age or older. Letters of solicitation (sent out by the Allegheny County Assistance Office to preserve the anonymity of nonrespondents) described the study as an ongoing survey on raising young children and family life (see also Jackson et al. 2009). The time-2 interviews took place between October 2005 and January 2006. The final sample consisted of 100 mothers who returned a form indicating a willingness to participate in the study at time 1—representing a response rate of 75%—and 99 mothers at time 2 (one mother died in the interim between time 1 and time 2). For each interview, mothers completed a computer-administered questionnaire focusing on parenting and family life. They were paid $75 at time 1 and $150 at time 2. Weekly phone calls to the mothers in the interim between the two interviews resulted in a retention rate of 100%. In addition, 89 teachers (89.9% of those sent a mailed questionnaire) completed an assessment of the children’s adjustment in kindergarten at time 2. Teachers were paid $25 for their time. This investigation focuses on the 89 mothers and children for whom there were teacher assessments.

Measures

The measures that follow are multiple- and single-item scales. Items were reversed as necessary so that a higher score indicates more of the attribute named in the label. Alpha coefficients were obtained for scales with multiple items.

Parenting stress (seven-item scale, alpha = .65) was assessed by a six-point scale (ranging from 0 = not at all to 5 = completely) asking mothers to indicate how true statements such as the following were for them: “My child seems much harder to care for than most,” “There are some things my child does that really bother me a lot,” and “I find myself giving up more of my life to meet my child’s needs than I expected” (Parenting Stress Index-Short Form; Abidin 1990). The frequency of spanking was measured by mothers’ answers to the question, “About how many times, if any, have you had to spank your child in the past week?” The frequency of father contact was indicated by mothers’ answers—on a seven-point scale (ranging from 1 = almost every day to 7 = 0 times)—to the following question: “In the past 12 months, about how often has your focal child seen his/her father?” Satisfaction with the father’s presence in the child’s life (our measure of quality) was indicated by mothers’ answers to the question, “All things considered, how satisfied or dissatisfied have you been with the father–child relationship over the last 2 months.” Response options ranged from 1 = completely dissatisfied to 5 = completely satisfied.

Financialchild support was indicated by mothers’ responses of “yes” (coded 1) or “no” (coded 0) to a question asking whether the birth father paid child support. Income-to-needs was measured by mothers’ answers to a question about their household income (including job, welfare, food stamps, child support, and any other source) in the prior month. This amount was divided by the number of people living in the household. Educational attainment was indicated on a five-point scale (1 = grade school to 5 = BA/BS degree) that asked mothers to give the highest level of education they had completed. The child’s gender was coded 1 if girl and 0 if boy.

Child behavior problems (30-item scale, alpha = .82) were assessed by kindergarten teachers who indicated on a three-point scale (ranging from 1 = very much like the child to 3 = not at all like the child) the extent to which behaviors such as the following described the focal child’s behavior in school during the past 3 months: “is rather high strung, tense, and nervous,” “is disobedient at school,” “has trouble getting along with other children,” “is impulsive or acts without thinking” (Problem Behaviors Index; Peterson and Zill 1986).

Overview of Analyses

The principal statistical procedure was ordinary least squares (OLS) regression analyses. First, teacher-reported child behavior problems in kindergarten were regressed on mothers’ parenting stress and spanking frequency, contacts between nonresident fathers and the child, and mothers’ satisfaction with these contacts, together with the mothers’ educational attainment, the child’s gender, and two income variables (the family’s income-to-needs ratio and child support payments). To estimate the main effects of the variables in this core model, the following equation was generated:

$$ BP = \beta_{1} ME + \beta_{ \, 2} CG + \beta_{ \, 3} IN + \beta_{ \, 4} CS + \beta_{ \, 5} PS + \beta_{ \, 6} SP + \beta_{ \, 7} FC + \beta_{ \, 8} SF, $$

where BP is teacher-reported child behavior problems, ME is the effect of mother’s education, CG is the effect of child’s gender, IN is the effect of income-to-needs, CS is the effect of child support, PS is the effect of parenting stress, SP is the effect of spanking, FC is the effect of father contact, and SF is the effect of satisfaction with father. If variations in mother’s education, the child’s gender, the family’s income-to-needs ratio, and child support payments from the nonresident father at time 1 significantly predict teacher-reported child behavior problems at time 2, then the β 1, 2, 3, and 4 coefficients should be significantly larger than zero. Negative coefficients would indicate that children who were boys, whose mothers had lower educational attainment, whose families had lower income, and whose fathers paid no child support would have more behavior problems in kindergarten, whereas positive coefficients would indicate the reverse. The latter was considered to be unlikely.

If, as expected, parenting stress explains a consequential portion of the variance in teacher-reported behavior problems, then the β 5 coefficient should also differ significantly from zero. For example, a positive β 5 coefficient would indicate that the higher the mother’s parent stress, the more child behavior problems teachers would report. Likewise, a β 6 coefficient significantly greater than zero and positive would indicate that children who were spanked more frequently would have poorer behavioral outcomes in kindergarten, whereas a β 7 coefficient significantly greater than zero and negative would indicate that nonresident biological black fathers might play an important role in their children’s behavioral development. Finally, a β 8 coefficient significantly greater than zero and negative would indicate the importance of the quality of the father–child relationship as a predictor of child behavioral outcomes in this study.

Then, to determine whether the nonresident father’s presence was a buffering factor for children in the context of mothers’ frequent spanking and parenting stress, the following interaction terms were added—in separate models—to the core regression model to test the hypothesized moderation effects: β 9 SP × FC and β 9 PS × FC, where SP × FC is the interaction effect of spanking and father–child contact and PS × FC is that of parent stress and father–child contact. If, as expected, more frequent contact between nonresident fathers and their children moderates the negative effect of mothers’ frequent spanking and parent stress on teacher-reported behavior problems in kindergarten, then the respective β 9 coefficients should be significantly greater than zero and negative. In the analyses that follow, two-tailed tests were used throughout.

Results

Descriptive Analyses

At time 1, the mothers were, on average, 25 years old; at time 2 they were 27. The corresponding ages of the focal children were 3 and 5 years old at time 1 and time 2, respectively. Means, standard deviations, and correlations between variables are presented in Tables 1 and 2. All of the mothers had some education at the high school level or beyond; for example, 11% had completed 10 years, 24% had completed 12 years, 64% had completed up to 14 years, and 1% had completed 15 years of education. Fifty-four percent of the children were boys and 46% were girls. On average, the focal children saw their fathers about five times a month (M = 4.94, SD = 2.46). The mean income-to-needs for the sample was about $337 (SD = $150) a month. Sixty percent of the mothers received child support payments from the nonresident fathers.

Table 1 Means and standard deviations for the sample (n = 89)
Full size table
Table 2 Correlations between variables (n = 89)
Full size table

The correlations in Table 2 show that more educated mothers were better off financially with respect to income-to-needs (r = .21, p < .05) and somewhat more likely to get financial child support from the children’s fathers (r = .20, p < .10). Mothers with higher levels of parent stress spanked more frequently (r = .34, p < .01). More frequent spanking was related, in turn, to increased child behavior problems in kindergarten (r = .34, p < .01), as was greater parent stress (r = .23, p < .05). The association between father–child contact and the payment of child support was positive (r = .24, p < .05) and that between father–child contact and teacher-reported behavior problems was negative (r = −.24, p < .05).

Multiple Regression Analyses

Table 3 contains the results of multiple regression analyses of teacher-reported child behavior problems. Model 1 shows the simultaneous effects of the core time-1 variables on behavior problems at time 2. Inspection of the betas indicates that greater parenting stress (β 5 = .32, p < .05), less frequent father–child contact (β 7 = −.21, p < .05), mothers’ diminished satisfaction with the father’s relationship with the child (β 8 = −.22, p < .05), and more frequent spanking of the child (β 6 = .30, p < .05) were significant predictors of increased teacher-reported child behavior problems in kindergarten at time 2, after controlling for mothers’ education, child’s gender, and income-to-needs, accounting for 21% of the variance in behavior problems, F (8, 76) = 3.80, p < .001.

Table 3 Multiple regression of teacher-reported behavior problems at time 2 (n = 84)
Full size table

Models 2 and 3 present the interaction terms between the father-contact variable and mothers’ spanking and parenting stress. Model 2 reveals a significant interaction between father–child contact and mothers’ spanking frequency in the hypothesized direction (β 9 = −.46, p < .05), suggesting that more contact between fathers and their children at time 1 reduced (buffered) the effect of harsh discipline practices by mothers on subsequent child behavior problems at time 2. This step increased the explained variance in behavior problems to 24%, F (9, 75) = 3.97, p < .001. Figure 1 illustrates the pattern of this interaction: when contact with the father was low, more frequent maternal spanking resulted in elevated behavior problems; with average father–child contact, the negative effect of spanking on teacher-reported behavior problems was muted; and with high contact, frequent spanking was not associated with increased behavior problems. Model 3 reveals a nonsignificant interaction between father–child contact and mothers’ parent stress. It is worthy of note that this interaction was in the hypothesized direction, F (9, 75) = 3.67, p < .001.

Fig. 1
figure 1

Moderating effect of the relationship between father contact and maternal spanking at time 1 on teacher-reported behavior problems at time 2

Full size image

Discussion

These results are consistent with a “buffering” interpretation of social support. This view holds that social support is mainly beneficial to those who need assistance from others due to conditions and circumstances such as single parenting, low educational attainment, parent stress, and limited access to social and financial resources (Cohen and Wills 1985; Hashima and Amato 1994). We viewed contacts between nonresident fathers and their children as a supportive resource for the mothers in this study. This view assumes, moreover, that such contacts may be beneficial for the children in part because of the greater emotional support father–child contacts provide to mothers (see, for example, Cowan et al. 2009). We expected at the outset that more frequent contact between nonresident fathers and their preschoolers would moderate the potentially negative effects of mothers’ parent stress and frequent spanking on child behavior problems and early school adjustment as evaluated subsequently by kindergarten teachers. We found no buffering factor (for children) of fathers’ presence in the context of mothers’ parent stress, and we can offer no compelling explanation for the nonsignificance of this interaction. However, our findings do suggest a “protective effect” of nonresident fathers’ presence in the context of mothers’ spanking behavior that appears to operate at least in part through increases in father–child contact. Recall that when contact with the father was low, maternal spanking was associated with elevated levels of behavior problems; with average contact, this effect was reduced; and with high father–child contact, there was no association between mothers’ physical discipline practices and teacher-reported child behavior problems.

These results are contrary to the notion that absent biological fathers may not be all that important in black single-mother families (Furstenberg 1995; King and Cherlin 2002); they also are contrary to studies showing no relationship between father–child contact and child well-being (Amato and Gilbreth 1999). While the “quality” of the father–child relationship is certainly important, as numerous studies have reported (see, for example, Amato and Gilbreth 1999), it is reasonable to expect that the “amount” of contact between nonresident black fathers and their young children would also be important (see, for example, Cowan et al. 2009; Jackson et al. 2009a, b; Seltzer and Bianchi 1988), precisely because—different from children (with absent biological fathers) in middle-class, mostly white stepfamilies (Amato and Gilbreth 1999)—many poor and near-poor black children have never lived with their biological fathers in the conventional sense. As indicated earlier, there are several important ways in which nonresident fathers can influence their children’s well-being beneficially. They can maintain contact, pay child support, and provide emotional support to mothers (see, for example, Cowan et al. 2009; Hawkins and Eggebeen 1991; King 1994; Jackson et al. 2009a, b). In the present study, there is evidence that such influences accounted, at least in part, for some of the variance in the children’s behavior in kindergarten. In addition, the mothers’ greater satisfaction with the fathers’ relations with the focal children early on was predictive of fewer behavior problems subsequently. If the latter can be considered an indicator of the “quality” of the father–child relationship, our findings do indeed suggest that both frequency and quality of contacts are important, at least for poor and near-poor black children whose biological fathers are often nonresident. While this is a matter for future research with larger samples, if our findings are valid, then policies and programs that encourage young black men to stay involved with their children should be a high priority. Indeed, studies of mostly white families that report no relationship between father–child contact and child well-being and the suggestion that children in poor black families do not need their (nonresident) biological fathers because of the presence of father figures may, if heeded by policy makers, discourage initiatives and interventions aimed at encouraging supportive relations among nonresident black fathers, their young children, and the children’s mothers.

Our results suggests that single mothers’ parenting stress—stemming, perhaps, from the single-handed negotiation of heavy parenting responsibilities (Brooks-Gunn and Duncan 1997; Ceballo and McLoyd 2003; Lamb 1997; McLoyd 1990)—may have serious consequences for the well-being of poor and near-poor black children, especially if emotionally stressed mothers are more likely than others to rely on aversive, coercive discipline techniques (Conger et al. 1984; Crnic and Greenberg 1987; Jackson et al. 1998). If nonresident fathers’ contacts with their children serve a buffering or “protective” function in such circumstances, then evidence-based interventions might focus on honing relationship skills between these fathers and the mothers of their children (see, for example, Brooks-Gunn and Markman 2005, for a review of intervention studies focused on improving parenting). Future research might also investigate models reflecting mediation effects among the predictor and outcome variables investigated in the present study. Such models would shed greater light on the mechanisms involved in poor and near-poor black children’s early adjustments to school (see, for example, Jackson et al. 2009).

Finally, there are several limitations of these data that should be acknowledged (see, for example, Jackson et al. 2009a, b). First, the sample was relatively small, and the mothers were residing in Pittsburgh. Further research with additional (and larger) samples from other cities is needed to explore more fully these very important issues. Second, although the present study used two waves of data, causal inferences about the observed relations among the variables with respect to the child outcome examined would be inappropriate. Third, our measures, for the most part, relied on mothers’ reports. However, it is important that in our model, the children’s behaviors at time 2 were based on teacher ratings. We acknowledge that objective reports of our constructs would have removed the potential for shared error variance. Still, these data are unique in that they were collected from a group of single mothers who are of special interest for two principal reasons: (1) single black mothers are disproportionately represented among the very poor and the welfare-dependent and (2) little is known about individual differences in this population with respect parenting stress, physical discipline practices, social support from nonresident fathers, and child outcomes.