Family of origin processes and attitudes of expectant fathers.

This study examined the association between family of origin experiences of expectant fathers and their attitudes about father involvement. Using structural equation modeling and multiple regression analysis with a sample of 152 couples, we tested an ecosystemic model of fathering and examined the relative strength of the modeling hypothesis and the compensation hypothesis for linking these constructs. We found that expectant fathers who were either very close to their parents or very distant from their parents during childhood had more positive attitudes about father involvement. In addition, expectant fathers who believed that their own fathers were competent in their paternal roles had stronger attitudes about fatherhood. The findings also showed that expectant fathers who believed that their parents disagreed about child rearing and discipline rules while they were growing up had more positive attitudes about fatherhood. Several current family factors were also shown to be positively associated with attitudes about fatherhood.

Key Words: fatherhood, family of origin, parenthood

**********

For decades, fatherhood scholars have hypothesized that family of origin processes are associated with future father involvement (Doherty, Kouneski, & Erickson, 1998; Lamb, Pleck, Charnov, & Levine, 1985; Pleck, 1997). However, relatively little research has been conducted to test this hypothesis, and the studies in the literature have focused on men's relationships with their own fathers, ignoring the influence of mothers and of the co-parenting relationship in the family of origin (see Floyd & Morman, 2000). The purpose of this study is to formulate and evaluate an eco-systemic model of how family of origin processes are associated with attitudes about father involvement prior to the birth of a man's first child.

THEORETICAL MODELS AND PRIOR RESEARCH

According to Doherty et al.'s (1998) eco-systemic model of father involvement, five interrelated factors determine responsible fathering: the co-parenting relationship, mother factors, father factors, contextual factors, and child factors. These factors interact with one another to determine how fathers will be involved with their children. Among the individual father and mother factors are family of origin experiences. Intergenerational processes from the past interact with current relational factors to determine father involvement. However, as Doherty et al. (1998) note, little research has examined how these processes work. This study undertakes this investigation.

There have been two prominent conceptual models for understanding intergenerational influences on current family relationships: intergenerational family systems theory and attachment theory. Intergenerational family systems models are based on the premise that family values, assumptions, and beliefs are transmitted from one generation to the next (Boszormenyi-Nagy & Ulrich, 1981; Bowen, 1978; Framo, 1981; Williamson, 1981). According to Bowen (1978), a healthy adult is a person who is able to self-differentiate from the negative influences of their family of origin. Presumably, problems can occur in current marital and parent-child relationships when adults are rigidly tied to relationship patterns learned in their families of origin that may be a result of emotional fusion with their parents (Bowen, 1978). There is mixed support for Bowen's hypothesis that a self-differentiated adult has better functioning adult relationships (Day, St. Clair, & Marshall, 1997; Prest, Benson, & Protinsky, 1998). Moreover, Bowen's assumption that fathers bring different relationship patterns from their families of origin about marital relationships and co-parenting practices needs to be studied. For example, which co-parenting practices do expectant fathers bring from their families of origin about how mothers and fathers are to be involved with children?

Attachment theory posits that nurturing and close parent-child relationships early in life are associated with positive functioning across the lifespan (Ainsworth, 1973; Belsky, 1984; Bowlby, 1969; Sroufe & Fleeson, 1986). Consistent with attachment theory, researchers have found that attitudes about father involvement develop through both positive and negative family of origin experiences (Belsky, 1984; Floyd & Morman, 2000; Snarey, 1993). Studies have demonstrated a positive association between family of origin processes and father-child attachment (see Belsky, 1999; Van Ijzendoorn, 1995). For example, Snarey (1993) in a 35-year-longitudinal study showed how warmth experienced by men during childhood from their parents was positively associated with displays of warmth toward their offspring. Three other retrospective studies have shown how mothers' and fathers' secure attachments during childhood were associated with the secure attachment of their own infants (Cohn, Cowan, Cowan, & Pearson, 1992; Cowan & Cowan, 1992; Steele, Steele, & Fongay, 1996). These findings support a modeling hypothesis, the notion that parents who experience positive and close relationships with their own parents growing up demonstrate high levels of affection to their children.

On the other hand, unaffectionate relationships between fathers and sons in childhood can lead to sons being more involved with their own children (Floyd & Morman, 2000; Pruett, 1987; Radin, 1988; Snarey, 1993). Floyd and Morman (2000) with a sample of 506 men and their adolescent sons found the men who were most affectionate with their sons had fathers who were either highly affectionate or highly unaffectionate--a kind of compensation effect. In qualitative studies, the majority of men report that their own fathers were not positive role models, and that they desire to be more positive role models for their own children (Daly, 1993; Ehrensaft, 1987). Floyd and Morman (2000) argue that father involvement is based on a combination of the modeling and compensation effects.

   The modeling hypothesis predicts that men who experienced high
   levels of involvement from their fathers in their family of origin
   are significantly more likely to become highly involved fathers
   with their own children. The compensation hypothesis argues that
   men who are dissatisfied with the fathering they received will feel
   compelled to remake the fathering experience into something
   more positive for their own sons, thus compensating for a perceived
   lack of caring, nurturing, or involvement from their own
   fathers. (p. 349)

Further investigation is needed to examine what specific types of father's family of origin processes have a linear relationship and which a curvilinear relationship with father involvement.

The majority of studies on family of origin influences have focused on mother-child attachment rather than father-child attachment. Studies that examine father-child attachment have primarily investigated father-son relationships, rather than including father-daughter and mother-son relationships. In addition to limited scope, there is a potentially important confounding factor in most of the studies. Bowlby (1988) argued that the connection between early attachment experiences and adult attachment experiences may be altered. For example, a secure attachment with a spouse or therapist in adulthood may allow an individual to reconstruct his original insecure attachment from childhood (Bowlby, 1988). An implication for research on fathering is that studies of the impact of family of origin factors on current fathering should include controls for current family relationship factors, such as marital satisfaction, that might be confounded with recollections of family of origin experiences.

Current family factors are not only important to examine as potential confounds, but as Doherty et al. (1998) have stressed, they are also important on a conceptual ground. Otherwise, the field does not move past competing unitary explanations. Mothers' current attitudes about father involvement have been found to be an important predictor of father involvement (Cowan & Cowan, 1992; Maurer, Pleck, & Rane, 2001; McBride & Rane, 1997; Pasley, Futris, & Skinner, 2002). Marital satisfaction research has also shown that both positive and negative marital relationships are related to father involvement. In general fathers are more involved with their children when they have positive marital relationships (Blair, Wenk, Hardesty, 1994; Cummings & O'Reilly; 1997, McBride & Mills, 1993). However, some studies have found that fathers are more involved with their children when they experience poor marital satisfaction (Crouter, Perry-Jenkins, Huston, & McHale, 1987; Russell, 1986). Pleck (1997) hypotheses that poor marital satisfaction is associated with increased father involvement when the marital outcome measures focus on conflict and disagreements, whereas positive marital satisfaction is associated with increased father involvement when the marital measures are global measures of marital satisfaction. Pleck believes that if both parents are highly involved, there may be more of a possibility for differences in child-rearing styles, and therefore more marital disagreements. Future marital satisfaction research needs to look closer at what exactly is being measured, for example marital conflict, marital support, or co-parenting issues. Assuming there are differences in findings for marital support or co-parenting issues, future family of origin research also needs to examine both marital and co-parenting relationships.

This study tests a model of how family of origin factors influence father attitudes prior to the transition to parenthood. We assume that fathers' attitudes prior to the birth of a first child will influence how they are later involved with their child; this assumption will be tested in future research that follows the fathers in this study. Attitudes are conceptualized as expectant mothers' and fathers' feelings and beliefs prior to the transition to parenthood about how fathers are to be involved with their children. Specifically, this study examines the relationship between expectant fathers' perceptions of three types of family of origin relationships and their current attitudes about father involvement. The family of origin factors are paternal competence of the expectant father's father, overall family closeness, and the parents' co-parenting relationship. Current family factors include expectant mothers' attitudes about father involvement, expectant fathers' current views about their marriages, and expectant parents' current relationships with their parents.

In addition to examining the utility of the eco-systemic model, we focus on two competing hypotheses: the modeling hypothesis and the compensation hypothesis. As mentioned before, the modeling hypothesis predicts that fathers who come from families that were more nurturing will have stronger attitudes about fatherhood. The compensation hypothesis predicts that fathers who come from families that were more distant will have stronger attitudes about fatherhood. We will explore which types of family of origin processes support each hypothesis.

In this study, the term "expectant parents" refers to the expectant fathers who are expecting their first child. The term "parents" refers to the expectant fathers' parents. Family of origin relationships are measured retrospectively. While retrospective reports are crucial to understanding the current perceptions adults have about their families of origin, there is some controversy over the reliability of adults' family of origin retrospective reports (Conger, Cui, Bryant, & Elder, 2000). Studies have indicated that adults' reports about their families of origin are fairly stable; however, some family of origin processes can be reconstructed as a result of current marital satisfaction and the transition to parenthood (Lewis & Owen, 1995). Therefore, as mentioned before, attitudes about father involvement and their association with family of origin processes need to be studied prior to birth, while controlling for current family factors.

METHODS

The current study used the first wave of data from the Parenting Together Project, a three-year longitudinal intervention study that consists of couples expecting the birth of their first child. The second and third waves of data are being collected following the birth of the child, at six months and at 12 months postpartum. The longitudinal intervention study includes both a control group and an experimental group. The experimental group received eight parenting classes, four prior to birth and four after the birth. Both the control and the experimental group completed three sets of assessments, during the second trimester, 6 months after birth, and 12 months after birth. The first wave of data was collected from 162 couples in their homes. The sample for this study included 152 couples that completed all the questions on the measures. The Time 1 assessments were given to all couples prior to their knowing if they were in the control or the experimental groups. The couples were asked to fill out the assessments without discussing their answers. Research assistants remained in the room in case there were any questions. The first wave of data was used to study expectant father attitudes and family of origin processes prior to birth, as these variables may change after the transition to parenthood.

PARTICIPANTS

The couples were recruited from an HMO and from television and radio advertisements. In order to be included in the intervention study, the couples needed to be married or living together, over the age of 18, and expecting the first child for both participants. The mean age was 32.6 for the expectant fathers and 30.7 for the expectant mothers. Ninety-four percent of the couples were married. Regarding race, 81% were White couples, 16% were interracial couples (primarily Hispanic and White, and African American and White), 2% were African American couples, and 1% were Asian American couples. The majority of the expectant parents had a college degree (73%). The combined personal income for the majority of couples ranged from $60,000 to $75,000. Most expectant mothers (69%) and fathers (71%) came from families where they lived with both of their biological parents for at least 18 years. Fifty-seven percent of the couples in this sample came from families where they both lived with their biological parents for at least 18 years.

DATA ANALYSIS PLAN

We conducted the data analyses in two steps. The first step used structural equation modeling (SEM) to examine the linear association between expectant fathers' family of origin processes and attitudes about father involvement, controlling for current family factors. The second step used regression analyses to test whether or not expectant fathers' family of origin processes have a linear and/or a curvilinear association with attitudes about father involvement, while controlling for current family factors. Curvilinear relationships can be studied using SEM; however, the sample size did not allow for curvilinear factors to be added to the theoretical model.

THEORETICAL MODEL

The theoretical model hypothesized how expectant fathers' family of origin processes are associated with attitudes about father involvement (see Figure 1). The family of origin factors for expectant fathers were (1) father competence, (2) family closeness, and (3) the co-parent relationship. The model also included three current family factors: (1) expectant fathers' views of their current marital relationships, (2) expectant fathers' and expectant mothers' current relationships with their parents, and (3) expectant mothers' current attitudes about father involvement. This model hypothesized that family of origin and current family factors are directly associated with expectant fathers' attitudes of father involvement.

[FIGURE 1 OMITTED]

In the original structural equation model, expectant mothers' family of origin processes were also examined in relation to attitudes about father involvement; however, no sufficient model was computed that fit the data, suggesting that mothers' family of origin processes were not associated with attitudes about father involvement with the data from this sample.

MEASURES OF FAMILY OF ORIGIN

Items from the Family of Origin Questionnaire (FOQ) were used in this study to measure two of the family of origin factors: father competence and family closeness. The FOQ was adapted by Lewis and Owen (1995) from Shereshefsky and Yarrow (1973) to study the stability of family of origin recollections over time. The scale consists of 20 items: seven items related to father's parenting, seven items related to mother's parenting, and six items related to parents' marriage. The FOQ was used by Lewis and Owen to measure the stability and change of family of origin recollections for 40 couples expecting their first child. Over a four-and-a-half-year period, the mean correlations for the 20 items were .63 for women and .68 for men.

Father competence. Father competence was assessed using one item from the FOQ that asked expectant fathers, "How anxious do you think your father was about fulfilling his role as a father?" Response options ranged from 1 (extremely anxious) to 5 (little or no anxiety). This item explores if expectant fathers perceived their own fathers as anxious in their paternal role.

Family closeness. Family closeness was assessed using three indicators. The mother-son (alpha = .89) and father-son (alpha = .93) closeness indicators each consisted of the same five items from the FOQ that asked expectant fathers how emotionally close they were to each of their parents. Response options ranged from 1 (distant) to 5 (very close). The mother-son and father-son closeness items focused on how expectant fathers felt each of their parents met their emotional needs during childhood. The marital-closeness (alpha = .88) indicator was based on five items from the FOQ that asked expectant fathers about their parents' marital relationship. For example, "How did they handle marital conflict?" and "How did they show support to one another?" Response options ranged from 1 (very well) to 5 (not very well). The marital-closeness items concentrated on the marital relationship as opposed to the co-parenting relationship.

Co-parent relationship. The Parental Disagreement on Expectations of the Child Scale (PDECS) was used to measure the co-parent relationship. The PDECS developed by Scheck (1979) is a 12-item scale that asks participants to reflect on the co-parenting of their mother and father prior to the age of 12. Split-half reliabilities were reported as .78 for fathers and .72 for mothers. The items from the PDECS measure consensus between parents about child-rearing issues. The co-parent relationship indicator (alpha = .94) consisted of all 12 items from the PDECS that asked expectant fathers about their parents' co-parenting relationship. For example, "My mother was almost never able to agree with my father on what I should be punished for." Response options ranged from 1 (very untrue) to 5 (very true).

MEASURES OF CURRENT FAMILY FACTORS

Father marital relationship. The Dyadic Adjustment Scale (DAS) was used to measure a father's view of his current marital relationship. The DAS (Spanier, 1976) is a 32-item scale that assesses four marital satisfaction constructs: satisfaction, cohesion, consensus, and affectual expression. For example, to what extent do expectant mothers and fathers agree or disagree on items like affection, finances, time spent together? This widely used instrument has consistently shown high levels of reliability and validity. All 32 items were used in this study, yielding an alpha of .90. Response options ranged from 1 (always disagree) to 5 (always agree).

Current parent relationship. The Social Support Network Inventory (SSNI) was used to measure expectant parents' current relationships with their parents. The 20item scale developed by Flaherty and colleagues (1983) is a widely used measure that assesses how expectant parents are currently feeling supported by significant others in their lives, like their parents. This scale has demonstrated high reliability and validity (see Flaherty, Gavira, & Pathak, 1983). The current parent indicator (alpha = .85) was based on 16 items from the SSNI that asked expectant fathers and expectant mothers about current support from their parents. Research indicates that both mothers' and fathers' current relationships with their parents can influence parental involvement (see Pleck, 1997). For example, "To what extent does your father give you emotional support by listening, talking, consoling or just being with you?" Response options ranged from 1 (very little) to 7 (very much).

Expectant mothers' attitudes. Expectant mothers' attitudes about father involvement were measured from items from the Father Attitudes Scale (FAS; Pleck, 1997) and the Caregiving and Breadwinning Reflected Appraisal Inventory (CBIRAI; Maurer, Pleck, & Rane, 2001). The FAS was adapted from Palkowitz's (1984) Role of the Father Questionnaire. The 13-item scale asks participants about their attitudes about a father's role. For example, "A father should be as heavily involved in the care of his child as the mother." The alpha reliability for the FAS scale was reported as .77 (Pleck, 1997). Response options ranged from 1 (disagree a lot) to 5 (agree a lot). For the purpose of this study, the mother-father role indicator (alpha = .71) consisted of 10 of the 13 items from the FAS. Only 10 of the 13 items were used from the FAS as three of the items did not demonstrate sufficient reliability.

Maurer, Pleck, and Rane (2001) studied parental identity for both mothers and fathers with the Caregiving and Breadwinning Identity Reflected Appraisal Inventory (CBIRAI). Maurer and colleagues asked 64 married couples to fill out the two scales in three different ways, (a) for themselves, (b) how their spouse should be, and (c) what they perceived their spouse said about them. For example, "I should not be very involved in the day-to-day matters of physically caring for my child." Response options ranged from 1 (strongly disagree) to 5 (strongly agree). Only the caregiving identity scale was used in the Parenting Together Project. Caregiving identity refers to the degree that fathers see certain parenting domains as important to being a good father. The alpha reliability for this scale was .75. The mother-father identity indicator (alpha = .70) consisted of seven items from the caregiving identity scale that studies father identity. The seven items were taken from the questions about how expectant mothers thought expectant fathers should be involved with their children.

Expectant fathers' attitudes. The expectant father's attitudes about father involvement factor was based on the same items as the expectant mother's factor. The father role indicator (alpha = .80) consisted of 10 items from the FAS. The father identity indicator (alpha = .60) consisted of seven items from the caregiving identity scale.

RESULTS

PRELIMINARY ANALYSES

The first step was to test the assumptions of the SEM method as it was applied to this study. Multivariate skewness and kurtosis scores were in the normal range. Curve estimations were computed to test for the presence of linear and/or curvilinear relationships among the family of origin indicators (father competence, mother-son closeness, father-son closeness, marital closeness, co-parent relationship) and expectant fathers' attitudes about the father role and father identity. The results confirmed that there were both linear and curvilinear relationships between some of the expectant fathers' family of origin indicators and attitudes about father involvement. Specifically, the analyses produced positive and significant curvilinear relationships between (a) mother-son closeness and father identity (B = .04, t = 2.52, p < .01), (b) father-son closeness and father identity (B = .02, t = 2.35,p < .01), (c) marital closeness and father identity (B = .02, t = 1.96, p < .05), (d) father-son closeness and father role (B = .08, t = 4.15, p < .001), and (e) marital closeness and father role (B = .04, t = 1.98, p < .05).

BIVARIATE ANALYSES

Table 1 shows the correlations, means, and standard deviations for all the observed indicators. As would be expected by the conceptual model, mother-son closeness, father-son closeness, and marital closeness were all positively and significantly associated with one another. For example, marital closeness and father-son closeness had a positive and significant correlation of .68, p < .01. With the exception of co-parent relationship, the correlations among the three family closeness indicators (mother-son closeness, father-son closeness, and marital closeness) were strong. This suggests that these three indicators are measuring a similar factor. This was also the case for the correlation between expectant fathers' attitudes about the father role and father identity (r = .61, p < .01).

The co-parent relationship indicator was highly correlated with the marital closeness indicator (r = .72), and therefore there may be some concern of multi-collinearity (Pedhazur & Schmelkin, 1991). The co-parent relationship indicator was also positively and significantly related to all other family of origin indicators (father competence, mother-son closeness, father-son closeness, marital closeness). For example, the co-parent relationship indicator was positively and significantly associated with the father-son closeness indicator (r = .58, p < .01).

As predicted, correlations revealed that current family indicators were associated with expectant fathers' attitudes about father involvement (see Table 1). Expectant fathers' current views of their marriages were positively and significantly correlated with attitudes about the father role (r = .36, p < .01) and father identity (r = .34, p < .01). Expectant mothers' attitudes about father identity (mother-father identity) were positively and significantly associated with fathers' attitudes about the father role (r = .28, p < .01) and father identity (r = .25, p < .01), and expectant mothers' attitudes about the father role (mother-father role) were positively and significantly associated with the father role (r = .22, p < .01). The mother-father role and mother-father identity indicators that represent the expectant mother's attitudes about father involvement factor are correlated higher with each other than with any other indicators (r = .44, p < .01). This suggests that the mother-father role and mother-father identity indicators are measuring a similar factor. Expectant parents' current relationships with their parents (current-parent) were positively and significantly correlated with father role (r = .23, p < .01) and father identity (r = .31, p < .01). Another interesting finding was that expectant parents' current relationships with their parents (current-parent) were positively and significantly related to four family of origin relationships, namely, mother-son closeness, father-son closeness, marital closeness, and co-parent consensus.

STRUCTURAL EQUATION ANALYSIS

The LISREL 8.3 (Joreskog & Sorbom, 1997) program was used to compute the structural equation model (SEM) estimates. Figure 2 summarizes the key findings from the SEM analysis. The three family of origin factors (father competence, family closeness, co-parent relationship) were not significantly associated with expectant fathers' attitudes about father involvement. These findings are not completely surprising as the preliminary analyses showed a curvilinear relationship between family closeness and expectant fathers' attitudes about father involvement. The SEM analyses only examined linear associations.

[FIGURE 2 OMITTED]

As expected, current family factors directly predicted attitudes about father involvement. As Figure 2 indicates, all three current family factors, expectant fathers' current marital relationships, expectant fathers' and expectant mothers' current relationships with their parents, and expectant mothers' attitudes about father involvement, were positively and significantly associated with expectant fathers' attitudes about father involvement.

As shown in Table 2, a number of the error terms were significantly correlated with one another. For example, the family closeness factor (E2) was positively and significantly correlated with the co-parent relationship factor (E3). The expectant father's current marital relationship factor (FA) was positively and significantly correlated with the current relationship with parents factor (E5). The expectant father's family closeness factor (E2) was positively and significantly correlated with the [beginning strike through]expectant father's current marital relationship factor (E4).[end strike through]

The analytical model produced a non-significant chi-square value ([chi square = 30.81, df = 27, p = .28) indicating a good fit for the data. The Goodness of Fit Indices (GFI = .96, AGFI = .91, CFI = .99) (Hu & Bentler, 1999) and the critical N = 217.43 value (Maruyama, 1998) also suggested a good model fit. The model was also properly specified (RMSEA = .02, p < .05), which indicates that there was a sufficient number of possible solutions for the data considering the sample size. The observed indicators were also very reliable (RMR = .04,p < .05) (Hu & Bentler, 1999).

REGRESSION ANALYSES

We conducted regression analyses primarily to study linear and curvilinear relationships between current and family of origin variables and the dependent variable, expectant fathers' attitudes (see Table 3). Model 1 includes all of the family of origin and current family variables. These variables were entered into the regression model simultaneously. The family closeness variable was a combination of the mother-son closeness indicator, father-son closeness indicator, and the marital closeness indicator. The dependent variable is a combination of the father role indicator and the father identity indicator (expectant father's attitudes of father involvement). All three current family variables, expectant fathers' marital relationships, current relationships with parents, and expectant mothers' attitudes, had a positive and significant relationship with expectant fathers' attitudes about father involvement. Both the linear ([beta]6 = 1.69, B = 1.19, p < .001) and curvilinear ([beta]= 1.16, B= .03,p < .001) family closeness variables were positively and significantly associated with expectant fathers' attitudes about father involvement. This provides evidence to support the notion that the family closeness family of origin variables for expectant fathers have a curvilinear association with attitudes about father involvement.

Model 1 also indicates that the expectant father's father competence variable ([beta] = .17, B = .88, p < .05) had a positive and significant linear association with expectant fathers' attitudes about father involvement. When expectant fathers believed that their fathers were confident in their paternal roles, attitudes about father involvement were more positive. Furthermore, the co-parent relationship variable ([beta] = -.69, B = -1.23, p < .01) had a negative and significant association with expectant fathers' attitudes about father involvement. The items from the co-parent relationship variable asked expectant fathers about how much their parents agreed about child-rearing practices. When expectant fathers' parents disagreed about child rearing practices, expectant fathers' attitudes about father involvement were more positive. The regression analyses showed that family closeness, which included marital closeness items, had a positive and significant relationship with attitudes about father involvement, whereas the co-parenting relationship items that focused on agreement about child-rearing practices had a negative and significant association with attitudes about father involvement.

The high betas for the family closeness variables ([beta]> 1.0) could indicate high multicollinearity between the variables; therefore, a subsequent regression was run with the linear and curvilinear family closeness variables and the dependent variable only (see Model 2, Table 3). The results indicated a positive and significant linear ([beta] = .71, B = .50, p < .001) and curvilinear ([beta] = .60, B = .01, p < .01) association between the factors. A graph was computed using select data points and entering both the linear and curvilinear unstandardized coefficients into the regression equation in order to visualize the curvilinear relationship (U-shaped) between family closeness and expectant fathers' attitudes about father involvement (see Figure 3). The graph shows that, when expectant fathers have either very close or very distant family of origin relationships, they have stronger attitudes about father involvement. The preliminary analyses showed no significant linear association between family closeness and expectant fathers' attitudes about father involvement (flat curve). However, the regression analyses now show a significant linear association between these two variables, as the curvilinear association combined with the linear association "pulls up" the flat curve, and the final results become U-shaped.

[FIGURE 3 OMITTED]

DISCUSSION

The results indicated that expectant fathers' family of origin factors are related to attitudes about father involvement. The regression analyses revealed that family closeness has both a linear and a curvilinear relationship with attitudes about father involvement. Although the SEM analyses revealed that there were no significant linear relationships between family of origin processes and expectant fathers' attitudes, these analyses did indicate that these family of origin factors fit the model well. As mentioned previously, the lack of significant relationships may be explained by the fact that SEM analyses assume a linear relationship between factors.

The findings show support for a combination of the modeling and compensation hypotheses (see also Floyd & Morman, 2000). As Figure 3 indicates, expectant fathers who come from either positive or negative family of origin backgrounds can develop strong attitudes about father involvement. Fathers may learn how to be involved with their children by watching how their parents were either highly affectionate or highly unaffectionate with one another and by receiving lots of affection or little affection from both their parents.

A unique contribution of this study is that expectant fathers' family of origin processes were studied in three different areas: closeness with mother, closeness with father, and parents' marital closeness. Previous family of origin studies have primarily concentrated on either father-child or mother-child closeness rather than focusing on multiple types of family closeness. Attachment studies have clearly shown how mother-child and father-child closeness influences the parenting practices of the next generation (Belsky, 1999; Van Ijzendorrn, 1995). However, less research has focused on children's observations of marital closeness and how this may influence their future parenting practices.

The idea that expectant fathers' family of origin closeness has a curvilinear association with attitudes about parenting is somewhat incongruent with some of the assumptions of developmental and intergenerational theories. Developmental models typically assume that secure attachments between parents and children will lead to children effectively parenting their own children, and insecure attachments between parents and children will lead to ineffective parenting. The results of this study suggest that both very close and very distant relationships with parents can lead to stronger attitudes about father involvement. Intergenerational models assume that adults who are too emotionally close to their parents may be ineffective parents. The results from this study showed that expectant fathers who were emotionally close to their parents, both currently and in the past, had stronger attitudes about father involvement. However, it is possible that the measures used in this study did not tap excessive closeness that intergenerational models posit as leading to poorer parenting.

The findings also suggest that a father's competence in his role as a father can influence his offspring's attitudes about father involvement. Father competence had a significant linear association with expectant fathers' attitudes about father involvement. That is, expectant fathers who believed that their fathers felt confident in their paternal role had stronger attitudes about father involvement. Lamb et al.'s (1985) conceptual model about father involvement focuses on how fathers who are confident in their paternal roles will be highly involved fathers; a unique contribution of this study is to show how paternal competence can influence the next generation's attitudes about father involvement.

The regression analyses also showed that fathers in this study had stronger attitudes about father involvement when their parents disagreed about how they should be raised. The co-parenting relationship variable had a significant linear association with expectant fathers' attitudes about father involvement. Pleck (1997) hypothesized that when marital satisfaction studies focus on conflict or disagreements between parents, there is a negative relationship between father involvement and marital satisfaction. Whereas, when marital satisfaction studies focus on global marital satisfaction, there is a positive association between father involvement and marital satisfaction. The co-parent relationship variable in this study primarily included items assessing parents' agreement or disagreement about raising a child. We speculate that an expectant father may have stronger attitudes about father involvement when his own father was more involved in parenting and therefore had more disagreements with his mother. Further research is needed to replicate this finding and directly test this speculation.

As expected, current family factors, such as fathers' marital relationships, current relationships with parents, and expectant mothers' attitudes were associated with expectant fathers' attitudes about father involvement. The SEM and regression analyses produced positive and significant linear relationships between all the current family factors and expectant fathers' attitudes about father involvement. The regression analyses also revealed that some of the family of origin processes were significantly associated with expectant fathers' attitudes above and beyond the contributions of current family factors. A unique contribution of this study was that current relationships with parents were shown to be an important predictor of attitudes about father involvement. Father involvement research related to current family factors tends to focus only on fathers' partners.

This study has a number of limitations. First, SEM analysis is a large sample technique; typically at least 200 participants are needed (Maruyama, 1998). Although the model was identified, indicating a sufficient sample size for the possible solutions to fit the data, a larger sample size would have been better. However, the SEM analysis allowed us to simultaneously study the associations between multiple factors and the endogenous variable. The majority of fatherhood research relies on regression techniques that do not allow the researcher to study the interactions of multiple factors with the endogenous variable simultaneously. A sample size of 152 couples is relatively large compared to other transition to parenthood studies. Future family of origin and fatherhood research needs to include larger samples with both mothers and fathers, in order to study the unique contributions of both mothers and fathers to child development.

Second, the sample is fairly homogenous in terms of ethnicity, income, and education. A more diverse sample may have revealed different findings. Third, expectant fathers' family of origin relationships were based on retrospective reports, and some researchers argue that these reports are not highly reliable. Future longitudinal research can use family of origin retrospective reports over time to see if they change substantially. Fourth, the father competence factor only included one item about how anxious an expectant father felt his own father was about fulfilling his paternal role. In the future, more items are needed to study family of origin processes that are directly related to father competence that may differ from the concept of father anxiety. More items are also needed that are related to fathers' anxiety, as it is possible that a fathers' anxiety about a paternal role may lead to more or less father involvement.

Finally, the results from this study are a first step in examining how family of origin processes influence father involvement. Positive father attitudes during the prenatal period may not always lead to positive father involvement after the baby is born. Future research with the Parenting Together Project data set will examine how family of origin processes influence father involvement behaviors after the transition to parenthood.

Furthermore, more family of origin research is needed on how mother's family of origin processes are associated with father involvement. Preliminary findings with this sample suggest that expectant mother's family of origin processes are not related to expectant father's attitudes about father involvement. This may be explained by the fact that father involvement outcome measures are primarily designed for fathers, rather than both mothers and fathers.

In conclusion, the findings from this study are a first step to show that family of origin closeness has both a linear and a curvilinear association with expectant fathers' attitudes about father involvement. Fathers seem to both model their positive family of origin experiences and compensate for their negative family of origin experiences. The other two family of origin factors, father competence and the co parent relationship had a linear association with expectant fathers' attitudes about father involvement. When expectant fathers felt their fathers were competent in their paternal role, they had stronger attitudes about father involvement. When expectant fathers believed that their parents disagreed about child rearing practices, they had stronger attitudes about father involvement. Future research that examines how family of origin processes are related to father involvement will need to account for both linear and curvilinear associations.

Table 1

Correlations, Means, and Standard Deviations Among Observed Indicators
(N = 152)

                     1         2          3          4

1.  Father
    Competence       1.00
2.  Mother-Son
    Closeness         .11       1.00
3.  Father-Son
    Closeness         .15        .47 **    1.00
4.  Marital
    Closeness         .18 *      .53 **     .68 **    1.00
5.  Co-parent
    Relationship      .29 **     .44 **     .58 **     .72 **
6.  Father-Marital    .09        .07        .16        .14
7.  Current-Parent   -.03        .43 **     .36 **     .31 **
8.  Mother-Father
    Role              .01       -.13        .02        .07
9.  Mother-Father
    Identity          .01        .04        .07        .14
10. Father Role       .10        .10        .17 *      .14
11. Father
    Identity          .04        .07        .11        .10

    Mean             3.22      20.49      16.57      18.60
    Standard
    Deviation        1.24       4.01       5.01       4.49

                    5          6           7          8

1.  Father
    Competence
2.  Mother-Son
    Closeness
3.  Father-Son
    Closeness
4.  Marital
    Closeness
5.  Co-parent
    Relationship     1.00
6.  Father-Marital    .16        1.00
7.  Current-Parent    .22 **      .37 **    1.00
8.  Mother-Father
    Role              .10         .10       -.02       1.00
9.  Mother-Father
    Identity          .20 *       .14        .02        .44 **
10. Father Role       .17 *       .36 **     .23 **     .22 **
11. Father
    Identity          .10         .34 **     .31 **     .10

    Mean            47.16      117.00      76.14      43.11
    Standard
    Deviation       10.02       11.45      19.48       4.85

                     9          10         11

1.  Father
    Competence
2.  Mother-Son
    Closeness
3.  Father-Son
    Closeness
4.  Marital
    Closeness
5.  Co-parent
    Relationship
6.  Father-Marital
7.  Current-Parent
8.  Mother-Father
    Role
9.  Mother-Father
    Identity          1.00
10. Father Role        .28 **    1.00
11. Father
    Identity           .25 **     .61 **    1.00

    Mean             26.73      41.21      28.76
    Standard
    Deviation         3.10       6.07       2.66

* significant at p < .05. ** significant at p < .01.

Table 2

Results of Correlations Between Error Terms (N = 152)

Parameters                          Estimates

Correlations Between Errer Terms
E1    E2                             .08
E1    E3                             .17 *
E1    E4                             .11
E1    E5                            -.12
E1    E6                             .16
E2    E3                             .80 ***
E2    E4                             .16
E2    E5                             .42 ***
E2    E6                             .18
E3    E4                             .16
E3    E5                             .22 **
E3    E6                             .23
E4    E5                             .37 ***
E4    E6                             .18
E5    E6                             .01

* significant at p < .05. ** significant at p < .01. *** significant at
p < .001.

Table 3

Regression Coefficients of the Associations between Current Family
Variables, Family of Origin Processes, and Expectant Father's
Attitudes (N = 152)

Variable                                  B        SE B    [beta]

Model 1

Current Family Variables
  Father Marital Relationship               .14    .05      .20 ***
  Expectant Mothers' Attitudes              .35    .08      .29 ***
  Current Relationship with Parents         .07    .03      .19 *

Family of Origin Variables                 1.19    .34     1.69 ***
  Family Closeness (linear)                 .03    .008    1.16 ***
  Family Closeness (curvilinear)          -1.23    .43     -.69 **
  Co-parent Relationship (linear)          -.06    .006    -.08
  Co-parent Relationship (curvilinear)      .88    .38      .17 *
  Father Competence (linear)                .83    .49      .12

Model 2
  Family Closeness (linear)                 .50    .16      .71 ***
  Family Closeness (curvilinear)            .01    .01      .60 **

Variable                                  95% B
                                          Confidence  Interval

                                          Lower       Upper

Model 1

Current Family Variables
  Father Marital Relationship               .03        .24
  Expectant Mothers' Attitudes              .18        .52
  Current Relationship with Parents         .01        .15

Family of Origin Variables                  .52       1.95
  Family Closeness (linear)                 .01        .04
  Family Closeness (curvilinear)          -2.07       -.38
  Co-parent Relationship (linear)          -.02        .01
  Co-parent Relationship (curvilinear)      .13       1.63
  Father Competence (linear)                .15       1.80

Model 2
  Family Closeness (linear)                 .19        .81
  Family Closeness (curvilinear)            .01        .03

Note: [R.sup.2] = .35 for Model 1; [R.sup.2] = .07 for Model 2.

* significant at p < .05. ** significant at p < .01. *** significant at
p < .001.

REFERENCES

Ainsworth, M.D. (1973). The development of infant-mother attachment. In B.M. Caldwell & H.N. Ricciuti (Eds.), Review of child development research (pp. 194). Chicago: University of Chicago Press.

Belsky, J. (1984). The determinants of parenting: A process model. Child Development, 55, 83-96.

Belsky, J. (1999). Interactional and contextual determinants of attachment security. In J. Cassidy & P.R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 249-263). New York: Guilford Press.

Blair, S.L., Wenk, D., & Hardesty, C. (1994). Marital quality and paternal involvement: Interconnections of men's spousal and parental roles. The Journal of Men's Studies, 2,221-237.

Bowlby, J. (1969). Attachment and loss: Vol. 1. Attachment. New York: Basic Books.

Bowlby, J. (1988). A secure base: Clinical applications of attachment theory. London: Routledge.

Boszormenyi-Nagy, I., & Ulrich, D.N. (1981). Contextual family therapy. In A.S. Gurman & D.P. Kniskern (Eds.), Handbook of family therapy (pp. 159-186). New York: Brunner/Mazel.

Bowen, M. (1978). Family therapy in clinical practice. New York: Jason Aronson.

Cohn, D.A., Cowan, P.A., Cowan, C.P., & Pearson, J. (1992). Mothers' and fathers' working models of childhood attachment relationships, parenting styles, and child behavior. Development and Psychopathology, 4, 417-431.

Conger, R.D., Cui, M., Bryant, C.M., & Elder, G.H. (2000). Competence in early adult romantic relationships: A developmental perspective. Journal of Personality and Social Psychology, 79,224-237.

Cowan, C.P., & Cowan, P.A., (1992). When partners become parents: The big life change for couples. New York: Basic Books.

Crouter, A.C., Perry-Jenkins, M., Huston, T., & McHale, S.M. (1987). Processes underlying father involvement in dual-earner and single-earner families. Developmental Psychology, 23, 431-440.

Cummings, E.M., & O'Reilly, A.W. (1997). Fathers in family context: Effects of marital quality on child adjustment. In M.E. Lamb (Ed.), The role of the father in child development (3rd ed., pp. 49-66). New York: John Wiley & Sons.

Daly, K. (1993). Reshaping fatherhood: Finding the models. Journal of Family Issues, 14,431-440.

Day, H.D., St. Clair, S.A., & Marshall, D.D. (1997). Do people who marry really have the same level of differentiation of self? Journal of Family Psychology, 11,131-135.

Doherty, W.J., Kouneski, E.F., & Erickson, M.F. (1998). Responsible fathering: An overview and conceptual framework. Journal of Marriage and the Family, 60, 277-292.

Ehrensaft, D. (1987). Parenting together: Men and women sharing the care of their children. New York: Free Press.

Flaherty, J.A., Gavira, S., & Pathak, D.S. (1983). The measurement of social support: The Social Support Network Inventory. Comprehensive Psychiatry, 24, 521-529.

Floyd, K., & Morman, M.T. (2000). Affection received from fathers as a predictor of men's affection with their own sons: Tests of the modeling and compensation hypotheses. Communication Monographs, 67, 347-361.

Framo, J.L. (1981). The integration of marital therapy with sessions with family of origin. In A.S. Gurman & D.P. Kniskern (Eds.), Handbook of family therapy (pp. 133-158). New York: Brunner/Mazel.

Hu, L., & Bentler, P.M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural equation modeling, 61, 1-55.

Joreskog, K.G., & Sorbom, D. (1997). LISREL 8: User's reference guide. Chicago: Scientific Software International.

Lamb, M.E., Pleck, J., Charnov, E.L., & Levine, J.A. (1985). Paternal behavior in humans. American Zoologist, 25, 883-894.

Lewis, J.M., & Owen, M.T. (1995). Stability and change in family of origin recollections over the first four years of parenthood. Family Process, 34,455-469.

Marsiglio, W., Amato, P., Day, R.D., & Lamb, M.E. (2000). Scholarship on fatherhood in the 1990s and beyond. Journal of Marriage and the Family, 62, 11731191.

Maruyama, G.M. (1998). Basics of structural equation modeling. Thousand Oaks, CA: Sage Publications.

Maurer, T.W., Pleck, J.H., & Rane, T.R. (2001). Parental identity and behavior: A contextual model. Journal of Marriage and Family, 63,394-403.

McBride, B.A., & Mills, G. (1993). A comparison of mother and father involvement with their pre-school age children. Early Childhood Research Quarterly, 8,457477.

McBride, B.A, & Rane, T.R. (1997). Role identity, role investments, and paternal involvement: Implications for parenting programs for men. Early Childhood Research Quarterly, 12, 173-197.

Palkovitz, R. (1984). Parental attitudes and fathers' interactions with their 5-monthold infants. Developmental Psychology, 20, 1054-1060.

Pasley, K., Futris, T.G., & Skinner, M.L. (2002). Effects of commitment and psychological centrality on fathering. Journal of Marriage and Family, 64, 130138.

Pedhazur, E.J., & Schmelkin, L.P. (1991). Measurement, design, and analysis: An integrated approach. New Jersey: Lawrence Erlbaum Associates.

Pleck, J.H. (1997). Paternal involvement: Levels, sources, and consequences. In M.E. Lamb (Ed.), The role of the father in child development (3rd ed., pp. 66 104). New York: John Wiley & Sons.

Prest, L., Benson, M.J., & Protinsky, H.O. (1998). Family of origin and current relationship influences on codependency. Family Process, 37, 513-528.

Pruett, K.D. (1987). The nurturing father: Journey toward the complete man. New York: Warner Books.

Radin, N. (1988). Primary caregiving fathers of long duration. In P. Bronstein & C.P. Cowan (Eds.), Fatherhood today: Men's changing role in the family (pp. 127-143). New York: Wiley.

Russell, G. (1986). Primary caretakers and role sharing fathers. In M.E. Lamb (Ed.), The father's role: Applied perspectives (pp. 29-60). New York: Wiley.

Scheck, D.C. (1979). Two measures of parental consistency. Psychology, 16, 37-39.

Shereshefsky, P.M., & Yarrow, L.J. (1973). Psychological aspects of a first pregnancy and early postnatal adaptation. New York: Raven Press.

Snarey, J. (1993). How fathers care for the next generation: A four-decade study. Cambridge, MA: Harvard University Press.

Spanier, G.B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Marriage and the Family, 38, 15-28.

Sroufe, L.A., & Fleeson, J. (1986). Attachment and the construction of relationships. In W.W. Hartup & Z. Rubin (Eds.), Relationships and development (pp. 51-77). Hillsdale, NJ: Erlbaum.

Steele, H., Steele, M., & Fongay, P. (1996). Associations among attachment classification of mothers, fathers, and their infants. Child Development, 67, 541-555.

Van Ijzendoorn, M.H. (1995). Adult attachment representations, parental responsiveness, and infant attachment: A meta-analysis on the predictive validity of the adult attachment interview. Psychological Bulletin, 117, 387-403.

Williamson, D.S. (1981). Personal authority via termination of the intergenerational hierarchical boundary: A new stage in the family life cycle. Journal of Marital and Family Therapy, 7, 441-452.

JOHN M. BEATON

University of Guelph

WILLIAM J. DOHERTY

MARTHA A. RUETER

University of Minnesota

Correspondence concerning this article should be addressed to John M. Beaton, Department of Family Relations and Applied Nutrition, University of Guelph, Guelph, Ontario, Canada, N1G 2W1. Electronic mail: [e-mail].