Effects of Family Foundations on Parents and Children: 3.5 years After Baseline

Abstract

This study investigated the ability of a psychosocial prevention program implemented through childbirth education programs to enhance the coparental and couple relationship, parental mental health, the parent-child relationship, and child outcomes. A sample of 169 heterosexual, adult couples expecting their first child was randomized to intervention and control conditions. The intervention families participated in Family Foundations, a series of eight classes delivered before and after birth, which was designed as a universal prevention program (i.e., applicable to all couples, not just those at high risk). Intent-to-treat analyses utilizing data collected from child age 6 months through 3 years indicated significant program effects on parental stress and self-efficacy, coparenting, harsh parenting, and children’s emotional adjustment among all families, and maternal depression among cohabiting couples. Among families of boys, program effects were found for child behavior problems and couple relationship quality. These results indicate that a universal prevention approach at the transition to parenthood focused on enhancing family relationships can have a significant and substantial positive impact on parent and child well-being.

Few family-focused prevention programs targeting the most malleable period of family and child development, the first three years of a child’s life, have been shown to be effective. A handful of programs -- such as Nurse-Family Partnership (Olds et al., 1998) targeting the transition to parenthood and the Incredible Years (Webster-Stratton, 2005) targeting families with young children -- have demonstrated beneficial impact on families in randomized trials. However, few such preventive programs targeting all families in a universal approach (i.e., targeting the general population, not a subgroup identified as high-risk; (O’Connell, Boat, & Warner, 2009) have been shown to be effective in promoting individual and family well-being (Prinz & Sanders, 2007). Family Foundations, a transition to parenthood program for couples that focuses on the coparenting relationship as a point of entry, has been shown in past studies to have a positive impact on families at six and twelve months after birth (Feinberg & Kan, 2008; Feinberg, Kan, & Goslin, 2010). However, it is important to extend the investigation of preventive effects of transition to parenthood programs past infancy: Evidence described below suggests that a particularly stressful and risky period of family life may occur after infancy, during early childhood.

Early Childhood and the Family

The large literature and debate over the extent of stress experienced by new families around the transition to parenthood (Cowan & Cowan, 1995; Lawrence, Rothman, Cobb, Rothman, & Bradbury, 2008; McHale & Huston, 1985; Perren, von Wyl, Bürgin, Simoni, & von Klitzing, 2005) has overshadowed the family stresses and risks of the early childhood years. In recent years, early childhood has come to be seen as period in which advances in emotion and behavioral regulation and cognitive development prepare children for meeting the social and academic challenges of primary school (Blair & Diamond, 2008). Deficits in either regulatory or cognitive development are seen as contributing to long-term trajectories of disruptive and eventually antisocial behavior and/or school failure. Thus, preventive interventions for this age period have been child-focused, providing compensatory support for regulatory control and cognitive stimulation through preschool programs (e.g., Head Start, Paths Preschool) or enhanced parenting (e.g., Incredible Years).

We propose that early childhood may also represent a stressful period for the whole family, with potentially negative consequences for all family members and family relationships. The stresses and strains of this period for the whole family may in turn negatively impact children. One indicator of the level of strain during this period is the elevated levels of aggression and violence exhibited in families with young children. Several studies have documented that rates of children’s physical aggression appear to be highest between two and four years old (Alink et al., 2006; Tremblay et al., 2004), and then start to decline as children develop better communication skills and regulatory abilities. Yet less well appreciated is that parent-to-child and parent-to-parent violence are elevated during early childhood years. National surveys suggest that severe parent-to-child violence (PCV) occurs in 5% of American families (Straus, 1990; Straus, Hamby, Finkelhor, Moore, & Runyan, 1998b), but this figure may substantially underestimate prevalence among families with young children, where 13% of families reported severe PCV in a representative community sample (Slep & O’Leary, 2005). Moreover, PCV appears to begin early in children’s lives, with parents reporting high rates of physical aggression toward toddler-aged children (Straus, Hamby, Finkelhor, Moore, & Runyan, 1998a). Similarly, the prevalence of physical intimate partner violence (IPV) has been estimated at approximately 14–16% in nationally representative samples (Gelles & Cornell, 1990), but nearly half of couples with young children may engage in IPV (Slep & O’Leary, 2005).

There is substantial evidence linking family violence with elevated levels of stress, depression, and relationship conflict (Black, Slep, & Heyman, 2001; Burman, John, & Margolin, 1992; Caetano, Vaeth, & Ramisetty-Mikler, 2008; Chaffin, Kelleher, & Hollenberg, 1996; Gelles, 1985; Lawrence & Bradbury, 2007; Milner & Chilamkurti, 1991; O’Leary, 1994; Schumacher, Feldbau-Kohn, Slep, & Heyman, 2001; Slep & O’Leary, 2007; Stith, Smith, Penn, Ward, & Tritt, 2004). It is possible, then, that elevated levels of family violence during early childhood reflect broader trends in stress, depression, and relationship conflict among most families during the early childhood-rearing years. Heightened family risk when children are young may be in part due to developmental processes. Young children continue to require a high level of parent involvement to address basic needs (eating, dressing, toileting, bathing, cleaning up toys), yet children’s development allows for increased self-management of some aspects of these activities, resulting in ongoing interpersonal negotiation around daily activities. Moreover, children’s increasing psychological autonomy may lead at times to resistance to parent demands, while fluctuations in psychological state may lead at other times to requests for more caretaking than parents want or have resources to provide.

The family context of managing young children’s emotional and physical needs may also exacerbate stress. Emotionally-laden negotiations with young children around basic needs can trigger conflict between parents when interparental coordination is problematic, or when parents are fatigued and irritable. Further, many parents of a preschooler have a second baby who requires additional childrearing management from parents, leaving less time for sleep, leisure, exercise, and self-care. Management of two children may require additional and complicated interparental coordination; and a second child may evoke complex emotions of both love and rivalry in the first child leading to further opportunities for family conflict (Dunn, 2000; Kreppner, 1988; Patterson, 1982). Indeed, managing sibling relationships are one of the most significant child-rearing challenges parents face (Perlman & Ross, 1997).

Against this background, it is important to assess whether transition-to-parenthood program effects extend to this developmental family period. Although short-term effects of FF during the first year after birth are important, the persistence of effects into early childhood would have practical and theoretical implications. Practically, the persistence of program effects would provide even further justification for the investment in preventive approaches around the transition to parenthood. Theoretically, the documentation of persisting effects would carry implications about the malleability of core aspects of family relationships. That is, if program effects are short lived, one might hypothesize that changes in attitudes and relationships served only as temporary compensation for enduring vulnerabilities (Karney & Bradbury, 1995) in individuals (personality, attachment security) or relationships (conflict resolution patterns). Accordingly, the current study aimed to investigate whether Family Foundations program effects were demonstrated through the child’s third birthday.

Family Foundations

Family Foundations is a series of eight interactive, psycho-educational, skills-based classes, designed for expectant couples who are cohabiting or married. The focus of the program is on enhancing the coparenting relationship—that is, the ways parents coordinate their parenting, support or undermine each other, and manage conflict regarding childrearing (Minuchin, Rosman, & Baker, 1978). Coparenting, a central aspect of family life, has been shown to be related to parental adjustment, parenting, and child outcomes (Feinberg, Kan, & Hetherington, 2007; Margolin, Gordis, & John, 2001; McHale, Kuersten-Hogan, Lauretti, & Rasmussen, 2000; Schoppe, Mangelsdorf, & Frosch, 2001; Van Egeren, 2004). Moreover, the coparenting relationship mediates and moderates the influence of individual parent characteristics, couple relationship quality, and environmental stress on parenting and child adjustment (Feinberg, 2003). This evidence supports a theoretical framework in which the coparenting relationship serves a central regulating function in the family due to its sensitivity to “upstream” factors such as individual parent characteristics; its causal influence on “downstream” factors such as parent adjustment, parenting, and child outcomes; and its ability to transmit or modify influences from upstream to downstream factors. Thus, the coparenting relationship is hypothesized to be a potential leverage point for enhancing family functioning and child outcomes. For further information regarding the theoretical framework and empirical evidence linking coparenting to family relationships, parent and child regulation and well-being, we direct the reader to our prior work (Feinberg, 2002, 2003; Feinberg & Kan, 2008).

Research suggests that the transition to parenthood is an opportune moment because of expectant and new parents’ particular openness to change (Duvall, 1977; Feinberg, 2002). Family Foundations was developed in collaboration with childbirth educators to appeal to first-time, expectant parents and to ensure real-world viability, and was delivered through childbirth education departments at local hospitals. This strategy was chosen because childbirth education is a non-stigmatizing educational framework, and represents an existing institutional niche capable of facilitating dissemination.

The content of the sessions was directly related to the risks emerging during the transition period (Feinberg, 2002). The program’s focus is on helping couples become aware of areas of coparental disagreement before parenthood, and managing disagreements through productive communication, problem solving, and conflict management techniques. The program material prepares parents for the strains of the transition period by noting that many new parents experience a high level of stress and strain (Sanders, Nicholson, & Floyd, 1997). By providing information on other couples’ experiences of parenthood, enhancing communication skills, and facilitating discussion of partners’ expectations for each other, we aimed to minimize the strains of the transition and increase parents’ coparental support and decrease coparental undermining. The program included limited material on parenting an infant: promoting parent-child bonding, infant sleep, and nutrition, largely in the context of the coparenting relationship.

Two prior studies of Family Foundations found positive short-term effects on family functioning. At post-test (approximately 6 months after birth), intervention parents reported more coparenting support, infant soothability, and duration of attention; less depression and anxiety among mothers; and less parent-child dysfunctional interaction than control parents (Feinberg & Kan, 2008). At child age 1 year, observational data indicated that intervention families exhibited more coparenting warmth and inclusion, parenting positivity, and child self-soothing; and less coparenting competition and parenting negativity than control families (Feinberg, Kan, & Goslin, 2009).

The Current Paper

In this paper, we extended our prior work to assess whether the Family Foundations program had positive effects on the outcomes identified above: parental adjustment (stress, efficacy, and depression), coparenting and couple relationship quality, parenting, and child behavior problems. Analyses of outcomes included three waves of data from posttest at six months post-birth through three-years post-birth when such data were available. Measures of parenting and child outcomes, for example, were only available at the last wave of data collection (three years post-birth), as these measures were not appropriate at earlier ages. For measures available at three waves, we analyzed all three waves in a single model to minimize the number of statistical tests. However, we also examined in these models whether there was a program effect on change (linear or quadratic) in these outcomes across the three waves. In this way, we were able to examine whether initial intervention effects apparent at six months post-birth declined (or increased) over the following 2.5 years. If there was no difference between intervention and control groups in change across these three measurement occasions, then we could conclude that any intervention effects detected across the three waves represented an initial intervention effect at posttest that persisted for the next 2.5 years.

All analyses included a consistent set of demographic control variables (age, education, income, financial strain), because these factors have been linked to parenting and the couple relationship in previous work. To control for potential response bias (Morsbach & Prinz, 2006) related to the demand characteristics of participating in the intervention condition, all analyses also control for parent social desirability.

Finally, we examine potential moderating effects of child and parent gender in these analyses. Although we do not offer specific hypotheses, it is possible that program effects could be stronger on mothers or on fathers for a variety of reasons. There may also be differential impact by child gender for a number of reasons. For example, fathers appear to be more involved in the parenting of boys than girls (McHale & Crouter, 2003), with the potential thereby for greater coparental conflict among parents of sons than daughters (Margolin, Gordis, & John, 2001; McHale, 1995). Thus, the program may have greater effects in families with boys given the greater salience of coparental coordination in such families.

Method

Participants

169 heterosexual couples made up the original sample. To be eligible for the study, first-time parents must have been at least 18 years of age, living together (regardless of marital status) and expecting a first child. At the start of the study, 82% of couples were married. Most participants (91% of mothers and 90% of fathers) were white, with the remaining participants including African American, Asian, Hispanic, or other. Median annual family income was $65,000 (SD = $34,372). Average educational attainment was 15.06 years for mothers (SD = 1.82) and 14.51 years for fathers (SD = 2.19); 86% of mothers and 71% of fathers had received at least some post-secondary education. Expectant mothers and fathers had an average age of 28.3 (SD = 4.9) and 29.8 (SD = 5.6). Families were recruited from medium-sized cities in Pennsylvania—Altoona and Harrisburg—and were generally representative of families from these regions. Further details on recruitment are available elsewhere (Feinberg & Kan, 2008).

Pretest data were collected during home interviews when mothers were pregnant (average weeks gestation = 22.9, SD = 5.3). Mothers and fathers separately completed questionnaires and engaged in videotaped interactions. Couples were randomly assigned to intervention (n = 89) or no-treatment control conditions (n = 80) after the first wave of data collection. Initial assessment of group differences based on a wide range of pretest variables -- including age, income, education, marital status, weeks of gestation, and mental health--suggested a successful random assignment process (Feinberg & Kan, 2008).

The no-treatment control group couples received a brief brochure in the mail about selecting quality childcare; intervention couples received the manualized FF program consisting of 4 prenatal and 4 postnatal sessions. An observer from the project team attended and rated over 90% of intervention sessions for implementation fidelity. The observer rated the proportion of the intervention content in the manual delivered by the group leaders. Observer ratings indicated the program was implemented as planned, with an average of 95 percent of the curriculum content delivered. The average number of prenatal sessions attended by each couple in the intervention group was 3.2; the average number of postnatal sessions attended was 2.3. About 80% of couples attended at least 3 prenatal sessions, and about 60% of couples attended at least 3 postnatal sessions.

Follow-up data collection occurred across three non-equally spaced waves: mail-in questionnaire when the child was roughly six months (wave 2), and home visits at child age 12 months (wave 3) and 36 months old (wave 4). Data from four families (two intervention and two control) were not utilized in analyses because of child medical problems, and two intervention families were excluded because of multiple births. Study attrition for mothers and fathers by wave 4 was 17% and 23%, respectively. 84.6% of families (n=137) provided data (from either parent) at wave 4.

Measures

Parental Adjustment.

We assessed parent efficacy with the 16-item Parenting Sense of Competence scale, asking mothers and fathers how they feel about their parental role (Gibaud-Wasston & Wandersman, 1978) (e.g., “I feel confident in my role as a parent”) using a seven-point likert scale. Reliability coefficients (alphas) were .84 for mothers and .83 for fathers. Self-report of parental stress was measured through the total score on the Parenting Stress Index (PSI; (Abidin, 1997). This scale consisted of 27 items asking parents to rate their agreement with certain statements (e.g., “I feel trapped by my responsibilities as a parent”), utilizing a five-point likert scale (alpha=.90 for mothers, .87 for fathers). We assessed parental depression using an abbreviated seven-item version of the Center for Epidemiological Studies Depression Scale (CES-D), which asks the respondent to indicate their feelings and outlook within the past week (Radloff, 1977) (alpha=.86 for mothers, .83 for fathers). The abbreviated version has been found to correlate strongly with the full scale. Subjects indicated the degree to which they felt lonely, or think people were unfriendly using a four-level scale ranging from rarely/none of the time to always/most of the time.

Coparenting and Couple Relationship.

We assessed coparenting relationship quality with the 31-item Coparenting Scale , which was created based on prior work (e.g., (Abidin & Brunner, 1995; Cordova, 2001; Frank, Olmstead, Wagner, & Laub, 1991; Margolin, Gordis, & John, 2001; McHale, 1997). The overall score represents an average of items covering theoretically important domains: coparental agreement, support, undermining, and exposure of the child to conflict (alpha=.72 for mothers, .65 for fathers). To assess the quality of the couple relationship, we included a measure of relationship satisfaction from the Quality of Marriage Index (Norton, 1983). This score was created from an average of six items – combining five items that ask parents to rate their agreement to statements about their relationship using a seven-point likert scale (e.g., “We have a good relationship”) with one item asking them to rank how happy they feel their relationship is on a scale of 1 to 10 (alpha=.97 for mothers, .95 for fathers; items were standardized before averaging.)

Parenting.

The Parenting Scale assesses discipline practices in parents of children from 18–48 months (Arnold, O’ Leary, Wolff, & Acker, 1993). Our assessment focused on three outcomes: The Laxness scale (based on 11 items assessing permissive parenting; alpha=.85 for mothers, .82 for fathers); the Overreactivity scale (9 items assessing the degree of authoritarian parenting; alpha=.76 for mothers, .78 for fathers); and a single item assessing the likelihood of the parent to “spank, slap, grab or hit” a misbehaving child, which we term Physical Punishment. Due to the skewed distribution of this score, the three highest values (5–7 on a seven-point likert scale) were combined to create a five-point ordinal ranging scale from 1 to 5 representing level of punishment likelihood.

Child Outcomes.

Child behavior problems were measured using the Child Behavior Checklist (CBCL; (Achenbach & Rescorla, 2000), reported by mothers only. From this 100-item questionnaire, separate sub-scales were calculated using scoring conventions. From these, we examined three overall scores which were normed for child age: Total Problems, Externalizing problems, and Internalizing problems (alpha=.90, .88, and .83, respectively). In addition, we examined two specific sub-scales given their relevance to this study: Aggression and Attention/Hyperactivity (alpha=.85 and .78, respectively). Measures of the child’s social and emotional adjustment according to mother report were taken from a measure designed for young children (Head Start Competence Scale; Domitrovich, 2001). A scale measuring Social Competence was comprised of an average of eight items assessing the child’s interactions (e.g., “resolves problems with friends on his/her own,” alpha=.84); an Emotional Competence scale was calculated from six relevant items (e.g., “copes with sadness,” alpha=.80).

Control Variables.

All program impact regression models included four background control variables--respondent age, marital status, family income, and respondent education level (in number of years of education)—and two variables based on parent report on established scales: Social Desireability and Financial Strain. The former -- based on a 33-item scale including such statements as “I am always courteous, even to people who are disagreeable;” (Crowne & Marlow, 1964) -- was included as a control for tendency to respond in a favorable manner (alpha=.75 for mothers, .69 for fathers). Financial Strain was measured with a three-item scale asking parents to indicate their degree of current financial hardships (utilizing a five-point likert scale; (Kessler, Turner, & House, 1988); alpha=.75 for mothers, .78 for fathers). We also included a measure of maternal Relationship Attachment Insecurity as a control variable in models assessing program impact on child social and emotional competence. This variable was created as an average of 20 items from the Relationship Scales Questionnaire (Griffin & Bartholomew, 1994) where mothers rate their agreement with relationship concerns (e.g., “I worry that others don’t value me as much as I value them.”) on a five-point likert scale (alpha=.73). For the single outcome for which we had pretest data available—depression—we included the pre-intervention (wave 1) score as a control variable in the regression model.

Statistical models

Separate statistical analysis was executed for each outcome. All tests of intervention effects were conducted as intent-to-treat analyses; data from all parents who responded at one or more of the post-intervention waves of data collection were included regardless of level of program participation. Condition was coded 0 for control and 1 for intervention. Our analytic strategy was to first test the main effect of the intervention. We then assessed variation in intervention impact (i.e., moderation) based on child gender. For outcome variables available from both parents, we also tested for differential intervention impact based on parent gender. For parent adjustment and couple relationship outcomes, we tested for moderation by marital status given that cohabiting couples generally display lower levels of well-being (Willitts, Benzeval, & Stansfeld, 2004). We plan to examine other potential moderators in future research.

Analytic models were structured to accommodate the number of waves of data available and the number of respondents per family (both parents versus one parent). For outcomes reported by both mothers and fathers, we used multi-level models with parent nested within family. SAS Proc Mixed was used for multi-level models when outcomes were approximately normally distributed, and Stata’s GLLAMM for multi-level ordinal models. For outcomes reported only by one parent (e.g., mother report in the CBCL), single-level regression was conducted with SAS Proc GLM.

For outcomes where data were available across multiple waves, multi-level models were used in which time-specific observations were nested within family, aggregated at the parent level. Such models are appropriate for modeling repeated measures within dyads, and rely on a data structure whereby separate level-1 intercepts and time-coefficients are specified for mothers and fathers to represent average parent levels and change rates for the outcome (Laurenceau & Bolger, 2005). Time at each wave was measured in terms of time elapsed since the prior assessment date, and varied slightly across families; time was coded as 0 at wave 2 allowing for the intervention coefficient to represent average post-intervention effect in the absence of a time interaction. Random intercept terms were specified for both parents; random slopes were specified for separate variables representing the impact of time on the outcome for mothers and fathers, distinctly. These random intercepts and slopes were allowed to co-vary. Models were parameterized to accommodate multiple intercepts at the second level of the multilevel model (using SAS PROC MIXED) (Kashy & Donnellan, 2008). A fixed-effect, the square of the time variable, was added to the model to assess curvilinear change—but removed from the model if non-significant (p>.05). We tested whether the intervention and control groups had changed at different rates across waves 2–4 by testing the impact of condition on variation in the random slope term. If this test was non-significant, we did not retain the effect in the final model.

In multi-level models in which parents were nested within family, both mother and father control variables were included. In models incorporating multiple measurement occasions, marital status was allowed to vary by wave, and economic strain was included as a person-level average across all periods measured (waves 1, 3 and 4).

Attrition.

We have reported previously that randomization was successful in that intervention and control groups demonstrated no significant differences on a range of background and study variables (Feinberg & Kan, 2008). For this report, we assessed potential differential attrition across condition at wave 4. First, we used logistic regression models to examine whether study participation between intervention and control groups was related to any key background characteristics (including the control variables listed above, as well as pretest levels of parent depression (CES-D) and a measure of anti-social behavior history). Only one variable was found to be related to differential study participation: Among non-participating mothers at wave 4, mother’s education level was lower in those originally assigned to the control group versus the intervention group (t=−2.90, p<.05); the opposite was true among those still participating (t=2.00, p<.05). These results imply that attriters in the control condition were less educated than attriters in the intervention condition. Because of this difference, we ran several alternative models testing program impact using multiple imputation techniques (Raghunathan, Lepkowski, Van Hoewyk, & Solenberger, 2001). Results from these analyses were very similar to the corresponding models without imputation in terms of coefficients and statistical significance. Accordingly, we report analyses carried out on wave 4 participants only. As noted above, parent education level is included in all regression models as a control variable.

Results

Descriptive statistics are provided in Table 1, including means/standard deviations separately by intervention status as well as measurement period for variables used in these analyses. Results of tests of program impact are detailed in Table 2. As noted, we tested all outcomes for both moderation of intervention effect by child gender, and (where appropriate) parent gender. For parental adjustment and inter-parental relations outcomes, we also assessed possible moderation of intervention effects based on marital status. We describe such moderation results below only if statistically significant (p<.05); there was no evidence of moderation by parent in any of the results. Effect sizes (Cohen’s d) are listed in Table 2 only for significant effects (calculated based on model-adjusted means).

Table 1.

Sample descriptive statistics, by control and intervention conditions

			Control		Intervention
	Wave*	Range	Mean	SD	Mean	SD
Demographics and Family characteristics
Mother age	1	18–42	28.54	5.28	28.61	4.38
Father age	1	20–55	30.11	6.18	30.05	4.92
Mother education (years)	1	10–17	15.61	1.66	15.06	1.72
Father education (years)	1	9–17	14.97	2.16	14.49	2.08
Family income (thousands $)	1	7.5–150	66.93	32.97	65.82	32.93
Financial Strain**	4	3–9.5	4.38	1.63	4.72	1.78
Social Desirability**	1	3–31	18.28	4.98	17.92	4.87
Parental Adjustment
Parental Stress	2	1–3.10	1.94	0.45	1.89	0.44
	3	1–3.25	1.95	0.45	1.90	0.41
	4	1–3.25	2.04	0.50	1.92	0.49
Parental Efficacy	2	60–112	89.72	12.53	91.36	11.62
	3	60–112	91.94	12.28	93.47	10.65
	4	60–112	91.35	12.61	94.39	11.28
Depression (CES-D)	2	0–2	0.32	0.38	0.27	0.31
	3	0–2	0.31	0.37	0.29	0.33
	4	0–2	0.36	0.42	0.31	0.39
Interparental Relationship
Coparenting Quality	2	2–4.0	3.21	0.44	3.33	0.38
	3	2–4.1	3.18	0.46	3.26	0.40
	4	2–4.1	3.12	0.43	3.22	0.42
Relationship Satisfaction	4	−3.3–1.3	0.01	0.85	0.01	0.96
Parenting
Overreactivity	4	1–5	2.52	0.76	2.31	0.66
Laxness	4	1–5	2.50	0.74	2.33	0.78
Physical Punishment	4	1–5	2.15	1.35	1.80	1.15
Child outcomes
CBCL-Total (T score)	4	28–71	46.17	8.54	45.23	8.67
CBCL-Externalizing (T score)	4	28–67	46.15	8.76	45.71	8.32
CBCL-Internalizing (T score)	4	28–67	46.49	8.46	46.03	8.69
CBCL-Attention/Hyper.	4	0–12	3.66	2.72	3.67	2.43
CBCL-Aggression	4	0–19	7.85	5.16	7.58	4.81
Child Social Competence	4	1–4	2.32	0.47	2.4	0.48
Child Emotional Competence	4	1–4	2.88	0.48	2.89	0.57

Notes:

^*Wave: 1 = pretest; 2 = 6 months post-birth [posttest]; 3 = 12 months post-birth; 4 = 3 years post-birth.

^**Financial strain and social desirability are pooled for mothers and fathers.

Table 2.

Intervention main effects and moderation of intervention effects by child gender

Outcomes	Coefficients (beta)			Effect Size
	Condition	Condition x Gender	Condition x Parent x Mar.Stat.
Parental Adjustment
Parental Stress^	−0.11*			0.16
Parental Efficacy^	2.77*			0.18
Depression (CES-D)^	−0.02		−0.42*	0.72
Interparental Relationship
Coparenting Quality^	0.12*			0.18
Relationship Satisfaction	−0.33	0.71*		0.43
Parenting
Overreactivity	−0.26*			0.35
Laxness	−0.23*			0.30
Physical Punishment	−1.09*			0.36
Child outcomes
CBCL-Total (T score)	0.07	−6.10*		0.81
CBCL-Externalizing (T score)	0.28	−5.95*		0.78
CBCL-Internalizing (T score)	1.98	−7.41*		0.70
CBCL-Attention/Hyper.	0.39	−1.84*		0.62
CBCL-Aggression	0.15	−3.49*		0.79
Child Social Competence	0.19*			0.43
Child Emotional Competence	0.13

Notes:

^*p<.05. Gender: Male = 1, Female = 0.

^{^}data include three waves of post-intervention data (waves 2–4). If interaction not significant, term dropped from model. Effect size (Cohen’s d) for outcomes moderated by gender represent effect for boys; effect size for depression (moderated by marital status) represents effect for cohabiting mothers. For the outcome analyzed with a non-linear model (Physical Punishment), effect size approximated with alternative linear regression models.

Rates of change.

For models that assessed multi-wave outcomes (Parental Stress, Efficacy, Depression, and Coparenting Quality), no significant differences between conditions in linear or curvilinear change were found. In other words, there was no evidence that intervention and control conditions differed in rates of change for those outcomes across posttest wave 2 through follow-up waves 3 and 4. Thus, all differences between intervention and control groups reported below represent average overall differences across the three post-intervention waves.

Parental Adjustment.

Models of outcomes across waves 2–4 indicated positive intervention effects for Parenting Stress, Efficacy, and Depression. Intervention group parents indicated lower rates of Parenting Stress on average across that time range (b=−.11, p=.031) and higher Parent Efficacy (b=2.77, p=.024). The overall intervention-control group difference on the Depression scale (CES-D) was non-significant. Marital status significantly moderated the effect of condition on change (i.e., a significant 3-way interaction, b=−.42, p=.008). A test on adjusted group mean differences indicated significantly higher levels of depression for control group non-married mothers compared to intervention group non-married mothers (p=.003); the same mean contrast was non-significant between groups for married mothers and among intervention and control groups for fathers. Marital status did not moderate intervention effects on Stress and Efficacy.

Coparenting and Couple Relationship.

An intervention main effect was found on the overall measure of positive Coparenting (b=.12, p=.011) considering data from waves 2–4. In the analysis of Relationship Satisfaction (available at wave 4 only), no main effect of the intervention was found. In this case, however, we did find a significant interaction between intervention status and child gender (b=.711, p=.007), indicating that condition was associated with Relationship Satisfaction among parents of boys. A test of adjusted group mean differences confirmed a significant between-group difference just for parents of boys (p<.05), with intervention families showing higher relationship quality than control families. The same mean contrast was non-significant for parents of girls. Marital status did not moderate intervention impact on these outcomes.

Parenting.

Multi-level regression models also indicated intervention effects for all three outcomes from the Parenting scale (parent-report at wave 4). Intervention parents indicated lower levels of Over-reactivity (b=−0.24, p=.019) and Laxness (b=−0.22, p=.049) and less likelihood to inflict Physical Punishment (b=−1.16, p=.014).

Child outcomes.

Significant intervention effects were found on the child measures of behavior (CBCL), although child gender was a factor in these results. For the Total Problems scale, overall intervention group differences were found (main effects model: b=−3.23, p=.022) indicating a lower levels of behavior problems in the intervention condition. When the interaction of child gender and condition was entered, this term was significant (b=−6.10, p=.027). Model-based group mean comparisons indicated that the intervention effect was driven by differences among control and intervention boys, where model-adjusted mean differences (control group minus intervention group) in Total Problems were 6.03 among boys (p=.002), and −0.07 among girls (n.s.). Table 2 provides the regression results for the final models where significant child gender-interaction terms were included. A similar pattern of findings was found for externalizing CBCL scales: Externalizing scale (main effects model: b=−2.93, p=.031; interaction term: b=−5.95, p=.025), Aggression scale (main effects model: b=−1.73, p=.029; interaction term: b=−3.53, p=.024). Model adjusted mean differences among boys were 5.66 (p=.002) and 3.33 (p=.002) for Externalizing and Aggression, respectively (non-significant differences between control and intervention girls).

For Internalizing Problems and the Attention/Hyperactivity scale, the test for intervention effects for the whole sample was not significant; however, the child gender interaction term was significant for Internalizing (b=−7.41, p=.012) and Attention/Hyperactivity (b=−1.84, p=.031). Model adjusted mean differences among boys were 5.42 (p=.007) and 1.45 (p=.013) for Internalizing and Attention/Hyperactivity, respectively (no significant differences between control and intervention girls).

Finally, results for models assessing impact on child Social Competence also indicated a significant intervention effect (b=.195, p=.032). Results for Emotional Competence did not demonstrate significant intervention impact (b=.134, p=.165). Tests of moderation by child gender were non-significant for these two outcomes.

Discussion

In this paper, we assessed the outcomes of a randomized trial of Family Foundations when children were three years old. Earlier reports indicated that families in the intervention condition demonstrated positive impact from the program in multiple domains, including parental adjustment, coparenting, parenting, and child regulatory capacity. The results of this paper extend these prior findings by a further two years, into the potentially stressful period of early childhood when risk for IPV and PCV may be high.

Consistent with our prior reports, we found positive intervention impact in all domains. For the three measures of parental adjustment—stress, efficacy, and depression—as well as coparenting relationship quality, we assessed intervention-control differences across three waves of data: posttest at 6 months post-birth, and follow-ups at 1 year and 3 years post-birth. For all four of these outcomes, we found that the intervention and control groups did not change at different rates. Instead, we found that there was a significant difference between conditions in the overall level of each measure across the three waves (for depression, this was true for cohabiting mothers only). Thus, from 6 months to 3 years postbirth, intervention condition parents reported less parental stress, more parental efficacy, less depression, and better coparenting quality than control parents. Given the absence of pretest differences on the parental adjustment measures (due to successful randomization), these results suggest than an intervention-related change occurred by posttest and remained through child age 3 years.

Although the primary target of the program was on the coparenting relationship, and our theory conceptualized the coparenting relationship as leading to changes in parental adjustment (Feinberg, 2002; Feinberg & Kan, 2008), the pattern of findings across coparenting and parental adjustment do not allow us to make a claim about the sequencing of effects. We will pursue formal mediational analyses in future work to explore this issue. However, we expect that impact on parental adjustment via coparenting occurred at a relatively short time-scale and would not be detected with our measurement design.

We also detected evidence of significant program impact on the quality of the couple relationship, parenting, and child behavior problems at child age three years. These findings were expected. Our theory regarding the role of coparenting suggests that couple relationship quality, at least during the early childhood years, is influenced by coparenting support and coordination (Feinberg, 2002). Across the transition to parenthood, parents’ time, energy, and psychological identity shifts from a preoccupation with the romantic couple relationship to coparenting and childrearing (Cowan & Cowan, 2000). Thus coparenting relationship quality is likely an important determinant of overall couple relationship well-being, as found in previous research (Schoppe-Sullivan, Mangelsdorf, Frosch, & McHale, 2004).

However, we did not predict that program impact on couple relationship quality would be found only for parents of boys. There is some evidence that fathers are more involved in the parenting of boys than girls (McHale & Crouter, 2003), with some reports indicating greater coparental conflict for parents of sons than daughters (Margolin, Gordis, & John, 2001; McHale, 1995). It may be that the program primed mothers’ to expect greater engagement by fathers in sharing and coordinating care of children, an expectation that may have tended to be violated more in families with daughters. On the other hand, the program’s preparation of couples for managing coparenting conflict may have been helpful for families with sons, in which fathers’ greater involvement evoked greater coparental conflict.

The intervention condition parents reported less harsh, physical, and over-reactive parenting than control parents, while simultaneously reporting less lax, permissive parenting as well. Thus, the effect of the intervention appears to have been to support high-quality parenting that avoids being too intrusive and harsh on the one side, and permissive on the other. We expected such a result given program impact on coparenting and parental adjustment. Maintaining a calm but involved parenting style can be undermined by stress, depression, low parental efficacy, and a lack of coparental support. The program’s ability to address these risk/protective factors may result in enhanced parenting even without a strong program focus on parenting.

Further, evidence suggests that the children in intervention families demonstrated better social competency and boys evidenced lower levels of internalizing and externalizing problems. These findings follow our earlier findings with this sample showing greater attentional control and soothability by parent report, and more self-soothing capacity according to observer rating for intervention children compared to control children. We appear to be observing the ongoing impact of the intervention on unfolding dimensions of behavioral expression of social-emotional-behavioral competence. This impact on capacities and competencies as they come on line may follow from initial impact on the infant, and/or may be mediated by persisting effects on parent adjustment, couple relationship, and parenting competence (itself a construct undergoing developmental change).

Effect sizes were small to moderate in this investigation for parental adjustment and the coparenting relationship, and moderate to large for parenting and child outcomes. The magnitude of these effects suggests that a modest impact on several risk/protective factors may combine to produce a larger impact on more distal outcomes. Further research is needed, however, to understand more precisely the cumulative meditational pathways.

Of course, replication of intervention results are important and a second trial of Family Foundations is underway. Moreover, limitations of this study include the fact that it is restricted to parent self-report data, which may involve a reporting bias as participants could not be blinded to condition. Although the sample had a large range of education and income, the majority of participants were well-educated and middle class. This feature of the sample is in part a result of the eligibility requirements for the study which excluded young teen parents and parents who were not cohabiting or married. Moreover, there was limited racial or ethnic diversity. Finally, although randomization was apparently successful, we do not have a pretest measure of relationship satisfaction to use a control in that analysis (naturally, there are no pretest measures of factors that only arise after birth, such as coparenting and parenting).

Despite these limitations, the results from this trial are promising, suggesting that Family Foundations may have a positive impact on multiple important domains of parent, child, and family well-being. The results also indicate that a strategically crafted preventive intervention, based on developmental family theory and research, can have a persisting impact on family relationship quality and child outcomes during the stressful period of early childhood. And finally, these results support the validity of the theoretical framework we have presented elsewhere, supporting the view that coparenting relationship quality plays an important role in shaping other family relationships as well as parent and child well-being (Feinberg, 2003; Feinberg, Neiderhiser, Reiss, & Hetherington, 2005).