Randomized Effectiveness Trial of the New Beginnings Program for Divorced Families with Children and Adolescents

Abstract

This study presents findings from a randomized effectiveness trial of the New Beginnings Program (NBP), which has demonstrated efficacy in 2 prior randomized efficacy trials. Family courts in 4 counties facilitated recruitment of divorcing and separating parents, and providers in community agencies delivered the program. Participants were 830 parents of children ages 3–18 who were randomized to receive either the 10-session NBP or an active 2-session comparison condition in which parents learned about the same parenting skills but did not complete home practice of these skills. Parents were ethnically diverse (59.4% non-Hispanic White, 31.4% Hispanic, 9.2% other race or ethnicity). Multiple rater assessments of parenting, interparental conflict, and child mental health problems were conducted at pretest, posttest, and 10-month follow-up. The results indicated positive moderated effects of the NBP as compared with the active control condition to strengthen parenting at posttest and to reduce child mental health problems at posttest and 10 months. Many of these moderated effects showed positive benefits for non-Hispanic White families but not for Hispanic families. The findings indicate support for the effectiveness of the NBP when delivered by community-based agencies but also indicate the need for further adaptations to make the program effective for Hispanic parents.

INTRODUCTION

This article reports on outcomes from a randomized effectiveness trial of the New Beginnings Program (NBP), a program that has demonstrated positive effects in two randomized efficacy trials (Wolchik et al., 2000, 1993). There are two primary functions of an effectiveness trial (Sandler et al., 2016). The first is to translate the intervention developed as an experimental prototype to a service that can be effective when delivered in community agencies (Glasgow & Chambers, 2013; Sandler et al., 2016). The second is to test whether the positive effects observed in the efficacy trials can be generalized across “a wide range of populations, settings, and times” (Gottfredson et al., 2015, p. 910) when the program is delivered as a community service. The public health significance of this effectiveness trial is supported by the high prevalence of divorce in the United States (National Center for Health Statistics, 2008) and the increased risk divorce confers for children in multiple areas of functioning (Amato, 2001). Next we describe findings from the efficacy trials of the NBP. We then summarize changes that were made to translate the program from an experimental prototype tested in the efficacy trials to a community service. Finally, we describe the questions addressed in this evaluation of the NBP as a community service.

Findings from Randomized Efficacy Trials of the NBP

The NBP is a group-based program designed to impact two factors that have been empirically associated with child outcomes following divorce—quality of parenting and child exposure to interparental conflict. The theory of the program is that program effects to improve these factors will lead to improved child outcomes following divorce. The initial randomized efficacy trial (N = 70, children ages 8–15) found that compared to families in a waitlist control group, families in the NBP showed an improvement in parenting and a reduction in child mental health problems at posttest (Wolchik et al., 1993). The second efficacy trial with mothers who had primary residency (50% or more time) of children ages 9–12 (N = 240) tested the effects of the parenting program alone, the parenting program plus a concurrently run child coping program, and a literature control condition. The results indicated that at posttest there was a significant effect of the parenting program to strengthen parenting and reduce children’s externalizing problems. At the 6-month follow-up, the parenting program reduced children’s externalizing problems as reported by parents, children, and teachers. Follow-up evaluations at 6 years and 15 years found a broad range of benefits, including reductions in mental health problems and disorders, substance use and abuse, and high-risk sexual behavior (Wolchik et al., 2013; Wolchik, Sandler, Weiss, & Winslow, 2007); less involvement in the criminal justice system; and use of mental health services (Herman et al., 2015), as well as improvements in indicators of positive functioning such as self-esteem and grade point average (Wolchik et al., 2007). Consistent with the program theory, many of these effects were mediated through improvements in quality of parenting (Tein, Sandler, MacKinnon, & Wolchik, 2004; Zhou, Sandler, Millsap, Wolchik, & Dawson-McClure, 2008). The findings from this trial indicated no significant additive benefit of the child coping program at any assessment, so most analyses compared the combined parenting program and the parenting-plus-child coping program to the literature control.

Translation from an Experimental Prototype to a Community Service

The NBP was delivered in both efficacy trials at a university prevention research center, and all aspects of implementation were controlled by the research team, including recruitment of participants as well as selection, training, and supervision of leaders. Translating the program into a community service required developing collaborations with the family courts to recruit families and with community agencies to deliver the program. A fuller description of the program translation process and the consumer research that informed these changes is presented elsewhere (Sandler et al., 2016; Wolchik et al., 2009) and is briefly reviewed next.

We collaborated with four family courts in Arizona, which agreed to show a 12-min invitational DVD during the brief (4-hr) parent information program that was mandated for all divorcing and separating families with minor children. For the program to be acceptable to the court, we needed to offer the program to divorcing and separating families, fathers as well as mothers, and families with a broad age range of children (ages 3–18). Further, the number of eligibility criteria was greatly reduced compared to the efficacy trials. Thus, the families in the effectiveness trial were expected to be quite different from those in the efficacy trials. Because Arizona has a large Hispanic population, we expected the sample to be ethnically diverse, as compared with the efficacy trials, where the sample was primarily non-Hispanic White. As compared to the sample of primary residential mothers in the efficacy trials, we accepted fathers and mothers with a broad range of parenting time as long as it was sufficient to practice the parenting skills taught in the program. The collaboration with the court in inviting parents and the broadening of the eligibility criteria enabled us to offer the program to the population of divorcing and separating families, thus greatly enhancing the potential public health benefit of the program.

The program was delivered by providers in community agencies. The agencies were selected using a competitive solicitation process similar to the way the court normally solicits agencies to deliver services. As compared to the efficacy trials where we screened and selected leaders, the agencies selected the leaders to deliver the program. As described next, the program was modified to make it easier and less costly for community-based providers to deliver the program and to make it appropriate for delivery to an ethnically diverse population.

Evaluation of the Effectiveness of the NBP

An effectiveness trial provides a critical bridge between demonstrating a program’s potential impact in an efficacy trial and assessing its impact when it is delivered at scale (Gottfredson et al., 2015; Spoth et al., 2013). The current article addresses two sets of questions that are important in establishing effectiveness of the NBP. First, does the NBP improve protective factors and reduce risk factors as well as improve children’s adjustment? Program effects were assessed on the two theoretical mediators targeted by the program—quality of parenting and child exposure to inter-parental conflict. These mediators have been found to relate to outcomes for children of all ages (Amato, 2010; Kelly, 2012), so it was theoretically appropriate for the program to target a broad age range of children. Program effects on children’s mental health problems and competencies were assessed because short-term effects on these variables have been found to predict long-term effects in cascading models over 15 years (Wolchik, Tein, Sandler, & Kim, 2016). Second, are the effects of the NBP moderated by characteristics of the participants? Moderators of program effects were tested because prior evidence showed moderated program effects in the efficacy trials (e.g., baseline status on the outcome; Tein et al., 2004; Wolchik et al., 2000, 1993) or because the sample included sufficient variability on participant characteristics to which we wished to generalize inferences about program impact (e.g., parent gender, child age, ethnicity). Testing generalizability of programs across populations is one of the major objectives of an effectiveness trial (Gottfredson et al., 2015). We also tested level of interparental conflict as a moderator because of the intense interest in high conflict as a risk factor for poor outcomes of children following divorce (Amato, 2001, 2010).

METHOD

Participants

The sample consisted of 830 families randomized to NBP (n = 445) or comparison (n = 385) in which either the mother (n = 474; 57.1%) or father (n = 356; 42.9%) enrolled. The sample size was selected to have adequate power (> .80) to detect main and moderated effects. At the pretest assessment, 262 (31.6%) of the 830 parents were divorced, 442 (53.3%) were legally married but divorcing, and 126 (15.2%) were never legally married but were in court to establish or change a parenting time agreement following separation. Parents were ethnically diverse(59.4% non-Hispanic White, 31.4% Hispanic, 9.2% other race or ethnicity). The Hispanic parents were primarily(84.3%) of Mexican heritage, with the remainder from Puerto Rico (4.6%), Cuba (0.8%), or another Latino country (10.3%). Parents had a wide range of education (3.7% less than a GED or high school diploma, 15.7% GED or high school diploma, 36.5% some college or vocational training, 14.5% associate degree, 29.6% bachelor’s degree or higher). Parents ranged from 18 to 63 years old (M = 37.46, SD = 8.16), and children ranged from 3 to 18 years old (M = 8.43, SD = 4.15); 47.8% of the children in these families were female. At the pretest assessment, mothers reported more overnights with their children in the past month (M = 23.18, SD = 7.82, Mdn = 26) than did fathers (M = 14.15, SD = 9.42, Mdn = 15), t (680.96) = −14.68, p < .01.

The trial was conducted in partnership with family courts in two urban and two small-town, rural counties in Arizona. Parents were primarily recruited (92.4%) by viewing a 12-min invitational DVD shown during a 4-hr parent information program mandated for all parents seeking a divorce or separation. Parents provided demographic information, completed a 15-item risk index previously shown to predict children’s postdivorce adjustment (Tein, Sandler, Braver, & Wolchik, 2013), and indicated their interest in participating in a randomized effectiveness trial. Parents were also recruited through judge, lawyer, and self-referrals and through media announcements. Parents who expressed interest were screened by phone based on the following eligibility criteria: filing for divorce or modification of a divorce decree within the past 2 years or, if never married, being in court in the past 2 years to establish or change a parenting time agreement following separation; having at least one child age 3 to 18 with whom the parent spends 3 hours or more each week or one overnight every other week; being able to complete the program and assessments in English; and not being mandated to a parenting class by the Juvenile Court or Child Protective Services. Following completion of the pretest interview, 886 parents were randomly assigned using computer-generated random numbers, with a 53%-to-47% ratio for the NBP to comparison condition. These 886 parents included 112 paired parents in which both father and mother (for 56 families) participated in either the NBP or comparison condition. However, we excluded data from the second parent enrolled in each of these 56 pairs, resulting in a sample of 830 parents for our analyses. The trial was conducted in four cohorts between 2012 and 2014. The CONSORT diagram in Figure 1 summarizes how many parents were invited, screened, randomized, and assessed at pretest, posttest, and 10-month follow-up. Data were collected in four Arizona counties (two urban and two rural) using telephone interviews with parents and children and using online or paper-and-pencil questionnaires for teachers.

FIGURE 1.

CONSORT diagram summarizing how many parents were invited, screened, randomized, and assessed at pretest, posttest, and 10-month follow-up.

During the pretest assessment, parents were asked for permission to interview their 9- to 18-year-old children and for permission to interview the teacher(s) of their 6- to 18-year-old children who were currently enrolled in school and not being homeschooled. Data were obtained at pretest, posttest, and/or 10-month follow-up from 559 (73.8%) of the 757 children age 9 or older and from teachers of 687(96.5%) of the 712 eligible children whose parent provided permission to collect teacher-reported data. Parents and children were paid $50 and $30, respectively, for their interviews. Teachers in the first cohort were paid $10, and those in the second through fourth cohorts were paid $5 for their assessments. Interviewers were masked to condition. Success of masking was assessed at posttest and 10-month follow-up by asking interviewers whether they thought they knew which condition the parent was in, and, if so, which condition; 95.4% at posttest and 95.3% at 10-month follow-up reported not knowing or answered with the incorrect condition. The study was approved by the Institutional Review Board at Arizona State University.

Intervention and Comparison Conditions

NBP

The NBP targeted three potentially modifiable parenting domains—parent–child relationship quality, children’s exposure to interparental conflict, and effective discipline, all of which have been identified as being associated with children’s postdivorce adjustment (Amato, 2010; Kelly, 2012). The small theory of the intervention was that program effects to strengthen parent–child relationship quality and effectiveness of discipline and to reduce children’s exposure to interparental conflict would lead to a reduction of child mental health problems (Wolchik et al., 2007). Drawing from social learning and cognitive-behavioral approaches, the NBP used parent training, videotaped modeling, role-playing, and weekly home practice to teach and review parenting skills during 10 group sessions and two individual phone sessions. Parents who did not attend a session were invited to come early to the following session and observe a 15- to 20-min self-administered make-up DVD that presented the skill taught in the missed session.

The program was adapted from the NBP tested in the efficacy trials with divorced mothers of children ages 8–15 (Wolchik et al., 2007) to prepare it for delivery on a much larger scale in community agencies (Sandler et al., 2016). Parents received tips for using the program skills with children across the broad age range. A comprehensive process to broaden the program material and skills to meet the needs of the parents from diverse cultural groups was undertaken. Data from these sources were used to make modifications that maintained fidelity to NBP’s core components but tailored presentation of the program skills to meet the needs of the parents from these groups (Gonzales et al., 2006). Prevention scientists and providers with expertise in working with Mexican American and African American families reviewed the protocol, and Mexican American and African American parents who participated in a pilot test of the program provided feedback. Across all the cultural informant types, the topics covered in the NBP were seen as appropriate cross-culturally. However, numerous suggestions for “surface structure” changes were made and implemented. For example, the program was modified to include more culturally relevant examples, the program activities were broadened to incorporate extended family members, and the videotaped skills demonstrations included greater ethnic diversity of the actors. Other changes included offering culturally sensitive rationales and explanations of program skills that were compatible with goals and values of ethnic minority parents; acknowledging the potential cultural barriers to specific program skills or recommendations, such as systematic reinforcement for positive behaviors and the NBP’s recommendation against spanking; and providing more opportunities for all parents to consider ways to make program skills fit their cultural styles and preferences. For example, extensive reliance on good listening skills was perceived as conflicting with Mexican American and African American parents’ role-based desires to provide advice and guidance to their children. We followed our cultural informants’ suggestions to acknowledge these strong desires and the difficulty that parents might have using listening skills prior to or in place of parental advice and to augment the discussion of how just using good listening could benefit their children.

The manual was revised to ease leader burden by providing DVD-assisted delivery, and a more economical training protocol was developed that included online training for each session. To reduce cost, the program was shortened from 11 sessions to 10 sessions and was delivered by a single leader rather than two as in the efficacy trials. The program was delivered to 26 mother groups and 24 father groups (average group size = 9.54, range = 4–15).

Active Comparison

Whereas prior efficacy trials of the NBP used a waitlist or literature control condition, the effectiveness trial used an active comparison condition due to the necessity of offering a credible intervention to all families enrolled through the family courts. During two group sessions, the leader didactically presented the risk and protective factors targeted in the NBP. Parents were encouraged to exchange their ideas for strengthening postdivorce parenting skills and reducing children’s exposure to interparental conflict. Parents were also asked to set goals that they wanted to accomplish in the program. Important to note, parents did not role-play parenting skills and were not instructed to practice the parenting skills at home. The program was delivered to 22 mother groups and 22 father groups (average group size = 9.30, range = 4–14).

Group Leaders

The programs were delivered by 38 group leaders (23 non-Hispanic White, 12 Hispanic, and 3 other race or ethnicity). Leaders each led only one experimental condition.

Measures

The time frame used for all measures was “within the past month.” Parents completed the full battery of parenting skills and child behavior problem measures for a randomly selected “target child.” Parents also completed the Child Monitoring Scale (described next) for all (target and non-target) children age 9 or older.¹ Data were also obtained from all (target and nontarget) children age 9 or older whose parent provided permission for the interview and from teachers of children age 6 or older who were enrolled in school and whose parent provided permission for the interview. We report internal consistency reliability (Cronbach’s alpha) at pretest where applicable as well as references for reliability and validity of measures.

Parenting Skills

Parents and children responded to several measures assessing two dimensions of parenting skills: parent–child relationship quality and discipline. The same measures (with different wording for the reference person) were administered to parents and children unless specified. Family routines were assessed by a seven-item (e.g., “You and child” routinely enjoyed activities together) adaptation of the Family Routines Inventory, which Jensen, James, Boyce, and Hartnett (1983) reported as having adequate reliability and validity (parent-report α = .82, child-report α = .78 in the current sample). Involvement was measured by six items adapted from Menning (2006) assessing activities parents engaged in with their children (reliability not applicable; e.g., “During the past month, did you play a sport with child?”). The scale has good predictive validity for father involvement following divorce (Menning, 2006). Communication was assessed by the 10-item (e.g., “Child found it easy to discuss problems with you”) Open Communication subscale of the Parent–Adolescent Communication Scale, which Barnes and Olson (1982) reported as having adequate reliability and validity (parent-report α = .82, child-report α = .93 in the current sample). Closeness was measured by a single item, “How close do you feel to your child/parent?” which has been found to predict well-being of children following divorce (Stewart, 2003). Parents and children completed an 11-item (e.g., “How much do you know about your child’s choice of friends, who they are, what they are like?”) version of the Child Monitoring Scale, which has been found to have good reliability and validity (Hetherington et al., 1992; parent-report α = .90, child-report α = .85); parents completed this measure for each target and nontarget child age 9 or older. The remaining measures completed by parents and children each had good reliability and predicted outcomes in a previous efficacy trial of the NBP (Tein et al., 2004; Wolchik et al., 2000). Parents and children completed three subscales of the Child Report of Parental Behavior Inventory (Schaefer, 1965): acceptance (16 items, e.g., “You saw child’s good points more than (his/her) faults”; parent-report α = .87, child-report α = .95), rejection (16 items; e.g., “You were not very patient with [child]”; parent-report α = .80, child-report α = .88 in this sample), and consistency of discipline (eight items; e.g., “You only kept rules when it suited you”; parent-report α = .80, child-report α = .80 in this sample). Parents (but not children) completed three subscales of the Oregon Discipline Scale (Oregon Social Learning Center, 1991): follow-through (11 items; e.g., “How often did child get away with things that you feel (he/she) should have been punished for?”; α = .75 in this sample), appropriate use of discipline (nine items; e.g., “When child misbehaved, how often did you restrict privileges?”; α = .76 in this sample), and inappropriate use of discipline (five items; e.g., “When child misbehaved, how often did you yell?”; α = .70 in this sample). A ratio of appropriate to appropriate plus inappropriate use of discipline was computed.

Parent-Report Confirmatory Factor Analysis Model.

Parent-report of parent–child relationship quality and discipline were assessed using factor scores from a two-factor confirmatory factor analysis model estimated in Mplus 7.3 (Muthén & Muthén, 1998–2014; see Sandler et al., 2018, for more details about the confirmatory factor analysis model). Indicators of parent–child relationship quality were involvement, family routines, parent–child communication, closeness, and acceptance. Indicators of discipline were consistency, follow-through, and appropriate use of discipline (ratio). When fitting the two-factor model, we allowed involvement and family routines to correlate due to their conceptual similarity and acceptance and consistency of discipline to correlate due to being measured by items on the same scale. The two-factor model closely fit the data at pretest, χ² (17) = 69.86, RMSEA = .06, CFI = .96; posttest, χ² (17) = 57.77, RMSEA = .06, CFI = .97; and 10-month follow-up, χ²(17) = 41.25, RMSEA = .05, CFI = .97. When investigating measurement invariance across parent gender, the metric invariance model did not significantly deteriorate fit relative to the configural invariance model, and the scalar invariance model did not significantly deteriorate fit relative to the metric invariance model. The scalar invariance model closely fit the data at pretest, χ² (46) = 102.18, RMSEA = .05, CFI = .96; posttest, χ² (46) = 84.76, RMSEA = .05, CFI = .97; and 10-month follow-up, χ²(46) = 94.05, RMSEA = .06, CFI = .95. When investigating measurement invariance across parent ethnicity, the metric invariance model did not significantly deteriorate fit relative to the configural invariance model, and the scalar invariance model did not significantly deteriorate fit relative to the metric invariance model. The scalar invariance model closely fit the data at pretest, χ² (46) = 102.53, RMSEA = .06, CFI = .96; posttest, χ² (46) = 101.35, RMSEA = .06, CFI = .96; and 10-month follow-up, χ²(46) = 77.28, RMSEA = .05, CFI = .96. Thus, the measurement model for parenting was invariant across parent gender and ethnicity. Parent-report of rejection and monitoring were analyzed as two separate measures beyond parent–child relationship quality and discipline. We treated rejection as a separate measure because it did not load highly with the other indicators of parent–child relationship quality (see Sandler et al., 2018). We treated monitoring as a separate measure because parents completed this measure only for children age 9 or older but completed all other measures for children of all ages.

Child-Report Confirmatory Factor Analysis Model.

Consistent with parent-report, we initially fit a two-factor model of parent–child relationship quality and discipline. Because these factors were highly correlated (r = .92 at pretest), we then estimated a one-factor model of positive parenting. Indicators of positive parenting were involvement, family routines, parent–child communication, closeness, acceptance, rejection, consistency of discipline, and monitoring. When fitting the one-factor model, we allowed involvement and family routines to correlate due to their conceptual similarity and acceptance, rejection, and consistency of discipline to correlate due to being measured by items on the same scale. The one-factor model closely fit the data at pretest, χ² (17) = 53.25, RMSEA = .07, CFI = .98; posttest, χ² (17) = 32.63, RMSEA = .05, CFI = .99; and 10-month follow-up, χ² (17) = 44.58, RMSEA = .06, CFI = .98.

Interparental Conflict

Parents reported on interparental conflict using four items that asked about children’s exposure to conflict adapted from the Children’s Perception of Interparental Conflict Scale that has previously been reported to have good reliability and validity (Grych, Seid, & Fincham, 1992) plus two items written for this study that reported on arguing in front of the child (α = .90). Only parent-reports about arguing in front of the child were used, because the program was designed to reduce exposure to conflict rather than the occurrence of conflict per se. Parents who reported no contact with the other parent were assigned the lowest score on this scale. Children reported on their exposure to interparental conflict using 13 items from the Children’s Perception of Interparental Conflict Scale (e.g., “In the past month, your parents got really mad when they argued.”; α = .87 in this sample) plus the two items concerning the parents arguing in front of them. Children also reported their perception of “being caught between their parents” using the seven-item (e.g., “In the past month, how often does your mother ask you to give messages or information to your father?”) Caught in the Middle Scale, which has previously been found to have good reliability and validity (Buchanan, Maccoby, & Dornbusch, 1991; α = .78 in this sample), and on their exposure to badmouthing using two items (e.g., “In the past month your dad said bad things about your mom;” Sandler, Wolchik, & Braver, 1988).

Child Mental Health Problems

Parents completed the Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2001) for children aged 6 to 18 (α = .89, .90, and .95 for internalizing, externalizing, and total problems) and the Preschool CBCL (Pre-CBCL; Achenbach & Rescorla, 2000) for children ages 3–5 (α = .89, .91, and .95 for internalizing, externalizing, and total problems). The T scores were calculated for CBCL and Pre-CBCL subscales based on child age and gender and combined to assess internalizing, externalizing, and total problems across the broad age range (T. M. Achenbach, personal communication, 2015).

Children ages 9 or older completed the Brief Problem Monitor (Achenbach, McConaughy, Ivanova, & Rescorla, 2011) to assess internalizing (six items, α = .79), externalizing (seven items, α = .71), and total (19 items, α = .86) problems. Teachers completed a parallel version of the Brief Problem Monitor to assess internalizing (six items, α = .88), externalizing (six items, α = .88), and total (18 items, α = .91) problems for children age 6 or older who were currently enrolled in school. The T scores for child-and teacher-reports were calculated based on child age and gender.

Developmental Competencies

Teachers completed the Teacher–Child Rating Scale, which has previously shown good reliability and validity (Hightower et al., 1986) to evaluate learning problems (six items, α = .95; e.g., “learning problems”), task orientation (eight items, α = .96; e.g., “well organized”), frustration tolerance (10 items, α = .93; e.g., “accepts things not going his/her way”), assertive social skills (seven items, α = .92; e.g., “defends own view”), and social competence (five items, α = .95; e.g., “listens carefully to others”).

Data Analytic Strategy

We assessed the equivalence of families randomized to the NBP and comparison conditions on 33 parent-reported demographic and baseline measures using t tests and chi-square tests. Univariate and multivariate outlier analyses were conducted to identify influential data points (Kutner, Nachtsheim, Neter, & Li, 2005). To assess the impact of attrition on internal and external validity, we compared the attrition rates across conditions using chi-square tests and performed 2 × 2 (condition by attrition status) analysis of variance or logistic regression on each baseline measure (Jurs & Glass, 1971).

Intent-to-treat analyses were conducted by regressing each parent-, child-, and teacher-reported outcome on condition and the following baseline covariates: baseline status on the outcome, baseline risk (Tein et al., 2013), county of residence, parent gender, and an indicator of whether both parents were enrolled in the effectiveness trial (true for 56 of the 830 families). All analyses were performed in Mplus7.3 (Muthen & Muthen, 1998–2014) using full information maximum likelihood estimation, which includes cases with missing scores. Intraclass correlations (ICCs) by intervention group were low (average cluster size = 8.83; parent-report M_ICC = .02 across pretest, posttest, and 10-month follow-up), whereas ICCs by family were high (average number of children per family = 1.84; parent-report of monitoring² M_ICC = .58, child-report measures M_ICC = .26, teacher-report measures M_ICC = .16 across pretest, posttest, and 10-month follow-up). A sandwich estimator was used for the standard error computations to adjust for clustering by family for parent-report of monitoring, child-report, and teacher-report (Yuan & Bentler, 2000).

We investigated whether the program effects were moderated by baseline status on the outcome, parent gender, parent ethnicity (non-Hispanic White vs. Hispanic), child age, and interparental conflict,³ one moderator at a time. Significant moderated effects were probed via simple main effects, which were computed at the mean and at 1 SD above/below the mean of a continuous moderator and at each level of a categorical moderator (Aiken & West, 1991). To control the Type I error rate, we applied a false discovery rate correction (FDR; Benjamini, Drai, Elmer, Kafkafi, & Golani, 2001) separately for each domain of outcomes with multiple measures: parent-report of parenting; child-report of interparental conflict; and parent-, child-, and teacher-reports of child mental health problems and developmental competencies. For these measures, we report the main and moderated effects with FDR p-values ≤ .10. For domains of outcomes with a single indicator, we report main and moderated effects with ps ≤ .05. Cohen’s d is reported for the significant main effects and simple main effects (Cohen, 1988). For all significant main or simple main effects on internalizing, externalizing, or total problems, we assessed clinical significance by comparing the proportion of children in the NBP versus comparison who moved from the clinical or borderline clinical range (T score ≥ 60) at pretest to the normal range (Tscore < 60) at posttest or 10-month follow-up.

RESULTS

Preliminary Analyses

The two conditions significantly differed on three of the 33 demographic and baseline characteristics. Relative to those in the comparison condition, parents in the NBP were higher on child-report of positive parenting and lower on child-report of interparental conflict and badmouthing at pretest (see Table A1 in the online appendix). To adjust for these imbalances, we included these three baseline covariates in the outcome models and assessed the robustness of the results. None of the conclusions changed, so because including significant baseline characteristics to the outcome models is an inferior option to using covariates related to the outcome (Begg, 1990; Permutt, 1990), we report the results without including these three baseline covariates in the outcome models. No influential data points were identified based on univariate or multivariate outlier analyses.

The attrition rates did not significantly differ across conditions at posttest, χ² (1) = 0.44, p = .51, or 10-month follow-up, χ² (1) = 0.58, p = .45. However, four of the 60 attrition status main effects were significant and two of the 60 Condition × Attrition Status interaction effects were significant. On average, parents who did not complete the posttest interview reported less education, lower risk, and fewer child internalizing problems at pretest than those who did, and parents who did not complete the 10-month follow-up interview reported less education than those who did. Parents in the comparison condition who did not complete the posttest interview reported fewer child externalizing and total problems at pretest than those who did. Parents in the NBP who did versus did not complete the posttest interview did not significantly differ on child externalizing or total problems at pretest.

Treatment Integrity

Of the 445 parents randomized to the NBP, 107 (24.0%) never attended, 92 (20.7%) attended between one and four sessions, 192 (43.1%) attended between five and nine sessions, and 54 (12.1%) attended all 10 sessions. The mean number of sessions attended (either regular or make-up) was 5.59 (55.9% of the 10 sessions). Interested readers should refer to Mauricio et al. (2017) for additional information on mothers’ and fathers’ attendance as well as attendance trajectory class differences on baseline characteristics. Of the 385 parents in the comparison condition, 65(16.9%) never attended, 55 (14.3%) attended one session, and 265 (68.8%) attended both sessions. A significantly greater proportion of NBP parents than comparison parents never attended, χ² (1) = 6.44, p = .01.

Fidelity of implementation of the NBP was assessed by objective rater coding of leader behavior for one activity per session, selected to represent the different kinds of program activities (e.g., home practice review, didactic, role-play). Activities were coded as totally completed (3), partially completed (2), or not completed at all (1). The mean fidelity of implementation score across activities was2.86, indicating a high level of fidelity to the manual.

Effects on Parenting Skills and Child Exposure to Interparental Conflict

Table 1 presents the results of the analyses for parenting skills and child exposure to interparental conflict at posttest and 10-month follow-up.

TABLE 1.

Program Effects on Parent and Child Reports of Parenting Skills and Interparental Conflict at Posttest and 10-Month Follow-Up.

Continuous Outcome	Adjusted Mean (NBP)	Adjusted Mean (Control)	Program Main Effect [95% CI]	Program Main Effect p/pFDR	Significant Moderators p/pFDR	Cohen’s d
Posttest
Parent Report of Parenting
Relationship Quality	0.08	−0.08	0.16 [0.06, 0.25]	< .01/< .01	Ethnicity (< .01/< .01)  Conflict (.03/. 12)	dWhite = 0.45
Discipline	0.12	−0.06	0.17 [0.07, 0.27]	< .01/< .01		d = 0.22
Rejection	1.41	1.44	−0.02 [−0.06, 0.01]	.19/.19	Ethnicity (.02/.05)	dWhite = 0.22
Monitoring	4.46	4.39	0.08 [−0.01, 0.17]	.07/.10		d = 0.13
Child Report of Parenting	−0.01	−0.09	0.08 [−0.03, 0.18]	.14/NA
Parent Report of Conflict	1.28	1.33	−0.05 [−0.10, 0.01]	.09/NA
Child Report of Conflict
Interparental Conflict	1.49	1.52	−0.03 [−0.11, 0.05]	.49/.61	Ethnicity (.06/.17)
Caught in the Middle Scale	1.85	1.88	−0.03 [−0.16, 0.09]	.61/.61
Badmouthing	1.41	1.45	−0.04 [−0.15, 0.08]	.51/.61
10-Month Follow-Up
Parent Report of Parenting
Relationship Quality	0.08	0.01	0.07 [−0.04, 0.17]	.22/.41
Discipline	0.03	−0.03	0.06 [−0.05, 0.17]	.31/.41
Rejection	1.41	1.41	0.00 [−0.04, 0.05]	.85/.85
Monitoring	4.43	4.32	0.11 [0.02, 0.21]	.02/.09	Child Age (< .01/.01)	d−1SDage = 0.13
					Conflict (.03/.10)	d+1SDage = 0.19
						dMcoflict = 0.16
						d+1SDconflict = 0.19
Child Report of Parenting	−0.08	−0.10	0.02 [−0.13, 0.16]	.83/NA	Parent Gender (.04/NA)
Parent Report of Conflict	1.25	1.26	−0.02 [−0.07, 0.04]	.58/NA	Baseline (.05/NA)
Child Report of Conflict
Interparental Conflict	1.49	1.51	−0.02 [−0.11, 0.06]	.61/.86
Caught in the Middle Scale	1.93	1.94	−0.01 [−0.16, 0.13]	.86/.86	Ethnicity (.09/.25)
Badmouthing	1.47	1.55	−0.08 [−0.22, 0.06]	.27/.80

Note: Main and moderated effects with p_FDR ≤ .10 are in bold. The last column reports Cohen’s d for simple main effects with p ≤ .10 and main effects with p_FDR ≤ .10 (when no moderated effects were significant). The simple main effects for Hispanic families were nonsignificant and favored the comparison condition (d_Hispanic = 0.14 and 0.15 for parent-reported relationship quality and rejection at posttest). NBP = New Beginnings Program; CI = confidence interval; p = p value from the original analysis; p_FDR = p value after the false discovery rate correction; NA = false discovery rate correction was not used because only one outcome was investigated.

Posttest

Parents in the NBP reported more effective discipline (b = 0.17, z = 3.21, p < .01, d = 0.22) and greater monitoring (b = 0.08, z = 1.81, p = .07, d = 0.13) relative to those in the comparison condition. Parent ethnicity moderated the effect of the program on parent-report of parent–child relationship quality (b = −0.40, z = −4.07, p < .01) and rejection (b = 0.10, z = 2.27, p = .02), such that non-Hispanic White parents in the NBP reported greater parent–child relationship quality (b = 0.32, z = 4.97, p < .01, d = 0.45) and lower rejection (b = −0.52, z = −2.43, p = .02, d = 0.22) relative to those in the comparison condition. There were no significant main or moderated effects of the program on child-report of exposure to interparental conflict at posttest.

10-Month Follow-Up

The program effect on parent-report of monitoring was moderated by child age (b = 0.06, z = 3.12, p < .01) and baseline interparental conflict (b = 0.29, z = 2.24, p = .03). Relative to those in the comparison condition, parents in the NBP reported less monitoring of younger children (age 9 at pretest, b = −0.10, z = −1.89, p = .06, d_{−1SD =} 0.13) but greater monitoring of older children (age 13 and older; b = 0.24, z = 2.80, p = .01, d_{+1SD =} 0.19). Parents in the NBP who were at the mean or at 1 SD above the mean of baseline interparental conflict reported greater monitoring of their children relative to those in the comparison condition (49.8% of children were in the region of significance; b = 0.11, z = 2.32, p = .02, d_Mean = 0.16; b = 0.26, z = 2.77, p = .01, d_{+1SD =} 0.19). At 10-month follow-up, there were no significant main or moderated effects of the program on parent-report of child exposure to interparental conflict or on child-report of parenting skills or exposure to interparental conflict.

Effects on Child Mental Health Problems and Developmental Competencies

Table 2 presents the results of the analyses for child mental health problems and developmental competencies at posttest and 10-month follow-up.

TABLE 2.

Program Effects on Parent-, Child-, and Teacher-Reports of Child Mental Health Problems and Developmental Competencies at Posttest and 10-Month Follow-Up

Continuous Outcome	Adjusted Mean (NBP)	Adjusted Mean (Control)	Program Main Effect [95% CI]	Program Main Effect p/pFDR	Significant Moderators (p/pFDR)	Cohen’s d
Posttest
Parent Report
Internalizing Problems	51.55	53.48	−1.93 [−3.12, −0.74]	< .01/< .01	Ethnicity (.06/.09)	dWhite = 0.36
Externalizing Problems	51.08	52.22	−1.14 [−2.17, −0.10]	.03/.03	Ethnicity (.02/.06)	dWhite = 0.28
Total Problems	51.28	52.96	−1.68 [−2.73, −0.62]	< .01/< .01	Ethnicity (.10/.10)	dWhite = 0.33
Child Report
Internalizing Problems	56.05	56.05	0.00 [−1.02, 1.03]	.99/.99	Child Age (.02/.06)	d−1SDage = 0.14
Externalizing Problems	53.97	54.11	−0.15 [−1.07, 0.78]	.76/.99
Total Problems	56.72	56.66	0.06 [−0.99, 1.11]	.91/.99	Child Age (.08/.12)
Teacher Report
Internalizing Problems	55.51	55.57	−0.06 [−0.81, 0.70]	.88/.94
Externalizing Problems	53.65	53.57	0.08 [−0.63, 0.79]	.82/.94	Ethnicity (.02/. 16)
					Child Age (.01/.03)	d−1SDage = 0.16
Total Problems	55.20	55.31	−0.11 [−0.80, 0.57]	.74/.94
Learning Problems	2.01	1.98	0.03 [−0.11, 0.17]	.66/.94	Conflict (.03/.19)
Task Orientation	3.59	3.62	−0.03 [−0.14, 0.09]	.65/.94	Child Age (.02/.03)
					Conflict (.05/.19)	d−1SDage = 0.16
Frustration Tolerance	3.96	3.95	0.01 [−0.09, 0.11]	.88/.94	Child Age (< .01/.03)	d−1SDage = 0.17
						d+1SDage = 0.14
Assertive Social Skills	3.67	3.67	0.01 [−0.10, 0.11]	.94/.94	Child Age (.01/.03)	d−1SDage = 0.16
Social Competence	4.54	4.50	0.04 [−0.10, 0.18]	.60/.94	Parent Gender (.04/32)
10-Month Follow-Up
Parent Report
Internalizing Problems	51.71	52.23	−0.52 [−1.93, 0.90]	.47/.90	Baseline (.08/.24)
					Ethnicity (.03/.10)	dWhite = 0.17
Externalizing Problems	50.69	50.50	0.19 [−1.03, 1.41]	.76/.90
Total Problems	51.41	51.33	0.083 [−1.20, 1.37]	.90/.90	Ethnicity (.10/. 15)
Child Report
Internalizing Problems	56.50	57.55	−1.06 [−2.27, 0.16]	.09/.27	Ethnicity (< .01/< .01)	dWhite = 0.48
Externalizing Problems	53.97	54.18	−0.21 [−1.33, 0.90]	.71/.71	Parent Gender (.06/. 18)
Total Problems	57.54	58.31	−0.77 [−1.96, 0.43]	.21/.31	Ethnicity (.01/.01)	dWhite = 0.34
Teacher Report
Internalizing Problems	55.05	55.32	−0.28 [−1.19, 0.64]	.56/.89
Externalizing Problems	53.45	53.76	−0.31 [−1.30, 0.67]	.53/.89
Total Problems	55.31	55.44	−0.13 [−1.04, 0.78]	.78/.89	Child Age (.06/.44)
Learning Problems	2.03	2.03	0.00 [−0.18, 0.18]	.98/.98
Task Orientation	3.51	3.47	0.04 [−0.13, 0.20]	.67/.89
Frustration Tolerance	3.90	3.85	0.05 [−0.08, 0.17]	.48/.89
Assertive Social Skills	3.66	3.57	0.10 [−0.04, 0.24]	.17/.68
Social Competence	4.56	4.39	0.17 [−0.01, 0.35]	.06/.47

Note: Main and moderated effects with p_FDR ≤ .10 are in bold. The last column reports Cohen’s d for simple main effects with p ≤ .10 and main effects with p_FDR ≤ .10 (when no moderated effects were significant). The simple main effects for Hispanic families were nonsignificant and sometimes favored the comparison condition (d_Hispanic = 0.05, 0.10, and 0.06 for parent-reported internalizing, externalizing, and total problems at posttest. NBP = New Beginnings Program; CI = confidence interval; p = p value from the original analysis; p_FDR = p value after the false discovery rate correction; d_Hispanic = 0.17 for parent-reported internalizing problems at 10-month follow-up; d_Hispanic = 0.20 and 0.23 for child-reported internalizing and total problems at 10-month follow-up; . Specifically, the direction of the simple main effects favored the comparison condition for parent reported externalizing problems at posttest; parent-reported internalizing problems at 10-month follow-up; and child-reported internalizing and total problems at 10-month follow-up.

Posttest

Parent ethnicity moderated the effect of the program on parent-report of internalizing problems (b = 2.61, z = 1.90, p = .06), externalizing problems (b = 2.79, z = 2.34, p = .02), and total problems (b = 1.96, z = 1.63, p = .10). Non-Hispanic White parents in the NBP reported fewer internalizing problems (b = −3.08, z = −3.96, p < .01, d = 0.36), externalizing problems (b = −2.02, z = −3.09, p < .01, d = 0.28), and total problems (b = −2.46, z = −3.59, p < .01, d = 0.33) relative to those in the comparison condition. Child age moderated the effect of the program on child-report of internalizing problems (b = 0.45, z = 2.30, p = .02), such that younger children in the NBP reported fewer internalizing problems relative to those in the comparison condition (age 9, b = −1.36, z = −1.64, p = .10, d_{−1SD =} 0.14).

Child age also moderated the effect of the program on teacher-report of externalizing problems (b = −0.27, z = −2.48, p = .01), task orientation (b = 0.04, z = 2.38, p = .02), assertive social skills (b = 0.04, z = 2.51, p = .01), and frustration tolerance (b = 0.04, z = 2.86, p < .01). Teachers reported more externalizing problems (for children age 8 and younger, b = 1.18, z = 2.07, p = .04, d_{−1SD =} 0.16), lower task orientation (for children age 8 and younger, b = −0.19, z = −2.14, p = .03, d_{−1SD =} 0.16), and lower assertive social skills (for children age 8 and younger, b = −0.17, z = −2.12, p = .03, d_{−1SD =} 0.16) in the NBP relative to the comparison condition. Teachers reported lower frustration tolerance for younger children in the NBP (age 8 and younger, b = −0.17, z = −2.22, p = .03, d_{−1SD =} 0.17) but higher frustration tolerance for older children in the NBP (age 13 and older, b = 0.12, z = 1.80, p = .07, d_{+1SD =} 0.14) relative to those in the comparison condition.

10-Month Follow-Up

Parent ethnicity moderated the effect of the program on parent-report of internalizing problems (b = 3.54, z = 2.12, p = .03) and child-report of internalizing problems (b = 4.45, z = 3.19, p < .01) and total problems (b = 3.83, z = 2.84, p < .01). Relative to those in the comparison condition, non-Hispanic White parents in the NBP reported fewer internalizing problems (b = −1.69, z = −1.85, p = .06, d = 0.17) and children of non-Hispanic White parents in the NBP reported fewer internalizing problems (b = −2.94, z = −4.40, p < .01, d = 0.48) and total problems (b = −2.11, z = −3.12, p < .01, d = 0.34). There were no significant main or moderated effects of the program on teacher-report of child mental health problems or developmental competencies at 10-month follow-up.

Clinical Significance

Follow-up analyses on the eight significant main or simple main effects on child mental health problems indicated that only two showed a significantly or marginally significantly greater proportion of NBP versus comparison children moving from the clinical or borderline clinical range at pretest to the normal range at posttest or 10-month follow-up. For parent-reported externalizing problems, a significantly greater proportion of non-Hispanic White children in the NBP than in the comparison condition moved from the borderline or clinical range at pretest to the normal range at posttest, χ² (1) = 4.64, p = .03,16.4% versus 9.3%. For child-reported internalizing problems, a marginally significantly greater proportion of non-Hispanic White children in the NBP than in the comparison condition moved from the borderline or clinical range at pretest to the normal range at 10-month follow-up, χ²(1) = 3.14, p = .08, 24.0% versus 14.8%. The percentage of children who shifted from the borderline or clinical range to the normal range are based on the full samples where either the parent or the children completed the pretest and 10-month follow-up assessments. The percentage of children in the borderline or clinical range at pretest, and thus were eligible to shift, did not differ significantly between the NBP and comparison condition for either child-report of internalizing problems (33.7% NBP, 40.2% comparison) or parent-report of externalizing (34.3% NBP,28.4% comparison).

DISCUSSION

This effectiveness trial tested whether the positive short-term effects of the NBP found in two efficacy trials could be obtained when the program was delivered by community agencies to a heterogeneous population of divorced and separated families. Significant moderated effects of the NBP to strengthen postdivorce parenting and to reduce children’s mental health problems were found, with ethnicity of the parents and age of the children being the primary significant moderators. The findings are discussed in terms of explanations for the subgroup differences in effects and implications for translating the NBP into a community service.

Similar to the two efficacy trials (Wolchik et al., 2000, 1993), the NBP had a significant effect to strengthen multiple aspects of positive parenting as reported by both parents and children. However, the effect sizes were in the small range, whereas they were in the medium range in the efficacy trials. Such decreases in effect sizes from efficacy to effectiveness trials are common when psychosocial interventions are implemented in community settings (Chambers, Glasgow, & Stange, 2013; Harvey & Gumport, 2015) and may be due to differences in the implementation of the intervention or characteristics of the participants. The fidelity of implementation in the current trial was quite high and comparable to that reported in the second efficacy trial (Wolchik et al., 2000). However, participant attendance was considerably lower in this trial (average percentage of sessions attended = 55.9%) as compared to the second efficacy trial (Wolchik et al., 2000; average percentage of sessions attended = 82.8%). In the current trial, 24.0% of those randomly assigned to the NBP never attended a single session and 44.7% attended four or fewer sessions. It is likely that effect sizes on parenting outcomes in the effectiveness trial were attenuated by families who attended no or very few sessions.

Although the NBP led to positive changes in discipline and monitoring for non-Hispanic White and Hispanic parents, the NBP improved two aspects of parenting (parent-report of parent–child relationship quality and rejection) and decreased child mental health problems at posttest and 10-month follow-up for non-Hispanic White parents only. In addition to the program effects being nonsignificant for the Hispanic parents, the effect sizes were very small for this group and often favored the comparison condition (see the footnotes for Tables 1 and 2). These differential effects may be due to lower attendance among Hispanic parents (Berkel et al., 2018) and a greater likelihood of dropping out after the first few sessions among Hispanic mothers (Mauricio et al., 2017). A second possible explanation is the lack of ethnic match between the Hispanic parents and their group leaders (Castro, Barrera, & Martinez, 2004). For Hispanic families, 76% in the NBP had a non-Hispanic group leader, whereas only 57% had a non-Hispanic group leader in the comparison condition. These possibilities are not independent; for example, it may be that lack of ethnic match with group leaders had a negative impact on attendance of the Hispanic families. Ongoing work is exploring these as possible explanations for the lack of program effects for Hispanic families. A third potential explanation for the lack of effects for Hispanic families is lack of measurement equivalence. However, findings from the test of measurement invariance across ethnicity in the assessment of parenting just reported makes this explanation unlikely. Research is also needed to investigate whether the failure of the NBP to impact parent–child relationship quality and rejection for Hispanic families might reflect cultural differences in the meaning of these constructs or a cultural mismatch with the intervention strategies.

The lack of Program × Parent Gender interaction effects on parenting or child mental health outcomes indicates that the program, which was originally tested with mothers, is also appropriate for fathers. The only indication of differential program effects across parent gender was a marginally significant Program × Parent Gender interaction effect on child-report of positive parenting. Tests of the simple main effects indicate a positive effect of the program to improve child-report of father positive parenting (b = 0.21, z = 2.32, p = .02, d = 0.33) but not mother positive parenting at posttest. Because divorced fathers have increasing parenting time over the past decades (Meyer, Cancian, & Cook, 2017) and prior interventions for divorced parents have focused on mothers (Forgatch & DeGarmo, 1999; Wolchik et al., 2000), there is a pressing need to understand the effects of preventive interventions for divorced fathers. A separate publication focuses on the father subsample and presents additional encouraging evidence of NBP effectiveness on father parenting and child mental health (Sandler et al., 2018).

For teacher-report there was a pattern of iatrogenic effects at posttest for children age 8 and younger on externalizing problems, task orientation, assertive social skills, and frustration tolerance, although there was a positive effect on frustration tolerance for the older children. In the efficacy trial (Wolchik et al., 2000), there was also a negative effect of NBP on teacher-reported shy-anxious behavior at posttest. Concern about long-term negative effects of NBP is mitigated because in the current trial (as well as in the efficacy trial), none of the negative effects persisted at follow-up. In the current trial, the only marginally significant difference between the groups at 10-month follow-up (on social competence) favors the NBP.

It may be that the changes parents engaged in during the program were difficult for younger children to integrate immediately and led to some short-term distress on their part, which was manifested in their behavior at school. The fact that this effect was only reported by teachers may be because teacher ratings of child problems are only modestly correlated with those or parents or children (range = .09–.35, Mdn = .22 in this sample), indicating that teachers provide a unique perspective on child behavior problems.

The use of an active control condition is an important strength of this study. Other research has found that although there were positive effects of prevention programs when compared to passive controls, there are often no significant effects when interventions are compared to active controls (Merry et al., 2011). Given that the active control condition and NBP both included didactic presentation of material on the risk and protective factors, the positive effects of the program in the current trial may be due either to the dosage (10-session NBP vs. two-session control) or to the active components of the NBP (i.e., home practice of program skills rather than simply raising awareness of these skills or providing positive attention to parents). Consistent with the latter explanation, Berkel et al. (2018) found that effective home practice of program skills significantly predicted improvements in parenting in this sample.

Two limitations of this study need to be noted. First, there were significant Condition × Posttest Attrition Status interaction effects on child externalizing and total problems at pretest, which potentially threatens the internal validity of the findings at posttest. However, using full information maximum likelihood estimation (which includes cases with missing scores) and including baseline status on the outcome as a covariate in the model mitigate bias from this threat to internal validity (West & Sagarin, 2000). Program effects at the 10-month follow-up are less likely to be affected by this threat to internal validity. Second, it is not known whether the program effects generalize to other ethnic or demographic subgroups that were not tested. Although the current sample was diverse in terms of ethnicity (30% ethnic minority), age of child (3–18), and gender of parent, further research is needed to test program effects on other subgroups (e.g., African American and Asian families).

The findings from this effectiveness trial are encouraging of further work to translate the NBP from an experimental prototype to a community service. Similar to others (Chambers et al., 2013), we see the translation from efficacy to sustainable community service as an ongoing process, guided by feedback from implementation in community settings. Three aspects of the feedback from the current trial will be critical for ongoing implementation of the NBP. First, the partnership with the family courts allowed the program to be presented to a large number of divorced and separated families. Building on this partnership is a promising direction for future implementation of the NBP. Second, the low level of enrollment and attendance, which is typical of implementation of parenting programs in the community, indicates the need for research on ways to increase the uptake of this program. Third, the lack of program effects on Hispanic children’s outcomes indicates the need for further study on how to strengthen program effects with this subgroup and to study the NBP’s effects in other ethnic groups. This research should be conducted within the framework of an ongoing collaboration with community providers with the goal of constantly learning how the NBP can best be used to improve outcomes for children following parental separation and divorce.

APPENDIX A

TABLE A1.

Baseline Characteristics for Full Sample and by Condition

Baseline Characteristic	Total	Comparison	NBP	Difference Across Conditions
Overnights in Past 30 Days, M (SD)	19.30 (9.64)	18.91 (9.63)	19.64 (9.64)	t(827) = −1.09, p = .28
Parent Report of Parenting, M (SD)
Relationship Quality	0.00 (0.91)	0.01 (0.90)	−0.01 (0.92)	t(828) = 0.33, p = .74
Rejection	1.52 (0.37)	1.51 (0.38)	1.53 (0.36)	t(824) = −0.74, p = .46
Discipline	0.00 (0.89)	0.03 (0.85)	−0.03 (0.93)	t(828) = 1.02, p = .31
Monitoring	4.36 (0.67)	4.35 (0.65)	4.36 (0.69)	t(738) = −0.12, p = .90
Child Report of Positive Parenting, M (SD)	0.00 (0.74)	−0.10 (0.75)	0.09 (0.72)	t(510) = −2.84, p < .01
Children’s Perception of Interparental Conflict, M (SD)
Parent Report	1.52 (0.47)	1.52 (0.47)	1.52 (0.47)	t(824) = 0.10, p = .92
Child Report	1.65 (0.44)	1.71 (0.44)	1.60 (0.43)	t(501) = 2.96, p < .01
Child Report of Caught in the Middle, M (SD)	1.92 (0.63)	1.95 (0.63)	1.88 (0.63)	t(499) = 1.26, p = .21
Child Report of Badmouthing, M (SD)	1.59 (0.68)	1.66 (0.67)	1.53 (0.69)	t(502) = 2.06, p = .04
Parent Report of CBCL/Pre-CBCL, M (SD)
Internalizing Problems	55.69 (10.74)	55.24 (10.82)	56.08 (10.66)	t(827) = −1.11, p = .27
Externalizing Problems	54.26 (10.00)	54.00 (10.26)	54.49 (9.79)	t(827) = −0.70, p = .48
Total Problems	55.19 (10.64)	54.77 (10.83)	55.55 (10.47)	t(827) = −1.05, p = .30
Child Report of BPM, M (SD)
Internalizing Problems	57.30 (6.83)	57.63 (6.85)	57.01 (6.82)	t(502) = 1.01, p = .31
Externalizing Problems	54.37 (5.95)	54.63 (5.88)	54.14 (6.01)	t(505) = 0.92, p = .36
Total Problems	57.45 (7.02)	57.78 (6.93)	57.15 (7.11)	t(504) = 1.00, p = .32
Teacher Report of BPM, M (SD)
Internalizing Problems	55.44 (6.24)	55.39 (6.16)	55.47 (6.34)	t(457) = −0.14, p = .89
Externalizing Problems	53.17 (5.78)	52.75 (5.39)	53.57 (6.12)	t(459) = −1.51, p = .13
Total Problems	54.94 (6.14)	54.64 (5.88)	55.23 (6.38)	t(457) = −1.03, p = .30
Teacher Report of Developmental Competencies, M (SD)
Learning Problems	1.95 (1.05)	1.97 (1.04)	1.94 (1.06)	t(459) = 0.35, p = .72
Task Orientation	3.61 (1.03)	3.59 (1.01)	3.63 (1.04)	t(459) = −0.43, p = .67
Frustration Tolerance	3.91 (0.82)	3.93 (0.79)	3.90 (0.85)	t(456) = 0.35, p = .73
Assertive Social Skills	3.59 (0.89)	3.52 (0.87)	3.66 (0.90)	t(450) = −1.66, p = .10
Social Competence	4.51 (1.12)	4.51 (1.08)	4.52 (1.15)	t(455) = −0.10, p = .92

Note: NBP = New Beginnings Program; CBCL = Child Behavior Checklist; Pre = Preschool; BPM = Brief Problem Monitor.

APPENDIX B

TABLE B1.

Means and (Standard Deviations) by Condition

	Pretest		Posttest		Follow-Up
Continuous Outcome	Comparison	NBP	Comparison	NBP	Comparison	NBP
Parent Report of Parenting		−0.01 (0.92)	−0.09 (0.96)	0.08 (0.85)		0.03 (0.91)
Relationship Quality Rejection	0.01 (0.90)	1.53 (0.36)	1.43 (0.30)	1.42 (0.31)	−0.04 (0.91)	1.42 (0.33)
	1.51 (0.38)	−0.03 (0.93)	−0.07 (0.88)	0.06 (0.92)	1.41 (0.30)	0.02 (0.85)
Discipline M	0.03 (0.85)	4.36 (0.69)	4.36 (0.71)	4.43 (0.68)	−0.02 (0.87)	4.41 (0.68)
onitoring	4.35 (0.65)				4.27 (0.84)
Child Report of Positive Parenting	−0.10 (0.75)	0.09 (0.72)	−0.13 (0.82)	0.12 (0.70)	−0.13 (0.92)	0.14 (0.76)
Children’s Perception of Interparental Conflict
Parent Report	1.52 (0.47)	1.52 (0.47)	1.41 (0.39)	1.35 (0.37)	1.30 (0.36)	1.30 (0.36)
Child Report	1.71 (0.44)	1.60 (0.43)	1.56 (0.42)	1.46 (0.44)	1.46 (0.41)	1.39 (0.38)
Child Report of Caught in the Middle	1.95 (0.63)	1.88 (0.63)	1.91 (0.62)	1.82 (0.63)	1.87 (0.63)	1.79 (0.61)
Child Report of Badmouthing	1.66 (0.67)	1.53 (0.69)	1.52 (0.63)	1.43 (0.62)	1.57 (0.69)	1.44 (0.58)
Parent Report of CBCL/Pre-CBCL
Internalizing Problems	55.24 (10.82)	56.08 (10.66)	53.92 (11.02)	52.34 (10.96)	52.76 (10.69)	52.27 (11.62)
Externalizing Problems	54.00 (10.26)	54.49 (9.79)	52.97 (10.43)	51.63 (10.06)	51.42 (9.88)	51.31 (10.81)
Total Problems	54.77 (10.83)	55.55 (10.47)	53.57 (11.04)	51.75 (10.80)	51.78 (10.79)	51.71 (11.56)
Child Report of BPM
Internalizing Problems	57.63 (6.85)	57.01 (6.82)	56.36 (6.82)	55.84 (6.46)	57.12 (7.35)	55.52 (7.08)
Externalizing Problems	54.63 (5.88)	54.14 (6.01)	55.11 (6.13)	54.40 (5.93)	54.27 (5.97)	53.64 (5.62)
Total Problems	57.78 (6.93)	57.15 (7.11)	57.33 (7.02)	56.66 (7.13)	57.62 (7.27)	56.14 (7.08)
Teacher Report of BPM
Internalizing Problems	55.39 (6.16)	55.47 (6.34)	55.70 (6.49)	55.49 (6.32)	55.11 (5.63)	54.63 (5.53)
Externalizing Problems	52.75 (5.39)	53.57 (6.12)	53.76 (5.99)	54.01 (6.21)	53.51 (5.92)	53.59 (5.66)
Total Problems	54.64 (5.88)	55.23 (6.38)	55.44 (6.13)	55.51 (6.56)	55.18 (5.85)	55.20 (5.97)
Teacher Report of Developmental Competencies
Learning Problems	1.97 (1.04)	1.94 (1.06)	2.07 (1.12)	2.06 (1.14)	2.10 (1.10)	2.04 (1.09)
Task Orientation	3.59 (1.01)	3.63 (1.04)	3.50 (1.08)	3.54 (1.12)	3.42 (1.03)	3.51 (1.14)
Frustration Tolerance	3.93 (0.79)	3.90 (0.85)	3.87 (0.82)	3.91 (0.88)	3.84 (0.80)	3.90 (0.85)
Assertive Social Skills	3.52 (0.87)	3.66 (0.90)	3.58 (0.87)	3.71 (0.86)	3.51 (0.83)	3.70 (0.83)
Social Competence	4.51 (1.08)	4.52 (1.15)	4.46 (1.09)	4.57 (1.17)	4.30 (1.12)	4.53 (1.11)

APPENDIX C

Figures Depicting Moderated Program Effects

FIGURE C1.

Adjusted means by parent ethnicity for parent reported parent-child relationship quality at posttest. Note. The asterisk denotes a significant mean difference (p ≤ .10) between the New Beginnings Program (NBP) and comparison conditions.

FIGURE C2.

Adjusted means by parent ethnicity for parent reported rejection at posttest. Note. The asterisk denotes a significant mean difference (p ≤ .10) between the New Beginnings Program (NBP) and comparison conditions.

FIGURE C3.

Adjusted means by parent ethnicity for parent reported internalizing problems at 10-month follow-up.Note. The asterisk denotes a significant mean difference (p ≤ .10) between the New Beginnings Program (NBP) and comparison conditions.

FIGURE C4.

Adjusted means by parent ethnicity for child reported internalizing problems at 10-month follow-up.Note. The asterisk denotes a significant mean difference (p ≤ .10) between the New Beginnings Program (NBP) and comparison conditions.

FIGURE C5.

Adjusted means across child age for parent reported monitoring at 10-month follow-up.Note. The vertical lines and arrows indicate regions of significance (p ≤ .10).