Table 1.

Mediation analysis of long-term program effects on child problems and competencies

PROGRAM and SAMPLE	CONDITIONS COMPARED IN MEDIATIONAL MODEL	PROGRAM EFFECTS ON MEDIATOR (S)	MEDIATOR EFFECTS ON OUTCOME (S)	TEST OF MEDIATION
Schoolchildren and their Families Project To facilitate child transition to school n = 100 families with child preparing to start kindergarten (mean age = 5) Programs included 16-week couples groups focused either on marital skills (Marital Intervention) or parenting skills Parenting Intervention) compared to control group receiving yearly parent consultation Cowan, Cowan, & Hemming, 2005 (in Cowan & Cowan, Ablow, & Johnson, 2005)	Marital intervention vs. Control	Posttest: Reduced mother and father authoritarian parenting Posttest: Reduced father authoritarian parenting 1 yr: Reduced marital conflict 1 yr: Reduced mother and father authoritarian parenting	1 yr: Mediated effects on children’s perceived academic and social competence 1 yr: Mediated effects on increased child academic achievement (Peabody Individual Achievement Test) 1 yr: Mediated effects on increased teacher-reported academic competence 2 yr: Mediated effects on increased social competence and higher academic achievement	Tested whether the program affected the mediators and whether the mediators were related to the outcome
	Parenting intervention vs. Control	Posttest: Reduced father authoritarian parenting	1 yr: Mediated effects on reduced child externalizing and internalizing problems
Early Risers “Skills for Success” program To prevent conduct disorder and aggression n=151 “at-risk” 1st graders with high levels of teacher-rated aggression Program included 3 components delivered to all families in intervention condition (i.e., Summer program, child group, family skills parent groups) and 2 components delivered as needed by families (i.e., School Support program, Family Support program) Bernat, August, Hektner, & Bloomquist, 2007	Early Risers vs. No-intervention control	3 yrs: Increased effective discipline 3 yrs: Improved social skills	6 yrs: Mediated to decreased Oppositional-Defiant symptoms 6 yrs: Mediated to decreased Oppositional-Defiant symptoms	Joint significance test and test of mediated effect (Baron & Kenny, 1986; Sobel, 1982; MacKinnon, 1994 method)
Coping Power To prevent aggression and behavior problems n = 183 5th grade boys rated as aggressive by teachers, primarily low income Program consisted of Coping Power child groups (8 sessions in first year and 20 sessions in second year) or Coping power child groups plus Parent Program (16 monthly groups). In analyses, 2 intervention conditions were combined and compared to a no-intervention control group. Lochman & Wells, 2002	Child Program with/without Parent Program vs. Control	Posttest: Improved parent and youth program targets (five-factor model consisting of parental inconsistent discipline, youth attributions, outcome expectations, internal control, and perceptions of others) Posttest: Improved parent and youth program targets (five-factor model, see above) Posttest: Improved parent and youth program targets (five-factor model, see above)	1 yr: Three variables mediated to self-reported youth delinquency: Parental inconsistent discipline, youth outcome expectations, and youth internal control 1 yr: Mediated to decreased youth reports of substance use (no individual mediator was significant) 1 yr: Mediated to improved teacher-rated school behavior (no individual mediator was significant)	Test of whether the program effect on outcome become nonsignificant when adjusted for mediator and mediator to outcome relations, did not cite any test of mediation
Project LIFT: Linking the Interest of Families and Teachers To reduce the risk of delinquency n = 361 5th graders in schools within neighborhoods with high rate of juvenile delinquency 10-week program included a classroom-based social skills and problem solving curriculum, playground behavior modification program, and group parent training sessions weekly for six weeks versus no interventions for school and $2000 in unrestricted funds Degarmo, Eddy, Reid, & Fetrow, 2009	LIFT program vs. Control	Posttest: Improved family interactions during observed problem solving task Posttest: Improved family problem solving and reduced observed playground aggressive behavior Posttest: Reduced observed playground aggressive behavior	7 yrs: Partially mediated reductions in average tobacco use 7 yrs: Partially mediated effects on growth of tobacco use over 7 years 7 yrs: Partially mediated effects on reductions in acceleration of illicit drug use	Path model, joint significance test of program on mediator and mediator on outcome and bootstrap test of mediated effect (MacKinnon & Dwyer, 1993; Preacher & Hayes, 2004)
Strong African American Families To prevent substance use in African American youths n = 667 African American children, age 11, from semi-rural Southern U.S. Program consisted of 7-session parent group and separate 7-session youth group compared to no-intervention control group Brody, Murry, Gerrard, Gibbons, Molgaard, McNair, Brown, et al., 2004; Brody, Kogan, Chen, & Murry, 2008	SAAF program vs. control	Posttest: Increased regulated, communicative parenting (variable consisted of involved-vigilant parenting, communication about sex, racial socialization, and expectations for alcohol use) Posttest: Improved scores on protective parenting index (consisted of regulated communicative parenting variables, see above) Posttest: Improved youth protective factor index (consisted of academic competence/engagement, self-esteem, future goals, risk attitudes)	Posttest: Mediated increases in youth protective factors (consisted of goal-directed future orientation, negative images of drinkers, negative attitudes about alcohol/sex, and acceptance of parental influence) (Brody et al., 2004) 7 years: Partially mediated reductions in conduct problems for high-risk youths with deviant peer group or low self control at pretest (Brody et al., 2008) 7 years: Partially mediated reductions in conduct problems for high-risk youth (Brody et al., 2008)	Baron & Kenny’s (1986) 4 conditions for mediation; test of joint significance (Brody et al., 2004) Baron and Kenny (1986)’s 4 conditions, joint significance and Freedman-Schatzkin test of mediated effect (Brody et al., 2008)
Iowa Strengthening Families Program To prevent substance use and other problem behaviors. n = 667 families of 6th graders from low-income rural areas, 238 assigned to ISFP and 208 to control Program consisted of video-adapted version of ISFP program that included 6 parent and 6 child sessions and 1 joint session compared to condition with mailed materials about substance use Spoth, Randall, & Shin, 2008	Iowa Strengthening Families Program Vs. Control	Multiple Linkage Model Posttest: Improved parenting competency (consisted of family rules, parental involvement of child in family activities and decisions, parent anger mangaement within the parent-child relationship, communication) ↓ 2 yrs: Improvements in parenting competency mediated increases in academic engagement	6 years: Academic engagement at 2 years mediated increases in academic success	Path model with joint significance test and test of mediated effect
Familias Unidas To prevent substance use and risky sexual behavior for Hispanic adolescents n=213 Hispanic 8th graders rated as at least 1 SD above nonclinical mean on Revised Behavior Problems Checklist by parents Program consisted of 9 2-hour parent group sessions and 10 1-hour family visits plus 1-hour booster sessions at 10, 16, 22, and 28 months after the program compared to no-intervention control group Pantin, Prado, Lopez et al., 2009	Familias Unidas vs. Control	Posttest: Improved family functioning (consisted of positive parenting, parent/teen communication, parental monitoring)	30 mo: Partially mediated to growth trajectory of past 30-day substance use across the 30 month period	Joint significance test and test of whether program effect became nonsignificant after adding mediator
Familias Unidas + Preadolescent Training for HIV Prevention (PATH) To prevent substance use and risky sexual behavior for Hispanic adolescents N=268 Hispanic 8th graders and their primary caregivers Familias Unidas program consisted of 9 2-hour parent group sessions and 10 1-hour family visits. PATH program consisted of 4 3-hour parent education sessions and one parent-child session. Two programs together were compared to PATH + English for Speakers of Other Languages (ESOL) classes for parents or ESOL classes + HeartPower! For Hispanics program (educates parents about adolescent cardiovascular health). Prado, Pantin et al., 2007	Familias Unidas + PATH vs. ESOL + PATH	Posttest: Improved family functioning: (positive parenting, parent-teen communication)	36 mo: Partially mediated to past 90-day cigarette use	Controlled for each mediator in predicting outcomes and determined that paths reduced to nonsignificance were mediated paths. No test of mediated effect
	Familias Unidas + PATH vs. ESOL+HeartPower!	Posttest: Improved family functioning: (positive parenting, parent-teen communication)	36 mo: Partially mediated to past 90-day cigarette use and past 90-day drug use
New Beginnings Program To prevent mental health problems in children who experienced parental divorce n=240 children from divorced families, ages 9–12, Primarily non-Hispanic Caucasian Program consisted of 11-session mother group with or without 11-session child group. Intervention conditions were combined and compared to control group that received literature about children and divorce. Zhou, Sandler, Millsap, Wolchik, & Dawson-McClure, 2008; Bonds, Wolchik, Winslow, Tein, Sandler, & Millsap, in press	Mother program with/without child program vs. Control	Posttest: Improved maternal discipline, n/s for improved mother-child relationship quality Posttest: Improved mother-child relationship quality, n/s for improved discipline	6 yrs: Mediated effects on grade point average 6 yrs: For high-risk group(externalizing problems and family adversity at baseline), partially mediated effects on externalizing, internalizing and symptoms of mental disorders (especially adolescent-reports) (Zhou et al., 2008)	Tested mediated effect using confidence intervals (MacKinnon, Lockwood et al., 2002; MacKinnon et al., 2004)
		Multiple Linkage Model Posttest: Increased mother-child relationship quality ↓ 3 mo and 6 mo: Led to decreased child internalizing problems Multiple Linkage Model Posttest: Increased effective discipline ↓ 3 mo, 6 mo: Led to decreased child externalizing problems	6 yr: Led to increased self-esteem and decreased internalizing symptoms 6 yr: Led to less substance use, better academic performance (Bonds et al., in press)	Multiple linkage model, did not specifically test mediation
Parenting Through Change To prevent mental health problems for youth who experienced parental divorce n = 238 divorced/separated mothers of boys ages 6–9, majority Caucasian Program consisted of 16-session parent management training (later condensed to 14 sessions) compared to no-intervention control group Fogatch & DeGarmo 1999; DeGarmo, Patterson, & Forgatch, 2004; DeGarmo & Forgatch, 2005; Forgatch, Patterson, DeGarmo, & Beldavs, 2009	Parenting Through Change vs. control	6 mo: Increased positive parenting (consisted of positive involvement, skills encouragement) 1 yr: Improved positive parenting 1 yr: decreased maternal and child depression 1 yr: Improved positive parenting 1 yr: Improved positive parenting	1 yr: Mediated effects on child, mother, and teacher-reported adjustment (Fogatch & DeGarmo 1999) 30 mo: Mediated effects on child problem behaviors (DeGarmo, Patterson, & Forgatch, 2004) 30 mo: Mediated effects on child problem behaviors (DeGarmo, Patterson, & Forgatch, 2004) 36 mo: Mediated effects to reduced delinquency (DeGarmo & Forgatch, 2005) 9 yr: Mediated effect on decreased adolescent delinquency (Forgatch et al., 2009)	Path model, no explicit mention of mediation (Forgatch & DeGarmo, 1999) Tested joint significance of program to mediator and mediator to outcome (DeGarmo, Patterson, & Forgatch, 2004) Tested whether the program effect was not significant when mediator was added (DeGarmo & Forgatch, 2005) Tested whether the program effect was not significant when mediator was added (Forgatch, et al., 2009)
		Multiple Linkage Model: 1 yr: Improved positive parenting ↓ 8 yr: Led to decreased association with deviant peers across 8 years	9 yr: To decreased delinquency (Forgatch et al., 2009)