Study 1
Gross and colleagues (2009) used a quasi-experimental design to determine the impact of the Chicago Parent Program (CPP) on parenting and child behavior outcomes 1 year after program completion. Participants included 253 parents of 2- to 4-year-old children enrolled in seven daycare centers in Chicago, Ill. Daycare centers were included in the study if they a) had more than 90 percent of families eligible for low-income, childcare assistance; b) were licensed by the Department of Children and Family Services; c) provided full-day child care; d) enrolled at least sixty 2- to 4-year-old children; e) had space to hold the weekly parent group; and f) agreed to the study randomization. Before randomization, the daycare centers were matched on size, racial/ethnic composition, percentage of single-parent households, and median income. Daycare centers were randomized to either the CPP treatment group or a waitlist comparison group using stochastic matching, a method that uses aggregate information to create equivalent groups. The data was analyzed at the parent and child level.
Study participants were the English-speaking parents or legal guardians of children enrolled in the daycare center. Parents in the treatment group (n = 135) received the CPP intervention in addition to child care. Parents in the comparison group (n = 118) received child care as usual with no intervention services. Parents in the treatment group were mainly Black (51.9 percent) or Latino (37.0 percent), mothers (91.9 percent), and single parents (60.7 percent). Parents in the comparison group were mostly Black (67.0 percent) or Latino (28.0 percent), mothers (85.6 percent), and single parents (62.7 percent). There were no statistically significant baseline differences between treatment and comparison group participants on parent age, educational level, employment status, marital status, or child age. However, there were statistically significant baseline differences on parent race/ethnicity and child gender. The treatment group comprised more Hispanic parents (37 percent versus 28 percent) and fewer boy children (48.9 percent versus 63.6 percent) than the comparison group. These differences were controlled for in the outcome analysis.
Parents were assessed at baseline, post-intervention (3 months after baseline), 6 months post-intervention, and 1 year post-intervention. At each of the four assessment intervals, parents completed a set of questionnaires and were videotaped during semistructured play and cleanup sessions with their children. The play session began with 15 minutes of free play between the parent and child. After 15 minutes, a new bag of toys was brought in, and the parent and child played with the toys for 5 minutes. The parent was then told to instruct the child to clean up the toys, without the parent touching the toys.
The primary outcomes of interest were parent reports of child behavior problems and intensity, child aversive behavior during free play and cleanup, and parent use of corporal punishment. Parent discipline strategies (including five items for corporal punishment) were measured with the Parenting Questionnaire. Aversive child behavior was assessed based on observed frequencies of seven problem behaviors (child noncompliance, destructive behavior, physically negative behavior, crying, whining, yelling, and smart talk) coded separately from the videotaped play and cleanup sessions using the Dyadic Parent–Child Interactive Coding System-Revised. Child behavior problems were measured using the Eyberg Child Behavior Inventory, a 36-item survey with each item measured along two scales: the Intensity Scale, which assessed parent perception of the frequency of each child problem on a scale of 1 (the behavior never happens) to 7 (the behavior is always happening), and the Problem Scale, which indicated whether each of the child behaviors listed was considered by the parent to be a problem with the target child on a binary scale (scored as yes or no). The study used an intent-to-treat model and the 1-year assessment was analyzed with growth curve modeling. No subgroup analysis was conducted.
Study 2
Gross and colleagues (2012) used a quasi-experimental design to evaluate the impact of the CPP on parent discipline strategies and child behavior problems 1 year after program completion. Participants were 285 parents of 2- to 4-year-old children enrolled in eight childcare centers in Chicago, Ill. Daycare centers were included in the study if they a) had more than 90 percent of families eligible for low-income, childcare assistance; b) were licensed by the state; c) enrolled at least sixty 2- to 4-year-old children; d) had space to hold the weekly parent group; and e) agreed to the study randomization. All eight childcare centers acted as their own comparison groups for the first year of the study. Four centers were randomly chosen to begin the study in 2007 and the other four centers began the study in 2008.
Study participants were Spanish-speaking and bilingual parents or legal guardians of children enrolled in the daycare center. Parents in the treatment group (n = 151) attended CPP sessions at the childcare center; parents in the comparison group (n = 134) received only child care as usual. Study participants were mainly Black (55 percent) or Latino (45 percent), mothers (89 percent), unmarried (73 percent), and about 30.9 years old on average. There were no statistically significant differences between groups on these demographic characteristics at baseline. However, the baseline assessments indicated that parents in the comparison group were statistically significantly more likely to be employed full time and have a higher income than parents in the treatment group; therefore, these variables were controlled for in the analysis.
Parents were assessed at baseline, post-intervention (3–4 months after baseline), 6 months post-intervention, and 1 year post-intervention. Parent discipline strategies were measured with the Parenting Questionnaire, Corporal Punishment scale. Child behavior problems were measured using the Problem and Intensity scales of the Eyberg Child Behavior Inventory. The study used an intent-to-treat model, and the outcome data were analyzed using a multivariate repeated-measures analysis of variance. If multivariate effects were found to be statistically significant, then parallel univariate analyses were conducted to interpret the meaning of the overall multivariate effect including race/ethnicity as a covariate. Subgroup analyses were conducted on race and ethnicity.
Study 3
Breitenstein and colleagues (2012) pooled the data from Studies 1 and 2 (Gross et al. 2009; Gross et al. 2012), to examine the impact of the CPP on Black and Latino parents. The pooled parent and child data represented 504 families. The treatment group consisted of 267 parents who received the CPP. The comparison group consisted of 237 parents who did not receive the intervention. On average, parents in this combined sample were about 30.9 years old, mainly unmarried (72.2 percent), mostly mothers (90.0 percent), and mainly Black (57.7 percent) or Latino (42.3 percent). Two statistically significant differences were found on baseline demographics: the sample of children in Study 2 were slightly younger than children in Study 1, and across both studies Latino parents were more likely (38.4 percent) than Black parents (18.2 percent) to be married. These variables were controlled for in the analysis.
Outcomes were measured using teacher reports of child behavior problems measured with The Caregiver–Teacher Report Form. Two behavior problem scales were analyzed: the Externalizing Scale (34 items measuring disruptive behavior problems, aggression, and hyperactivity) and Internalizing Scale (32 items measuring anxiety, inhibition, depression, and social withdrawal). The study used an intent-to-treat model, and the outcome data were analyzed using a multivariate repeated-measures analysis of variance. If multivariate effects were found to be statistically significant, parallel univariate analyses were then conducted to interpret the meaning of the overall multivariate effect including race/ethnicity as a covariate. The authors conducted subgroup analyses on race and ethnicity.
