Sixty-three studies (n approximately 3,600; 83 comparisons) of 69 behavioural and 14 non-behavioural programmes were included.
The mean validity score was 6.16 (SD=1.11) for behavioural studies and 4.64 (SD=1.15) for non-behavioural studies out of a maximum of 7 points. The quality ratings did not correlate significantly with study effect sizes.
Parent training using behavioural or non-behavioural programmes significantly improved the outcomes. In all cases the effect size was small to moderate and Qw indicated statistically significant heterogeneity. The findings were as follows: child behaviour, 0.42 (95% confidence interval, CI: 0.35, 0.49; Qw statistic, p=0.01; 68 comparisons); parent behaviour, 0.47 (95% CI: 0.40, 0.54; Qw statistic, p=0.01; 51 comparisons); parental perceptions, 0.53 (95% CI: 0.44, 0.61; Qw statistic, p=0.01; 48 comparisons). Behavioural and non-behavioural programmes did not differ significantly in their effects.
Moderators of child behaviour: significantly smaller effect sizes were found for disadvantaged children (p<0.01) and for studies with a high proportion of single parents (p<0.01). Effect sizes were greater for children with a high level of disruptive behaviour at baseline (p<0.05), children with high levels of ADHD symptomatology (p<0.01), and in studies using only individually delivered parent training. Other variables analysed did not significantly influence effect sizes.
Moderators of parental outcomes: significantly smaller effect sizes were found for parent behaviour (p<0.01) and perceptions (p<0.02) among studies of disadvantaged families (p<0.01). Parent-only programmes were associated with significantly larger effect sizes for parent behaviour and perception than were programmes including separate interventions for the child (p<0.01 and p<0.05, respectively) and for parent behaviour outcomes in multi-systemic interventions (p<0.03). The effect was also significantly greater for parents of children with ADHD (p<0.01).
Heterogeneity among the subgroups.
Among subgroup analyses of 15 or more studies, twelve had significant statistical heterogeneity. However, further stratification of these subgroups by socioeconomic status eliminated the heterogeneity in most cases.
Moderators of outcomes among disadvantaged families.
The only treatment variable that significantly improved the effect size for behavioural changes among disadvantaged children and also among parents was the mode of treatment delivery, with individual parent training being significantly more beneficial than group training (p<0.01).
There were few studies on the longer term effects of non-behavioural parent training and many studies did not employ a control group during follow-up. Among controlled studies, behavioural parent training had a statistically significant but small effect on outcomes, without significant heterogeneity: child behaviour, 0.21 (95% CI: 0.08, 0.33; 21 comparisons); parent behaviour, 0.25 (95% CI: 0.11, 0.40; 14 comparisons); parental perceptions, 0.45 (95% CI: 0.24, 0.65; 9 comparisons).