2012 · Study · English
The Application of Meta-analysis within a Matched-pair Randomized Control Trial: An Illustration Testing the Effects of Communities That Care on Delinquent Behavior
Monahan, K. C., Hawkins, J. D., & Abbott, R. D.
Monahan, K. C., Hawkins, J. D., & Abbott, R. D. (2012). The Application of Meta-analysis within a Matched-pair Randomized Control Trial: An Illustration Testing the Effects of Communities That Care on Delinquent Behavior. Prevention Science, 14(1), 1–12. doi.org/10.1007/s11121-012-0298-x
Summary
Methodological article in Prevention Science proposing meta-analysis as a complement to conventional design-based analysis in matched-pair randomised trials, using CTC as the worked example. In the 24-community trial, the authors estimate the effect of CTC on eighth-grade delinquent behaviour separately within each matched pair, pool the pair-level effect sizes into an overall estimate and test for heterogeneity across pairs. The pooled result confirms that CTC reduced delinquent behaviour, and the heterogeneity test shows that the effect varied significantly across communities. The point of the approach is that it makes the variation visible and opens the way to identifying which features of communities or coalitions go with larger or smaller effects – something a single trial-wide estimate hides. For anyone planning a paired-community evaluation of CTC in Europe, this is the analytic template the Seattle group used to study why the same system worked better in some places than in others.