Meta-analysis of computer-aided psychotherapy: problems and partial solutions.

Meta-analysis can be valuable if it heeds its originators' caution that intimate communing with the data is essential. A critique of the authors' own meta-analysis shows that the danger of overly broad conclusions could be reduced by attention to specificities and awareness of potentially hidden sources of variance. Conclusions from even good meta-analyses are best placed in perspective, along with naturalistic reviews, open studies, and even anecdotes to yield a fair picture of what computer-aided psychotherapy or any other treatment can achieve under varying conditions. The most realistic picture comes from zooming in and out and melding meta-analyses with further types of evidence.