How do I use the data to perform a meta-analysis?

Meta-analyses and the PTSD Repository

Meta-analysis requires specific skills and expertise. For those familiar with these principles, data in the PTSD Repository can be downloaded into different formats--such as CSV, RDF, RSS, TSV, and XML--for analysis. Each dataset also has an API endpoint for access by common statistical packages (such as R, Python, SAS, Stata, and Tableau). For those who lack this training, we strongly suggest working with a statistician who has relevant experience.

Resources for meta-analyses

If you’re looking to learn more about how to conduct a meta-analysis, we recommend the following resources:
  • Dr. Pim Cuijpers, Professor Emeritus of Clinical Psychology at Vrije Universiteit Amsterdam and Director of the WHO Collaborating Centre for Research and Dissemination of Psychological Interventions, offers free online lectures, a downloadable practical guide for Meta-analyses in Psychological Health, and a hands-on guide for Doing Meta-Analyses with R.
  • Cochrane, an international network of researchers that perform meta-analyses and systematic reviews of health and health care evidence, offers information and resources to support researchers. Search both Cochrane and Cochrane Training for “meta-analysis” to learn more.

Single Treatment Study Effect Sizes

Others may just want to compare the PTSD treatment effect sizes from one study to another. In this case, we provide standardized effect sizes in the datasets. Within group effect sizes are an analog of Cohen’s d estimate of pre-post change for a specific treatment. Between group effect sizes are Hedge’s g estimates of the difference in change between the active treatment and the comparison condition.
Between-group estimates with a negative sign indicate a larger decrease in Arm A compared to Arm B, which corresponds to A outperforming B if the measure is scored such that higher scores indicate higher severity of symptoms. In within-group estimates, a negative sign indicates a lower score at endpoint compared to baseline, which corresponds to a pre-post decrease for the score.
It is important to understand that these within group and between group effect sizes for a given single study are not the same as effect sizes in a meta-analysis. In meta-analyses, the effect sizes of multiple studies are combined, and other factors are considered (such as sample size). Nevertheless, individual study effect size data can still provide a descriptive estimate of effectiveness.

Meta-analyses using Metapsy

For a subset of studies in the PTSD Repository–specifically psychotherapy vs control trials–the Metapsy website will conduct a meta-analysis for you at the click of a button.