Practicing physical therapists seek the highest levels of evidence (ie, systematic review and meta-analysis) to inform clinical decision making. Performing a meta-analysis with assimilated data from a systematic review is important for comprehensive decision making in clinical and educational practice. In educational practice, the Pearson's product–moment correlation coefficient (r) is often used as the measure of effect size to examine causal relationships between variables. Our purpose was to present the best model for conducting a meta-analysis using r.
A meta-analysis is prescribed through five steps: (1) calculation of the effect sizes, (2) selecting the appropriate effects model and computational method, (3) applying the computational method, (4) performing heterogeneity analysis, and (5) assessing for publication bias.
Key statistical outcomes include: (1) total sample size, (2) number of correlations, (3) mean true score correlation, (4) variance of true score correlations, (5) 80% credibility interval, and (6) 95% confidence interval. To help with the interpretation of these results, figures (ie, forest plots and funnel plots) are used.
Central to achieving evidence-based practice is the selection and application of high-quality research. A systematic review with random-effects meta-analysis using high-quality cohort studies can provide high-quality evidence to physical therapy educators to answer their pertinent education questions.
Mitch Wolden is an associate professor in the Physical Therapy Program at the University of Jamestown, 4190 26th Avenue South, Fargo, ND 58104 (email@example.com). Please address all correspondence to Mitch Wolden.
Brent Hill is an associate professor in the School of Education at the North Dakota State University.
Sara Voorhees is an associate professor and program director in the Physical Therapy Program at the University of Jamestown.
The authors declare no conflicts of interest.
Received June 26, 2018
Accepted September 07, 2018