Software and Programs for Conducting Meta-Analysis Research: A Monte Carlo Investigation of Potential Differences.
Interim technical rept. Sep 91-Dec 94,
ARMSTRONG LAB BROOKS AFB TX HUMAN RESOURCES DIRECTORATE
Pagination or Media Count:
With the increased popularity of meta-analysis, a number of articles have questioned and highlighted the role that judgment calls play in the implementation and, ultimately, outcomes of studies that use this technique. In the absence of standardized data analyses packages, coupled with the wide range of options available to meta-analysts, the current study investigated the effect of choice of data analysis programs on meta-analysis study outcomes. The objective of this Monte Carlo study was to investigate the extent to which four commonly used Schmidt and Hunter validity generalization-based meta-analysis software programs, all based on the same conceptual and theoretical assumptions, produced identical outcomes when used to analyze the same dataset. The results indicate that while there were some differences in values obtained from the programs, these differences tended to be very small, typically occurring in the fourth and sometimes fifth decimal place, and did not influence the meta- analytic outcomes. Finally, differences in the features and capabilities of each of the programs are presented and discussed.
- Statistics and Probability
- Computer Programming and Software