An Application of Meta-Analysis and an Application for Bankruptcy Prediction Studies
1987, Organizational Behavior and Human Dec Proc
Peggy A. Hite
Attention is focused on the findings of prior research studies on the accuracy of predictions in bankruptcy studies. While studies have reported conflicting ratings of prediction accuracy, they also have put forth varying explanations for the variations in predictive ability. The Hunter-Schmidt method of meta-analysis is used to examine bankruptcy prediction studies from behavioral research. The method is a systematic, replicable approach that is used to accumulate results statistically across studies. The analysis technique is used first to examine whether the reported results from the studies were significantly different. Next, the results are examined to test if the difference was due to sampling error. Finally, the studies are categorized in order to test for differences due to a moderator variable. It is found that the various results do vary significantly. A source for this variation, the setting of priors, is identified.
Hite, Peggy A. (1987), "An Application of Meta-Analysis and an Application for Bankruptcy Prediction Studies," Organizational Behavior and Human Decision Processes, Vol. 39, March, pp. 155-161.