"The Impact Of CEO Turnover On Security Analysts' Forecast Accuracy" - A Comment

A recent Journal of Applied Business Research article by Sheikholesami, Wilson and Slevin (1998) examined the accuracy of security analysts' earnings forecasts for CEO change firms relative to a control group. The authors applied ANOVA on Value Line percentage forecast error measures and found "marginally significant" results indicating "that precision improved more for CEO change firms than for control firms." Doran (1998) tests for superior methods when scrutinizing forecast error. He finds percentage forecast error data to be severely non-normal, and demonstrates that nonparametric tests based upon ranks are superior to parametric methods. If analysts' earnings forecast precision actually improves more for CEO change firms, test results should be stronger using rank values instead of discrete percentage error measures.