PERCENTAGE DIFFERENCE VS. REGRESSION IN CONSTRUCTING APPLICATION-BLANK KEYS.
SYSTEM DEVELOPMENT CORP SANTA MONICA CALIF
Pagination or Media Count:
Application-blank scoring keys for two different employee groups and several different criteria were constructed by both multiple-regression and percentagedifference methods and compared. Multiple regression is unlikely to be superior to percentage difference item selection procedures for constructing application blank keys with small samples. However, with larger samples the regression item-selection procedure is still a candidate. Author