Response-shift bias: A source of contamination of self-report measures
Abstract
Investigated the effects of response-shift bias on the evaluation of a training program to improve the skills of interviewers engaged in selection interviews. 21 individuals, heterogeneous with respect to employer and geographic location, whose jobs include the responsibility for interviewing potential job candidates or subordinates, participated in a 5-day (40-hr) workshop on interviewing techniques. Self-report and performance measures demonstrated the effectiveness of the workshop. However, the results indicate that response-shift bias influenced Ss' estimates of training effects. Self-reported measures of change that used retrospective pretests to remove response-shift bias demonstrated significantly greater validity than measures of change that used traditional self-report pretests. (13 ref)