A Bayesian analysis of employment interviewers' errors in processing positive and negative information

It is a pervasive finding in the interviewing literature that unfavorable information has more influence upon interviewer's judgements than favorable information. Hollmann (1971) has proposed that interviewers process unfavorable information accurately while processing favorable information in an underweighted fashion. Due to several methodological shortcomings in Hollmann's design, a reinvestigation of his hypothesis was conducted using a Bayesian information processing paradigm. Possible interviewer information processing errors may be conceived as errors of misperception or misaggregation in the Bayesian framework. The error of misperception involves a misvaluing of the information value of an information item by the interviewer. An error of misaggregation involves inaccuracy in combining items of information. Hollmanns' hypothesis is a differential misaggregation hypothesis. The present study investigated the differential misaggrogation hypothesis in a student sample and a professional interviewer sample. Subjects revised probability of success estimates for hypothetical job applicants based upon items of favorable and unfavorable information. Probability revisions were compared to those dictated by the normative model and the relative accuracy of aggregation for favorable and unfavorable items of information was determined. In both subject groups, favorable and unfavorable infojmation was processed conservatively. In the student group, where differential aggregation accuracy occured it was in the opposite direction from that proposed by Hollmann, Favorable information was aggregated more accurately than unfavorable, There was no differential accuracy effect in the professional sample. It was argued that effects found in the student group could be attributed to a general student leniency effect and that an explanation of the dynamics of the negativity bias probably lies in a misperception hypothesis.