Judgment theory applied to the selection of life insurance salesmen

Abstract

The predictability of success of life insurance salesmen, and the judgment policies of 16 Agency Managers who select them, were investigated within the lens model paradigm. Each judge observed 200 profiles abstracted from the records of salesmen for whom 1-year, dichotomous criterion information was available. The profiles consisted of 64 cues, abstracted from personnel files, with ecological cue intercorrelations intact. An additional 160 cases were available for cross-validation. The criterion was moderately (and usefully) predictable from a cross-validated linear combination of the cues. There was relatively little agreement among the judges' decisions and considerable disagreement about the proportion of profiles judged acceptable. Judges were consistent but inaccurate in their predictions, substantially overusing scores on the Aptitude Index Battery. Unit weighting was superior to bootstrapping, but regression statistics were used to select the cues for unit weighting. Several cluster analytic techniques were tried, but none was really effective in differentiating preexisting groups. Accuracy of insight was generally poor and was not related to experience. Potential relationships between the judgment literature and two others, the interviewing and rating literatures, were explored.