| Abstract: | There is debate whether the social objectives met by banding can justify the loss in performance. To rationally assess this trade-off, one must know the magni- tude of performance lost. This article presents a mathematical model that allows computation of (a) the expected difference in performance between the top- ranked and bottom-ranked person in the band, and (b) the likelihood that the top-ranked person will actually outperform the bottom-ranked person, under various common testing situations. The model shows that, in most situations, a small loss in average performance may be expected, and that a person at the top of the band is only slightly more likely to actually outperform a person at the bottom of the band. The results support the use of banding and selection based on secondary criteria and generally show that the social gains of banding may be greater than the economic cost. |
