Some Implications of Parameter Sensitivity Research for Judgment Modeling in Accounting.

Abstract ABSTRACT: This article examines the issue of cue weighting in judgment models used to represent human information processors. Illustrative results are presented which show that predictions produced by linear models are very robust with respect to alternative cue-weighting schemes unless the number of predictors included in the models is small, the mean correlation among the predictors is low, and the dispersion of the weights is large relative to their mean. Implications of these results for judgment modeling research in accounting are discussed. Generally, these implications concern the potential improvement in human decision making that might obtain from the search for alternative predictors and/or measurement methods rather than the search for improved methods of weighting predictors in linear models.