Regression models of search time for color-coded information displays.

Abstract

Visual search time on color-coded information displays has previously been shown to depend upon the number of items displayed (density), the number of colors used (code size), and the number of items in the same color category as the target. The present paper employed regression analysis techniques to develop models of search time which explain the mode of operation of these factors. The initial model, based upon number of items per category as the effective parameter, was successful in explaining 58% of the variation in search time means. An improved model, which introduced the concept, “operative number of colors,” accounted for 84% of the variation. This model was able to postdict 82% of the actually obtained search time values within 0.76 s, with no postdiction in error by more than 2 s. The elevation in search times with larger display densities and code sizes observed in the data, and described by the improved model, was interpreted in terms of disruption of subjects' scan behavior on the more crowded displays.