Characterizing Spatial Ability: Different Mental Processes Reflected in Accuracy and Latency Scores.

Abstract

Abstract : Recent experimental studies have analyzed the time to perform tasks patterned after standard tests of spatial ability. Based on these analyses, information-processing models have been developed suggesting that subjects work through a sequence of component mental processes (e.g., code, transform, match) to perform spatial test items. If these models are correct, then response latencies, especially estimates of component-process durations, may be the best measures of spatial ability. By contrast, traditional psychometric analyses of these tasks have consistently used overall accuracy scores as measures of spatial ability. A model of the relationship between traditional accuracy measures of spatial ability and theoretically based latency measures is proposed. In this model overall accuracy and mean latency are viewed as composite scores consisting of the product (accuracy) or sum (latency) of component-process parameters. Three experiments investigated the relationship between spatial accuracy and latency scores, and established some psychometric properties (reliability, correlation across tests, predictive validity) of various measures. While accuracy and mean latency scores each proved to be reliable and consistent across different tests, the two measures were virtually independent. Further analyses using component-process latency scores suggest that different mental processes influence overall accuracy and mean latency. One hypothesis consistent with the data is that spatial accuracy scores reflect the ability to accurately code a pictorial stimulus, but mean latency scores on the same items reflect the ability to mentally transform the code. Implication for ability testing are discussed. (Author)