Reliability, Construct Validity, and Response- Set Bias of the Revised Learning-Style Inventory (LSI-1985)

Abstract

This research investigated the internal consistency and construct validity of the revised Learning-Style Inventory (LSI-1985). In particular, the study examined the possibility of a response-set bias based on the single-scale-per-column format of the instrument. Order effects were considered as well. To study the possible response set, two versions of the instrument were administered: (a) the standard LSI-1985 and (b) a scrambled version. Estimates of scale consistency were less in the scrambled version suggesting the presence of a response set for the standard LSI-1985. Factor analyses failed to support the construct validity of the revised instrument and also seemed to indicate the presence of a response set. In addition, completing the LSI after the administration of a second instrument seemed to affect the reliabilities of some scales. Further research is needed to clarify the effects of the psychometric properties and underlying assumptions of the LSI-1985.