Because of their value in assessing many aspects of information systems (IS) productivity, the development and psychometric evaluation of scales which measure unobservable (latent) phenomenon continues to be an issue of high interest among researchers in the IS community. Typically, the measurement properties of developed scales are evaluated through traditional techniques such as item-to- total correlations, coefficient alpha, and exploratory factor analysis. While potentially useful in exploratorysituations,thesemetricsprovidelittleassessmentofscaleunidimensionality. Scaleswhichare unidimensionalmeasureasingletrait. Thispropertyisabasicassumptionofmeasurementtheoryandis absolutelyessentialforaccurate(unconfounded)measurementofvasableintenelationships. Inthispapet a paradigm for developing unidimensional scales Is presented. Drawing from well developed techniques within marketing research, education, aid psychology, this paradigm incorporates the yse of c o n f m a t o v factor analysis (CFA) as a means of assessing measurement properties. Importantly, CFA provides a stricter interpretation of unidimensionality than traditional methods and in many instances will lead to different conclusions regarding scale acceptability.
Segars, Albert and Globensky, Michel, "Scale Development in Information Systems Research: A Paradigm Incorporating Unidimensionality and Its Assessment" (1994). ICIS 1994 Proceedings. 55.