Abstract
Since the publication of Cronbach and Meehl's (1955) article on construct validity in terms of a nomological network, this methodological approach has been widely acknowledged by researchers in psychology. When Cronbach (1989) reviewed the evolution of the concept 30 years later, he suggested a realistic strategy for improving construct validation. During that time, the general trend of psychology had drastically changed into an information-processing approach. Embretson (1983) proposed a new method of construct validation based on cognitive theorizing, calling it “construct representation”. In the present article, following a discussion of the history of construct validation, the author reviews available methods for construct validation, including one proposed by Kyllonen (1996). A summary is presented of the present author's development of a neuropsychological diagnostic test for dementia based on working-memory theory and item-response theory. A new method of construct validation, called “a mutually compensatory method of construct validation”, is proposed that would effect a synergy of the psychometric and cognitive approaches.
