Studying Manifest and Latent Profiles and Their Reliability and Validity
Cognitive profiles have been an item of debate within for several decades. Ever since the first conceptualization of intelligence testing, it was assumed that variations within profiles would yield clinically meaningful results. However, variability within cognitive profiles has not led to many clinical findings. There are three possible reasons for this. First, cognitive profiles may not contain statistically meaningful variance. Second, the individual subscale reliability and the profile reliability of cognitive profiles are not yet sufficiently high. Third, the methodology with which we assess profiles is currently inadequate to capture reliable, meaningful differences in cognitive profiles. This paper will assess various approaches to profile reliability, both at the subscore and profile level, in order to understand which is most sensitive to variance within cognitive profiles, and thus could subsequently be used to find clinically meaningful profiles. Additionally, this paper will address profile reliability, on both the manifest and latent levels. Manifest profiles may be of interest due to particular tools aimed at individual level analysis, and latent profiles may be more useful for discovering clinically relevant groupings or patterns. Methods of examining both manifest and latent profiles are explored.
Lipton, Abraham, "Studying Manifest and Latent Profiles and Their Reliability and Validity" (2015). ETD Collection for Fordham University. AAI10187158.