Conducting External Profile Analysis with Multiple Regression

Abstract

Internal profile analysis methods that employ multidimensional scaling or cluster analysis continue to receive application in educational and psychological research. One critique of these methods is that there is no guarantee that the identified profiles will exhibit criterion or diagnostic validity. Recently, an external profile analysis technique was developed using multiple regression. The external profile analysis method yields a profile that is derived to explicitly relate to a criterion. This paper introduces the new external profile analysis method and presents an application that uses 8th grade NAEP reading test scores as the criterion and background questions pertaining to students’ out-of-school reading and writing behaviors and experiences as predictors. The external profile analysis sought to uncover a type or pattern of out-of-school experiences that related to reading achievement. The profile analysis findings were also used to interpret group performance differences and to identify possible recommendations for closing the male reading achievement gap.