Measuring the relationship between two questionnaires: a comment on Burt (2005)

Richardson, John T. E. (2007). Measuring the relationship between two questionnaires: a comment on Burt (2005). Assessment & Evaluation in Higher Education, 32(3) pp. 383–384.

DOI: https://doi.org/10.1080/02602930600896746

Abstract

In a series of publications, my colleagues and I have obtained scores from students in higher education on different questionnaires, and we have described the relationships among these scores using the statistic known as Wilks' Λ (lambda). Burt (2005) has criticized our use of this measure as inappropriate, arguing (1) that our account failed to acknowledge that Wilks' Λ is based on the concept of "generalized variance" as opposed to "additive variance"; (2) that we ought to have acknowledged the asymmetry of the situations discussed; (3) that there are several competing definitions of "proportion of shared variance", and (4) that Wilks' Λ is inappropriate in certain situations. This note sets out to clarify the position.