Conceptual clarity in measurement—Constructs, composites, and causes: a commentary on Lee, Cadogan and Chamberlain

In an insightful and important article, Lee et al. (2013, this issue) clearly point out the problems with so-called formative measurement. In particular, they suggest that the MIMIC model formulation, as currently conceptualized, does not provide a solution. Their central thesis is that, in a MIMIC model, the supposedly formatively measured latent variable is empirically a reflective latent variable depending entirely on the endogenous variables included. They then look at composite variables as a possible solution. This commentary seeks to reinforce their central thesis, providing additional evidence and support. I also attempt to clarify the distinction between two types of models discussed in the article as MIMIC models. I then examine the use of composite variables, focusing on potential information loss and issues concerning conceptual clarity. I conclude that composite variables should not be routinely employed in theory testing research, and their use must be clearly justified.