Lawley and Maxwell’s book on factor analysis legitimized the psychometric development of factor analysis as a “real” statistical model. Although most now praise them for their breakthrough in deriving maximum likelihood estimators for the model parameters, I think the following sequence of photos shows Lawley and Maxwell’s great insight and most important contribution, received at the time as general heresy by the high priests of factor analysis.
I read this book around 1975. The paragraph I underlined and typed on the cover is one that significantly altered my career: I learned then, and further in the early 1980s, that statistical theories of psychological processes, at their very best, are only weak approximations to reality. At the time lots of psychometricians were giving their professional lives to determine if one blindly empirical factor rotation method was better than another since after all they were different by 3 percent and everyone had just “discovered” Joreskog and Sorbom’s work on structural equation modeling. I never spent more than 50 percent of my research time between 1977 and 1984 on psychometrics and statistics — the rest of the time went into modeling adolescent drug abuse behaviors and their precursors. In 1984 it was time to move on to 90 percent of my work being devoted to real psychological and social issues.
As for statistical-psychology theories, the fact that
- it’s new
- it’s hot
- you can publish it a lot
does not sway me anymore.
Please click on the pictures to zoom.