What if there were no psychometrics?: constructs, complexity, and measurement

McGrath (2005/this issue) has published a very significant article; it is significant because it examines the substantive issue of construct validity in a simple and informative way, noting the confusions apparent in how constructs and their variables are defined, used, and interpreted within the domain of personality and individual differences. This is a careful, thoughtful, article that quietly proceeds to outline the problems with current thinking and approaches to defining and using constructs in psychology. McGrath also suggests how some of these problems might be addressed. I note that McGrath appears to miss what might be considered an obvious rejoinder to some of his arguments, that is, the use of latent variable and item response theory. Some consideration is given to these arguments. However, a cursory examination shows that although these new methodologies offer many opportunities for new stochastic questionnaire data modeling and the construction of "instant" latent variables, the same problems caused by lack of attention to measurement and meaning remain. I find I am in agreement with many of the author's views and arguments, but I also find myself wondering whether modern psychometrics and individual differences research methods is now so dominated by psychological statisticians that any thought of substantive scientific innovation in this area that deals more properly with measurement and meaning is long gone. I think the answers to the question "What if there was no psychometrics?" would be most illuminating.