| Abstract: | This article is concerned with the comparison of slope estimator precision in regression analysis and the structural relationships approach (e.g., Humak, 1983, ch. 3; Kendall and Stuart, 1977), as it is relevant for their applications when testing for initial value dependence in biomedical and behavioral contexts of repeated assessments (e.g., Blomqvist, 1977; Wall, 1977). As a basis for the comparison of the two methods, the mean square error is adopted. In the general case, it is argued for an informed (data-dependent) choice between regression analysis and the structural relationships approach. For the apparent majority of biomedical and behavioral studies of the phenomenon of initial value depenence, this comparison suggests that structural relationships is the preferable approach leading to more trustworthy substantive conclusions. |
