This paper is about a causal model of role-taking recently suggested by BERTRAM and BERTRAM. The model tries to combine aspects of the cognitive-developmental approach as proposed by BRUNER, PIAGET, and WYGOTSKI,and symbolic interactionism as advocated by LINDESMITH and STRAUSS. While the selection of variables is handled rather carefully, the identification and testing procedures may be criticized in three respects: They are tautological because the same equations are used for identifying and testing the model, they are contradictory because identification procedures applicable to recursive models only are applied to a nonrecursive model, and they are fragmentary because only a few although the most important of the possible comparisons of implied and observed correlations are computed. Thus, some of the author’s major conclusions seem not to warranted by the rules of path analysis.