“…This is espe cially important in the case of anorexia ner vosa because therapeutic approaches are widely divergent (e.g., pharmacological, learning theory oriented, psychoanalytic, system oriented, multifactorial) [Vandereycken and Meermann, 1984], direct therapy comparisons have hardly been feasible until now, and the numerous studies on prognos tic factors only classified the latter as posi tive, negative, or irrelevant [e.g., Crisp et al, 1977;Bemis, 1978;Steinhausen, 1979;Fichter, 1981a;Kohle and Simons, 1986] without establishing an empirically founded hierarchical order based on the relative im portance of the areas under investigation. This study attempts to make up for that defi ciency.…”