“…If indirect assessment procedures such as interviews, rating scales, or checklists are used (Durand & Crimmins, 1988;O'Neill et al, 1990;Willis, LaVigna, & Donnellan, 1993), care should be exercised to ensure that a wide range of establishing operations, antecedent conditions, and consequences are considered. Similarly, correlational or descriptive assessments based on direct observation of naturally occurring events (Bijou, Peterson, & Ault, 1968;Doss & Reichle, 1991;O'Neill et al, 1990;Touchette, MacDonald, & Langer, 1985) should indude a full complement of observation conditions lest a constrained set of observations identify a constrained set of functions (e.g., if observations do not indude situations in which difficult tasks are presented) (Sasso et al, 1992). Rigorous functional analyses (e.g., manipulations of controlling variables) in either analogue or natural settings may also need to occur with careful attention to the range of stimulus conditions that are presented.…”