In this regard, Stringer (1999) defines action research as a three step spiral process of (1) planning which involves reconnaissance; (2) taking actions; and (3) fact-finding about the results of the action. The process alternates between action and critical reflection and which, in various stages, continuously refines its methods. While action research is most often applied to qualitative studies, it is possible to utilize it within a quantitative design such as will be done in the proposed research.
Because this study is only quasi-experimental in nature identifying variables as "dependent" or "independent" is not appropriate (see: Best & Kahn, 2002) due to the lack of controls inherent to the time-series design. However, it can be noted that the primary variables of this study are: (1) teaching strategies operationalized as the specific strategies used in Greenspan's (1998) Floor Time intervention for autistic children; (2) aggressive behavior operationalized as children's scores on the Aggressive Behavior Checklist (Shapiro, 2000); and (3) teacher feedback operationalized in terms of a researcher-designed instrument which collects teacher's ratings of each teaching strategy used in Greenspan's Floor Time intervention.