The purpose of the study was to specify conceptions of "effective teaching" in terms conducive to the future development of a generally accepted, reliable, and valid system of instructional evaluation. Multiple regression was used to model individual conceptions held by male faculty (N=4) and male undergraduates (N=4) at Rice University. Faculty and student judges reviewed and rated profiles of 1 hypothetical instructors. The profiles consisted of a course subject matter designation and seven quantified cues referring to the instructors' performances on the following dimensions: lecture and/or presentation style (LECT), general rapport with students (RAPR), amount of information imparted in the course (INFO), arousal of student interest (AROU), clarity of course requirements and grading procedures (PROC), intellectual demand of the course (DEMD), and instructor's general knowledge of the field (KNOW). The judgmental policies of eight participants varied according to the subject matter designation. However, they did not vary in any normative or systematic manner. The non-configural raters (N=72) were included in a factorial analysis of group-related (i.e., Status X Discipline) differences in judgmental policy. The relative importance of the content (INFO, DEMD, KNOW) to style (LECT, RAPR, AROU, PROC) dimensions was greater for faculty judges. There was no evidence that policies are related to the raters' respective academic disciplines. Considering all raters, INFO received the highest average weighting, followed by AROU, LECT and KNOW, RIGR and RAPR, and PROC dimensions. Four clusters of raters were identified by HIER-GRP (Human Resources Laboratory, USAF). The composition of each cluster was heterogeneous in terms of the status and academic discipline of the members. The policies characterizing the cluster memberships varied in two respects: (1) in the dimensions weighted most heavily, and (2) in the number of dimensions receiving substantial weight (i.e., policy complexity). Modifications of conventional student rating scales were suggested in view of findings from the present study.