It was hypothesized that individual differences in observer attitudes might affect the way observers code teacher behavior. A questionnaire assessing educational attitudes was given to 22 subjects who were trained on 3 classroom observation systems: the Flanders Interaction Analysis, the Classroom Observation Scales, and the Classroom Observation Record. Subjects viewed videotapes of eight teachers and then rated each teacher on a Post-Coding Questionnaire. The questionnaire requested responses on a seven-point scale to the following items: (1) How well did you like the teacher? (2) How well did the teacher present the subject matter? (3) How well did the teacher interact with the students? and (4) How well did the students interact with the teacher? Correlations with these four factors found for the attitude survey confirmed the study hypothesis. It is suggested that observers perceive an effective teacher via general impressions, and that codings indirectly indicate these perceptions. Tables and references are included. (JD)