ERIC Number: ED147276
Record Type: RIE
Publication Date: 1976-Jun-6
Reference Count: 0
Relationships Between Pupil Achievement and Characteristics of Observed Teacher Behavior Distributions. Research Report No. 6.
Calkins, Dick S.; And Others
This study attempts to discover relationships between teacher classroom behaviors and pupil achievement. Observers rated thirty-two fifth-level teachers on the Global Rating Scales (GRS) and the Teacher Practices Observation Record (TPOR). The data were analyzed using a linear model in which classroom mean posttest math score was the criterion and predictors were pretest classroom mean math score, a teacher behavior score (mean), the teacher behavior score squared, the variance of the teacher behavior variable, and the variance of the teacher behavior variable squared. Results indicated that only a very small number of the teacher observation variables accounted for a significant amount of variance in math achievement, probably within the boundaries of what is expected on the basis of chance. The lack of positive results can conceivably be due to one of five reasons: (1) teacher behavior is not related to pupil achievement; (2) the scaling method or (3) type of analysis is not sufficient for discovering such behavior relationships; (4) some teacher behaviors are related, but none were sampled in the investigation; or (5) the relationship is of a form in which pubil characteristics interact with teacher behaviors. Retrospectively, it was not surprising that the use of variance as a predictor of achievement produced fewer statistically significant results than the mean even though it was reasonable to expect that all were due to chance, because the samplinq distribution of variance is not normally distributed. In addition, the scalinq procedure utilized in this investigation for obtaining teacher means for the TPOR instrument was probably not conducive to the discovery of relationships between these variables and pupil achievement. The development of alternative scaling techniques that would produce normally distributed values for infrequently observed behaviors inherent in the TPOR and similar low inference instruments appears to be quite important when these variables are to be utilized in correlational investigations. (MJB)
Publication Type: Reports - Research
Education Level: N/A
Sponsor: National Inst. of Education (DHEW), Washington, DC.
Authoring Institution: Texas Univ., Austin. Research and Development Center for Teacher Education.