Gender Difference May Be a Factor in Teachers' Perceptions of Students

Abstract

In his reply (3), Prawat concludes that the analyses used are logically inappropriate, although not statistically inappropriate. The appropriateness of any statistical analysis follows from a logical evaluation of problems the analysis might create. I applaud the recognition of this, although elaboration of the logical versus statistical inappropriateness would have been interesting. The author minimizes the importance of the inappropriate analysis in the original conclusions. What is the basis of the conclusions? No statistical comparisons of relative importance are presented. The author says that conclusions were based on the size of standardized regression coefficients and effect magnitudes. Sex was a significant factor in three of six analyses. Only IQ was significant in more analyses (4). This certainly cannot be taken as evidence of the insignificance of sex as a factor. Prawat is critical of the Miller and Asher discussion of suppression. This brief discussion did not elaborate upon the two types of suppression that can be expected (1). Miller and Asher (2) contend that variables like grade and age gain significance on the his of their relationship to irrelevant variance in SORT (grade/SORT r = .81) rather than by predicting relevant variance in the dependent variables as Prawat suggests. The specifics of these suppression effects could not be discussed until the correlation matrix was made available. It is suggested that readers consult Conger (I), the correlation matrix, and the original article to draw condusions about suppression. Prawat ignores the main point of the criticism of his use of effect size; his measure of achievement was a measure of reading achievement while his teachers' perceptions were perceptions of general achievement. Of the eight items, only one perception item dealt with reading. Sex differences in reading are likely to be different from sex differences in mathematics, spelling, etc. Why didn't they compare sex differences in IQ (which is acrually general achievement) to perceptions of general achievement? Obviously the results would have been quite different. If we want to test whether perceptions of gender differences can be explained away by differences in reading ability, the appropriate method is ordered regression, testing to see whether sex remains significant after differences in ability are panialled out (with consideration for true scores). It appears, based upon their regression statistics, that sex does predict variance over and above that partialled out by reading ability. If teachers hold biases, the biases may begin with differences ia boys' and girls' behavior in school. Perhaps reading performance is the major determinant of percep tions of reading ability, which leads to more general perceptions. A biased attitude does not require that sex, race, origin, or whatever is the best predictor of attitudes. Wh~le the evidence is not well established, this author suspects that intelligence plays a blgger role in teachers' perceptions than sex. But if sex plays a role over and above ind~v~dual differences in ability, gender difference is an important factor in teachers' perceptions of students.