ERIC Number: ED229406
Record Type: RIE
Publication Date: 1983-Apr
Reference Count: 0
An Analysis of Bias in the Selection of School Outliers.
Connor, Ethel Green
A study was conducted to detect possible bias in the prediction of school achievement with family background variables and the effects of biased selection on the classifications of exceptional schools. Bias was examined by testing the equality of subpopulation regressions for predominately Black, Hispanic and White school subgroups. Significant intercept differences were detected among the subgroup regressions. The effectiveness indices obtained from a multiple regression analysis which failed to control subgroup differences were significantly different from those of a model which accounted for intercept differences. The average effectiveness index for the predominantly Black schools was significantly smaller when intercept differences were not controlled. The indices of Hispanic schools were significantly larger and those of White schools were unaffected. Failure to account for differences among subgroup regressions was found to alter the outcomes of studies aimed at identifying school outliers (schools in which the student enrollment performs significantly above or below expectation). (Author)
Publication Type: Speeches/Meeting Papers; Reports - Research
Education Level: N/A
Authoring Institution: N/A
Identifiers: FLorida Statewide Assessment Program; School Outliers
Note: Paper presented at the Annual Meeting of the American Educational Research Association (67th, Montreal, Quebec, April 11-15, 1983).