Black-White Differences in the Educational Attainment Process: What Have We Learned?

Differences in the status attainment process for groups are often 9 inferred from differences in corresponding regression coefficients when structural equations models are separately estim2ted for each group. This paper examines the credibility of inferences based on crossgroup compfirisons of regression coefficients using black-white differences in the educational attainment process as an example. It 'reviews evidence bout race differences from previous studies and finds inconsistencies. Reanalyses of the data from these studies using common models and methods fails to produce consistent results. Additional reason for caution in substantive interpretation of differences in regression coefficients is illustrated using NLS data in which a large portion of the subgroup differences in regression coefficients can easily be interpreted as due to black-white differences in the measurement properties of the observed variables. Other sources of regreinn slope filictuatior that may arise from methodological rather than substantive processes are also illustrated. Evidence implies that regression slope differences across groups in models of attainment provide ambiguous evidence on which to base statements about differences in the attainment process.