Analyzes two methods of testing group differences of a latent variable: group code analysis and structured means analysis. Describes these methods in terms of conceptual representation, unique underlying assumptions, and relative merits and limitations; points toward methodological extensions beyond this two-sample case. (RJM)

Responds to an article that explored heterosexist bias in seven widely used assessment instruments. Praises the article's authors for raising this issue and explores their recommendation that norms appropriate to lesbian, gay, and bisexual populations be developed. Argues that bias can pervade the whole assessment process--not just the testing…