The study evaluated the invariance of the construct validity and thus the interpretation of Graduate Record Examinations (GRE) Aptitude Test scores. A systematic procedure for investigation of test bias from a construct validity frame of reference was developed and applied. Invariant construct validity was defined as similar patterns of loadings and equal units of measurement across four populations, as well as equal test score precision as defined by the standard error of measurement. The advantage of investigating psychometric bias at the item type level was to restrict use of a biased item type. The study also evaluated the factor structure of the subtests--Verbal, Quantitative, and Analytical. The subpopulations in the study were black and white males, and black and white females. The results indicated the item type factors appeared to be measuring the same things in the same units with the same precision. There was no significant evidence of psychometric bias. The factorial contributions of the Analytical item types were investigated. Analysis results indicated that the three Analytical item types appeared to be varying functions of reading comprehension and quantitative ability. (DWH)