|| Checking for direct PDF access through Ovid
To date, African–American teacher candidates' certification rate has been significantly below that of non-Hispanic white candidates, and this differential has been consistent across all field tests and operational assessments. At issue is the source of the disparate impact observed. The article first discusses the distinction between disparate impact and bias in assessment. This is followed by a discussion of the five broad categories of sources of disparate impact (substantially differential rates of certification by subgroup): (1) demographic differences (such as candidates' years of teaching experience or the number of advanced degrees earned), (2) recruitment differences (the system of incentives to volunteer for certification and other factors related to recruitment) that may have resulted in samples of African–American and non-Hispanic white candidates who differ systematically in teaching performance, (3) contextual differences in the conditions under which candidates teach (for example, the level of administrative, technical, or collegial support or other contextual factors that distinguish urban schools (where the majority of African–American teachers teach) from rural and suburban schools, (4) biases or deficiencies in the assessment process itself (for example, a standard of practice that privileges certain forms of teaching over others, exercises that contain biases that either disadvantage African–American candidates or, advantage white candidates, and deficiencies in training materials and the training of scorers), differential and (5) actual differences in teaching performance that have resulted from unequal educational opportunity, discrimination, and other social and historical forces. The evidence collected and analyzed to date by the National Board's Technical Analysis Group is reported, and other studies and analyses planned for the future are mentioned.