The Compelling Need for Diversity in Higher Education
EXPERT REPORT OF PATRICIA GURIN

Gratz, et al. v. Bollinger, et al., No. 97-75321 (E.D. Mich.)
Grutter, et al. v. Bollinger, et al., No. 97-75928 (E.D. Mich.)

1. In each of the analyses I used common standards for judging the statistical significance of findings. Statistical significance is an approach that is used to judge the reliability of relationships in order to reduce the possibility that observed findings are simply due to chance. For the analyses based on total or white student samples, I use a probability level of .05 (5%) as the criterion for judging a finding as significant. This indicates that there is less than 1 in 20 chance that any relationship of the magnitude reported is simply due to chance, and is indicated by the notation p<.05. Since probability levels are related to sample size, I used a slightly different criterion for the smaller samples of African American and Latino students, p<.10. In other words, while there is always a slight chance that any individual finding is illusionary, we can be relatively confident any significant finding truly exists and is important in a statistical sense.

2. White students composed 85 percent of the students in the IRGCC study, and thus the findings from this study are included when I discuss white students. The total number of students in the study is soo small to analyze date separately for white students and students of color.

3. The MSS analyses do not include Latino/a students because their numbers at the University of Michigan are not large enough to permit reliable results in the regression analyses.