Graduate Admissions for a More Diverse Student Body Fall 2011 Joseph L. Brown, PhD Stanford University School of Humanities & Sciences Acceptable Practices for Graduate Admissions Consider all applicants in the same admissions pool Do an “individualized, holistic review” of each file to identify candidates qualified for admission Once a pool of qualified candidates is identified, then select admits from that pool based on each applicant’s contributions to the educational environment Diversity, broadly defined, can be considered, along with other criteria at each step of the process 2 1 What is diversity? Includes culture, socioeconomic status, race, gender, ethnicity, sexual orientation, disability, work and life experiences, skills, and interests The School particularly encourages departments and programs to seek applications from and to recruit – – – – Underrepresented minorities (URM): African Americans, Hispanic/Latino Americans and Native Americans Women in the sciences Low-socioeconomic backgrounds First generation college students A diverse student body will enhance the breadth, depth, and quality of our educational environment. 3 ETS Guidelines for Using GRE Scores Use multiple criteria – letters of recommendation, courses taken and their rigor, GPA, essays, GRE General and Subject Test scores Consider Verbal, Quantitative, and Analytical Writing scores as separate and independent measures Avoid decisions based on small score differences Do not use GRE cutoff scores Question predictive utility of scores for your own department: Conduct validity tests 4 2 Revised General GRE Began in August 2011; first score reports won’t be available until mid-November New verbal question types with increased focus on reading and analyzing text – – Elimination of analogies and antonyms Touted as benefit to students with diverse backgrounds New scoring range for verbal and quantitative sections – – From 200-800 to 130-170 in 1 pt. increments Reason? Tendency to make too much of small score differences with 200-800 range 5 Help on Interpreting Revised GRE Scores Handouts: – – – Tables 1A&1B, Percentile Rankings for Revised GRE Verbal, Quantitative, and Analytical Writing sections Table 1C, Verbal Reasoning Concordance Table Table 1D, Quantitative Reasoning Concordance Table For more on using Revised GRE scores, you can download THE GRE GUIDE TO THE USE OF SCORES from: http://www.ets.org/gre/institutions/ scores/interpret 6 3 Estimated Counts of Scores for GRE by Race for 2006-07 African Mexican Americans Americans Native Americans # who took exam in 06-07 29,432 8,447 1,978 Verbal: 600+ 353 (1.2%) 346 (4.1%) 142 (7.2%) Verbal: 700+ 29 (0.1%) 25 (0.3%) 14 (0.7%) Quant: 600+ 2502 (8.5%) 1596 (18.9%) 431 (21.8%) Quant: 700+ 471 (1.6%) 439 (5.2%) 125 (6.3%) 7 Causes of Group Differences in GRE Performance Socioeconomic Status (family income and first generation college status) – – An ETS study of high scorers (750-800) found 90% had fathers with graduate or professional degrees. Only 4% had fathers who had NOT completed high school (Penncock-Roman, 1994) 8 4 Causes of Group Differences in GRE Performance Racial and Gender Stereotypes – – – – The threat of confirming negative stereotypes places an additional evaluative pressure on minority test-takers and women taking math-oriented standardized exams (Steele & Aronson, 1995; Spencer, Steele & Quinn 1999) Impacts strongest academic performers, not the weakest Associated with increased load on working memory Stereotype threat constitutes a type of measurement bias that may complicate interpretation of GRE scores. 9 Group Differences in GRE Performance Use the tables in Appendix to consider within-group percentile rankings Consider the question, “if we have an applicant with a 650 verbal score, what is the likelihood we will see another applicant with a 650 verbal score or greater?” 10 5 GRE Summary GRE scores must be interpreted with care – Consider within-group percentile rankings in addition to national percentile rankings Understanding the limitations of the GRE tests frees you up to focus instead on the whole application You need not discard an applicant solely on the basis of a relatively low test score 11 The “Student in Context” Approach Consider all aspects of the application – past accomplishments, letters of recommendations, research statement, writing sample, GRE, GPA and contributions to diversity Consider multiple GPA measures – post-frosh GPA, GPA in major, GPA in courses relevant to field of grad study 12 6 Faculty Reviewer for Diversity Identify a faculty member to take another look at the diverse admissibles… Review applications (particularly letters of recommendation and essays) for evidence of opportunities available and extent to which applicant took advantage – – Who are the trailblazers in the pool? Was summer research the only opportunity available to the student? 13 Faculty Reviewer for Diversity Look through apps for traits associated with success that can’t be measured by the GRE – persistence, multitasking ability, creativity, intellectual curiosity, potential for productivity Report results back to full committee, make recommendations and draw attention to promising applications Full committee could consider applicants who are ‘on the bubble’ 14 7 The End Questions? Contact jlbrown@stanford.edu, GRE Scores How well do they predict performance in graduate school? How significant are small score differences such as 100 points? 16 8 GRE Scores and Graduate Performance ETS studies find that – – 12-25% of variation in 1st year grad school GPA can be explained by variation in GRE scores. Less variation explained in time to degree, faculty rankings of grad students or ratings of dissertation quality Thus 75% or more of variation in 1st year GPA is unexplained by the General GRE 17 Factors Influencing Success in Graduate School Reasoning Ability (General GRE) Motivation SUCCESS Creativity Interpersonal Skills Subject Matter Knowledge (Subject GRE) Financial Support Family Circumstances From ETS publication “What is the Value of the Graduate Record Examinations?” (available from www.gre.org) 18 9 Do Small Differences on GRE Scores Matter? IF the general GRE explains at best 25% of variation in first year GPA, leaving the lion’s share in variation unexplained…particularly when it comes to other performance measures such as time to degree or faculty rankings… THEN a gap of 100 GRE points between two otherwise equally qualified candidates may not be particularly meaningful AND students can buy 100 points on the GRE if they have $1500 for a good GRE prep course 19 Percentile Scores for GRE Verbal Test by Race GRE scores European American African Americans American Indians Mexican Americans 800 99.9 99.9 99.9 99.9 750 99.4 99.9 99.8 99.9 700 97.7 99.9 99.3 99.7 650 93.5 99.7 97.4 98.8 600 84.9 98.8 92.8 95.9 550 70.5 95.3 83.2 89.1 500 52.4 87.5 68.1 76.7 20 10 Percentile Scores for GRE Quantitative Test by Race GRE scores European American African Americans American Indians Mexican Americans 800 97.1 99.8 98.9 99.1 750 93.3 99.5 97.2 97.8 700 86.9 98.4 93.7 94.8 650 76.7 96.0 87.7 89.6 600 63.3 91.5 78.2 81.1 550 48.0 83.7 65.9 69.2 500 43.0 71.0 51.0 55.2 21 Percentile Scores for GRE Tests by Gender Verbal GRE scores Quantitative Men Women Men Women 800 99.6 99.8 93.0 98.0 750 98.6 99.2 86.2 95.4 700 96.0 97.6 76.4 90.5 650 90.3 93.9 64.1 82.6 600 80.2 86.7 50.0 71.6 550 65.5 75.5 36.0 58.0 500 48.0 60.3 23.6 43.3 22 11