One of EphBlog’s continuing missions is to bring the excellent work done by Williams undergraduates to a broader audience. For now, the focus continues to be on senior theses. Although I hope to convince the College to post all those prior theses related to specifically to Williams, we will do what we can in the meantime. Consider:

Admission criteria as predictors of achievement : a case study of Williams College, by Jennifer Doleac ’03.

Special thanks to Jennifer for providing me with a copy of her thesis. There is a lot of great stuff here, which I hope to be able to comment on at some point. Her thesis lacks a abstract, which is too bad for those looking to get a quick overview. (The Economics Department ought to require all theses to have abstracts, a handy tool for readers and a useful method for forcing students to focus on the specific point that they are trying to make.)

Below are selections from Doleac’s conclusion (pages 96 to 99):

As competition for admission to elite colleges and universities increases, it is ever more important to verify the ability of admission criteria to distinguish between better and better applicants. Given the emphasis Williams places on building a vibrant college community, and the increasing evidence that peer effects matter in higher education, it is important to select applicants who will contribute meaningfully both in and out of the classroom to maximize the value of the college experience for the student body as a whole.

Much attention has been paid to fine-tuning assessments of applicants’ academic potential, and with considerable success. Consequently, the AR [academic rating] used by Williams explains a fair amount of the variation in academic achievement, with other criteria adding relatively little, especially in models of basic measures like GPA. Other types of academic achievement that indicate significant intellectual engagement, such as departmental honors, are less predictable and more sensitive to socioeconomic background. The implications of this are twofold: First, the faculty should be reassured by how little of an incoming student’s intellectual trajectory is predetermined. College experiences, professors and courses clearly play a very large role in shaping students’ academic development. Second, more should be done to encourage students from less-advantaged backgrounds to live up to their potential. It is discouraging how much of the power of these models comes from socioeconomic indicators rather than academic talent and interests. Unfortunately this is not a problem that can be remedied by Admissions officers alone, but must be addressed after matriculation by the faculty and administration.

With these points in mind, my results do suggest a few ways to improve measuring the academic potential of applicants. Specifically, the Ph.D. attribute should be assigned differently, students’ SAT II scores should be given more weight, and the two indicators of writing ability — the writing attribute and the SAT II: Writing — should be largely disregarded. I will discuss these criteria in more detail below.

Students’ engagement in nonacademic areas has not been examined as seriously in the literature, however, either because extracurricular achievement is difficult to define, or because it is assumed to be shaped primarily by college experiences and thus to be unpredictable. The limited predictive power of the models in this paper certainly support the latter of these concerns, and it would be tremendously interesting to examine the interaction of precollege achievements and peer effects. Certainly there will be students who showed no signs of interest in extracurricular involvement in high school who then “come alive” when put on a vibrant college campus, and students for whom the opposite is true. The same could be said in terms of academics, however, and the myriad strong relationships between precollege and college achievements in that area — even at a school with as condensed a range of students as Williams — is somewhat astounding. I believe the same sorts of relationships are present between precollege and college nonacademic achievements, and that, given the importance of nonacademic involvement to the strength of a college community, finding these relationships should be of the utmost importance to college Admissions officers.

The NAR, unfortunately, does not correlate highly with many types of extracurricular engagement, and when it does its predictive power tends to vary with socioeconomic background. We shouldn’t be surprised that particular credentials are more indicative of actual motivation or ability for students from certain backgrounds, just as we aren’t surprised when the low SAT scores of a low-income student are less indicative of potential than are the same low scores of someone whose parents could afford testprep courses. If the NAR is to indicate one’s dedication, follow through, and motivation to be a community leader, achievements should be evaluated in the context of the opportunities available. If this is done, I have no doubt the NAR will be a valuable predictor of future involvement in and contribution to the college community.

That said, it is important to note the often large effects of socioeconomic background and gender, especially in predicting campus leadership. That such background characteristics have so much influence on community engagement is disturbing, especially in the context of ongoing discussions of increasing diversity on campus. Anecdotal evidence of a steady decline in leadership by female students is at least partially supported by the analyses above, and, considering the active (albeit informal) “affirmative action” practiced by many campus selection committees faced with a dearth of female applicants, the reality is likely far worse than these results reveal.

Meanwhile, if lower-income students are less likely to seek or attain leadership roles on campus, as this study suggests, the benefits of increasing diversity will be greatly reduced. It seems that the campus’ “ruling class” is limited to its more affluent students — or at least that this perception is prevalent enough to limit others’ attempts to crack any (real or imagined) glass ceiling. Considering that the bottom income quartiles at Williams consist primarily of students in the middle, uppermiddle and even upper classes highlights how restricted leadership is to the wealthiest of students.

It’s a problem if only the “ideal” or “typical” Williams students hold leadership positions, but an even greater problem if that ideal or type is defined primarily by income and gender. Those who do not see themselves as close enough to the “ideal” Williams student may not feel they are in a position to represent their peers, and/or may be too discouraged or intimidated by a “boys’ club” culture to pursue top leadership positions. These students are thus prevented from full engagement in what Williams touts as one of its primary assetsthe campus community, complete with the r�sum�-building leadership positions within it and the opportunities to work and network with other future leadersand so for them the value of a Williams education is greatly reduced. This underrepresentation of lower-income and female students — and consequent homogeneity of opinion and experience among those who are shaping campus life — is again a problem at least as much for the administration as for Admissions. Further investigation of the types of achievements predicting the involvement of underrepresented students will help identify more promising applicants, but an active role on the part of the administration in changing the campus culture may be required. Peer effects, again, are surely influential here, and could be used to indirectly steer the campus community in a more desirable direction.

Thus, while admission policy certainly plays a significant role in determining the character of the campus community and collective achievements of each class, campus experiences are often even more influential, with positive or negative effects. That said, my analyses have suggested possible areas of improvement in admission criteria, which are considered in more depth below. The relevance of the context of high school credentials suggests that current admission policies may be biased toward recognizing the potential of some students more than others. Using one-size-fits-all criteria implies turning away qualified applicants whose talents go unnoticed because the tools we’re using are inadequate. I firmly believe refining indicators of academic and nonacademic potential is an exceedingly important step in improving the value of a Williams education for all students and setting a strong example for other selective colleges and universities.

Print  •  Email