The Dec. 27 front-page article " How U-Md. sifts through 30,000-plus applications" raised bothersome issues about college admission procedures, such as the "26 factors" considered by this university. Presumably most other universities have similar factors. Most of the factors under consideration are nothing more than empirical hypotheses, which beg for testing. For example, years ago after a detailed discussion of admission procedures with the admissions director of a leading Ivy League university, I asked the following question: "You have listed a variety of (presumably) predictive criteria. . . . What do these actually predict?" He seemed bemused, but when he finally figured out the question, his answer was that SAT scores predicted first-year grades — nothing else.
I'd like to see a little science injected into the process. Establish reasonable criteria for selection procedure outcomes, test their actual relationship to these outcomes (not an easy task), and if the nature of the outcomes distribution is not amenable to statistical analysis, then consider the following approach. Of the entire applicant pool, using criteria of choice, determine the best applicant subset, such as the top 25 percent everyone agrees should be offered admission. Similarly, identify the lowest 25 percent subset for rejection. Then, simply use a lottery to offer the appropriate number of admissions to the middle 50 percent. This would simplify the entire process and face the reality that as hard as they try, admissions committees cannot predict the next Einstein, Albert Schweitzer, Steve Jobs, Mother Teresa or serial miscreant.
Paul G. Rochmis, Vienna