How to pick a mate efficiently

Sooner or later, most Princeton undergraduates will choose a mate with whom to share their lives. Most of them will engage in a time-consuming, often expensive ritual of evaluating the many dimensions of prospective mates — looks, height, biomass, intelligence, sense of humor, pedigree, earnings potential and so on. The search for the optimal mate can take years and may not even lead to the maximum attainable bliss because dating and courtship involves so much deceptive marketing.

I have a better idea for a truly efficient search for a mate: Line up all prospective candidates behind a curtain, letting only their feet stick out, and pick a mate on the basis of the feet. While not very thorough, the process is quick, and feet certainly do tell something about a person — probably as much or more as does the GPA by which academia evaluates human beings. Indeed, in my view, mate picking by feet only and reliance on the GPA operate on roughly the same intellectual plane.

The manufacture of GPAs is a wondrous process. In the large courses for undergraduates I have taught at Princeton, for example, we develop numerical percentage scores for each of the various components of the course that are formally evaluated. These scores are then collapsed, at year's end, into a weighted average overall percentage score. Dean of the College Nancy Malkiel requires me to convert these fairly accurate numerical scores into much cruder letter grades which, in turn, are then reconverted into even cruder numerical scores that are then used to calculate the one-dimensional GPA thought by graduate schools and prospective employers to distill accurately a student's possibly varied academic achievement during four years at Princeton.

Mapping numerical scores into letter grades is a wondrous process in itself. To do so, I rank the students in my courses by their overall numerical course scores, whereupon I gaze intently at this ranking, desperately looking for gaps I can use as breakpoints for the letter grades. A gap of half a percentage point is heavenly, because it so clearly divides students into those evidently deserving, say, an A and those worthy only of a B+. But often the breaks are much smaller. If your roommate received an A in my course and you only a B+, the numerical gulf separating you may be only as wide as 0.3 percentage points.

And it gets worse. Until Malkiel's recent attempt to impose at least some constraints on grading practices, the reigning theory on this campus had been that the distribution grades in particular courses was strictly the professor's prerogative. Thus, in an article in The Daily Princetonian on grading (March 27, 2006), history professor Stephen Kotkin is quoted as saying that he has given A's to about 10 percent of students in his upper-level courses — far below the dean's guidelines. In the same article, Professor Ed Zschau '61 is quoted as saying that he would like to have given far more A's to students in that year's course than the dean's still generous guidelines permitted.

Now it may be that all the smart students in campus gravitate to Professor Zschau's course and all the dumb ones (well, Princeton does not have "dumb" students — let's say "less smart" ones) gravitate to Professor Kotkin's class. An alternative hypothesis is that the entire traditional grading system on Princeton campus had best be described as an incoherent mishegas of which intellectuals should be ashamed.

To extract GPAs from this mishegas and then base crucial decisions about human beings on such a number borders on the irresponsible. Furthermore, coupled with the lack of coherence of grading standards among professors on any campus, widespread reliance on GPAs to rank students can seriously bias the selection of courses towards those taught by professors who are known to be easy graders, thus distorting the students' intellectual experience. You want a job at Goldman Sachs or admission into medical school? Think twice before taking a course from tough graders like Professor Kotkin.

Years ago I had proposed that crude letter grades be replaced with numerical scores out of 100. For each course taken, a student's transcript would then show the number of students in the course, the student's score for the course, the average or median (or both) score for the class, the range of scores from minimum to maximum and perhaps even the student's rank in that course. GPAs would not be calculated. If all Ivy League schools agreed to adopt such an approach, the rest of the country probably would follow.

As I recall it, the USG rejected the very idea out of hand. Apparently, Princeton students actually prefer the fuzzy mishegas that Princeton's grading practices always have represented, as long as lots of As are given and their number increases over time, in step with the ever rising levels of intelligence successive cohorts of students impute to themselves. Small wonder then that so many students also reject the dean's attempt to impose at least some modest degree of coherence on our grading mishegas. Uwe E. Reinhardt is the James Madison Professor of Political Economy and a professor in the Wilson School. He can be reached at reinhard@princeton.edu.