You need to enable JavaScript to run this app.
Our Grading System Needs an Overhaul
William Speer, Math Teacher
May 23, 2018

On August 5, 2014, less than a month before I started at Deerfield, I received an email from Diana Kocot stating that my schedule was available on DAInfo. Full of anticipation, I logged on to check my course schedule and to do some background research on my students. As I clicked around my classes and dorm rosters, I noted a recurring pattern under the “Grades” tab: student averages were incredibly tightly bunched, and it felt like half of my soon-to-be students sported averages between 88 and 91. This seemed like an odd coincidence at the time, but as I came to find out, my observations were far from anecdotal. Before even setting foot on campus, I was familiar with an ingrained aspect of Deerfield academic culture: grade compression.

According to the “Grades and Percentiles” tab on DAinfo, for the 2016-2017 school year, the top quintile of juniors had averages ranging down to 92.4 percent, while the bottom quintile ranged up to 87.87 percent (the spread is similar for seniors and slightly wider for underclassmen). This spread of four and a half percentage points covered the middle 60 percent of the class — so, roughly two points in either direction (maybe a missed negative sign in math, or a forgotten vocabulary word in English) might determine whether a name comes up in front of the Cum Laude Committee or the Academic Standing Committee.

Credit: Mark Chung

Deerfield is an inherently stressful place, and this narrow margin of error only intensifies that pressure. As described above, our compressed grades attach massive significance to each percentage point. These percentage points should theoretically represent one hundredth of the grade scale, but in reality each point carries much more weight. As a result, students (particularly juniors and fall-term seniors) fight incredibly hard for every point, given the disproportionate benefits that come from every slight boost. While we want to have driven, motivated students, I believe this narrow scale promotes an unhealthy focus on numerical grades. Compression inspires obsession with the minor distinctions that constitute a one or two-percentage point swing, to a level which I find counterproductive and unhealthy.

Additionally, compression at the other end of the scale may inspire the opposite behavior. A median grade of 90 would not be a problem if the grades span the full spectrum up to 100! Currently, this is not the case, particularly in certain classes and in some departments. This can disincentivize effort at the top end — why give 100 percent effort if the best you can get is a 93 percent? While it is certainly true that not every section contains a mid-to-high-90s caliber student, it is true that many of them do. If we wish to decompress but not lower our grades, simply opening up the full upper range would be an easy solution.

While the internal disadvantages of our compressed grades cause me enough concern, there is also a nontrivial external disadvantage created when our students apply to colleges. For any given school, many Deerfield applicants likely have similar, highly compressed grades. In this case, it is difficult for the schools to distinguish applicants by their academic record, and they may turn to a different metric. A convenient alternative would be to place increased weight on standardized test scores, as Math Teacher Sean Keller suggests. I would not want colleges to enhance the significance of a four-hour standardized test compared to 1–4 years of full-time study here, and I suspect the majority of the faculty feel the same way. However, if grade compression remains the same or gets worse, those schools have no choice.

How can we fix this? One idea would be to prescribe fixed (or “suggested”) averages and/or standard deviations for course level (100, 200, 300, etc) and/or track (honors, regular, etc). This would ensure that heterogeneous classes (where students of varying ability are mixed) have a different grade profile than homogenous classes (those where students are of similar ability level). A heterogeneous class, such as Africa and Latin America, would reasonably be expected to have a lower average level of achievement and a greater average spread than a more homogenous class, such as French V Honors. That distinction would help students at all levels: those students who excel in heterogeneous, lower-level classes will be more apparent “shining stars,” clearly identifiable by their distance from their class medians. Such a distinction would provide better evidence for future placement like moving into the honors track. Furthermore, students who take mostly upper-level honors classes will have higher averages on their transcripts, making them more easily identifiable to faculty and college admission officers alike. We will no longer have the problem of a few multiple-choice errors dropping a student from top decile to middle quintile, as grades will span a wider spectrum across the board.

For those who feel the above solution is too prescriptive, I offer one more alternative: simply weight grades for higher-level, honors, and AP classes. This would likely require abandoning the 100-point grading scale (to prevent averages from exceeding 100), but it would be an easy way to space out term averages and reward those students with challenging course loads.

Of course, compressed grading is not a Deerfield-specific issue; many high schools and colleges are struggling with this problem. Furthermore, the academic level of the school has likely narrowed over the years, so some grade compression would be natural. However, just as we have recalibrated our admissions and student life policies to reflect modernization and better serve the student body, there is no reason we can’t do the same for our grading system. It would take honest discussion, rigorous data analysis, and a willingness to compromise — but our students would be better for it.