Deflating Grades, or Students’ Futures: Probing the Grade Deflation Proposal

Professor Mark McPeek, who chaired the grade inflation committee.

On May 11, 2015, the Ad Hoc Committee on Grading Practices and Grade Inflation comprised of six Dartmouth professors released its “Proposal for Addressing Grade Inflation at Dartmouth College”.  The committee was chaired by biology professor Mark McPeek, and was charged by Provost Carolyn Dever with tackling the inflation question through thorough, actionable analysis.

The proposal reviews the history of grade inflation at Dartmouth, noting that students’ median Grade Point Average rose from 3.04 in 1974 to over 3.4 in 2014, and that in 2013-14 the median grade was an A-, 34.0% of all grades were a As, and 24.7% of all grades were an A-.  The proposal also considers the causes of grade inflation, including acknowledging faculty incentives to grade generously in order to attract students to particular courses and departments, and to encourage positive student reviews.  The proposal argues that professors should make courses more challenging and should assign grades based on the levels of mastery of course material contained in the Dartmouth Scholarship Ratings in the Organization, Regulations, and Courses (ORC) (A = excellent mastery; B= good mastery; C=adequate mastery; D=deficient mastery; and E=seriously deficient mastery).  The proposal does not recommend any specific limitation on the number or percentage of any particular grade, that instructors or departments set target medians or that grading be done according to any curve.   The proposal does, however, recommend a series of annual reports about grades, grading standards, grade distributions and course rigor among individual faculty, the chairs of each academic unit, Associate Deans and the Registrar that seem clearly intended to promote more coordinated and consistent grading and rigor standards.  Most significantly, the Proposal recommends that the frequency of A grades should decline and the frequency of B and C grades should return to “intelligible levels”.

The Dartmouth Review has been a consistent supporter of efforts to strengthen the academic rigor and overall academic life of the college.   (See “Toward a Greater Intellectualism” (February 19, 2015) and “Grading Hanlon’s Academic Plans” (February 26, 2015)).  In this light, the proposal merits commendation for advancing the discussion of an important, complex and sensitive subject.  Nonetheless, the proposal suffers from several methodological and substantive weaknesses and blind spots.  This article highlights these shortcomings in the hopes that the proposal is not the final word in Dartmouth’s consideration of grade inflation and academic rigor, and that these shortcomings will be addressed as the conversation continues.

Predetermined Conclusions and Insufficient Process

The proposal does not read as a fact-driven inquiry and balanced analysis that reasons its way to a recommendation, but rather as a predetermined policy outcome supported by cherry-picked arguments and data.  The chair of the committee, Professor McPeek, confirmed as much within days after the proposal’s release when he told The Dartmouth that the committee essentially formulated its policy recommendations at its first meeting because most members were already well-versed in the issue.  “Well-versed” in this context would appear to be a euphemism for “already had their minds made up”.  Subsequent public statements by committee members also indicated that the committee’s process from its initial meeting to publication of the proposal required only about four months.  One cannot help but infer that the committee felt considerable pressure and/or desire to finish its process and announce the proposal before the end of the 2015 Spring term. One consequence of this rapid process is that, as noted below, many areas of the proposal rely on the limited data that apparently was already available to the committee, and the anecdotal impressions of committee members in lieu of a rigorous process of gathering and analyzing relevant data.

Grading policy is among the most significant topics for any educational institution to tackle.   All Dartmouth stakeholders deserve a more open-minded, comprehensive, deliberate and data-driven evaluation process before any significant changes to grading policies are finalized.

The Dartmouth Bubble

While carefully documenting the rise in average grades given at Dartmouth from 1974 to 2014, the proposal is deafeningly silent about grade inflation trends over the same period across all colleges and universities nationally, and specifically the Ivies.  This silence does not do justice to the motivations behind the proposal nor the potential consequences of implementing the proposal.

Anyone who has peered outside the Dartmouth bubble of late appreciates the fact that grade inflation has been a pervasive issue throughout American higher education for more than four decades.  In the past few years, the issue has landed quite publicly, and to some minds embarrassingly, in the hallowed halls of the Ivies.  It was revealed that the median grade at Harvard is an A- and more than 60 percent of Yale grades were A or A-.  In addition, under pressure from students, parents and other stakeholders, Princeton reversed its decade-long unilateral experiment in imposing a numerical cap on the number of A-range grades that could be awarded (even though the data showed that the cap in practice had been consistently exceeded during its supposed existence).  By these indications, grade inflation at Dartmouth appears at worst consistent with, and perhaps less pronounced than, at the other Ivies.  A growing number of Ivy administrators and faculty (including, one can fairly assume, the members of the committee) believe that the reputation of the Ivies will be tarnished and the perceived market value of an Ivy degree will be undermined if a widespread perception takes hold that A-range grades are too easy to come by in the Ivies – and that such a perception would be very bad for all Ivy stakeholders.  Consequently, many in the Ivies believe that grade inflation must be halted and even reversed.

So why does the proposal fail to acknowledge the broader national and Ivy context?  One suspects that a bit of intra-Ivy pride is at play.   Heaven forbid that an official report from a Dartmouth faculty committee openly acknowledges that the school or committee has been motivated, much less spurred to action, by what is going on at “H-Y-P”.  Venerable Dartmouth sits in its Hanover bubble, soberly charting its own course without regard to trends and events elsewhere.

More fundamentally, however, one cannot help but infer that the proposal ignores the broader context because to do otherwise would require openly addressing the inherently relativistic nature of inflation in all its forms. Dartmouth tuition, room, and board costs all have risen dramatically since 1974. Does this mean we need a faculty committee to study and urge a return to 1974 costs? Of course not. The U.S. economy has experienced significant inflation in not only the costs of most goods and services since 1974 but also in salary and wage scales. The affordability of college today versus 1974 depends not on nominal college costs today versus 1974, but a complex web of relative prices today versus 1974. Similarly, a comparison of nominal grade averages at Dartmouth today versus 1974 offers relatively little insight on its own.

In addition, and perhaps most importantly, one suspects that the proposal’s failure to acknowledge the broader context of grade inflation is because the committee did not want to head on the challenges, risks and costs inherent in the proposal’s recommendations.  Admitting that grade inflation is pervasive in higher education would require them to admit that truly rectifying the situation, much like addressing global warming, would involve difficult collective action and enforcement problems.  The situation could be solved if all colleges (or carbon emitters) agree to, or have imposed on them, an enforceable regime to deflate grades (or reduce emissions).  But creating a regime that could prevent the inevitable temptation of individual participants in all cartels to not comply, would be a Herculean task.  In the absence of such a workable collective regime, well-intentioned but unilateral efforts by an individual participant to deflate grades likely will inflict disproportionate costs on that participant, and are likely to engender passionate resistance by affected stakeholders (as Princeton discovered).  This is not to say that the proposal is misguided in attempting to address grade inflation.  It does suggest that the proposal does the issue a disservice by failing to openly acknowledge (i) the broader context as regards grade inflation, (ii) that Dartmouth does not, in fact, exist in a bubble and (iii) that the unilateral action by Dartmouth recommended in the proposal entails significant challenges and potential costs.

What exactly are these potential costs?  First, awarding fewer A-range grades at Dartmouth would create intra-Dartmouth costs and benefits.  Those who receive fewer A-range grades under a deflated system would suffer by comparison in the job and graduate school markets to those Dartmouth students who continued to earn more A-range grades.  On the other hand, those continuing to receive A-range grades would benefit by having their work effort and achievements more clearly stand out.  Such a result seemingly would be a wash in terms of net costs and benefits to Dartmouth students in the aggregate and arguably would create an overall fairer and more meritocratic system.

Second, however, an overall deflation of Dartmouth grades would put Dartmouth students on average at a disadvantage in the job and graduate school markets relative to students from other schools unless employers and admissions committees adjust so as to accurately and fairly evaluate deflated Dartmouth GPA’s against GPA’s from other schools.  For such an adjustment to fully occur, three factors would be required: (i) Dartmouth would have to make publicly available all information necessary for schools to understand what a given GPA (say 3.5) really means at Dartmouth, (ii) employers and admissions committees would have to invest the time and effort to comprehend such data and map what various GPAs at Dartmouth equate to in terms of GPAs at a range of other schools and (iii) there must be no reasons why an employer or admissions committee would favor a nominally higher GPA from another school (say a 3.7) over a Dartmouth GPA (say a 3.5) when the employer or admissions committee otherwise considered the two GPAs to be substantively equivalent.

Although, as noted, the proposal never acknowledges the broader grade inflation context or the potential costs of Dartmouth unilaterally deflating, the committee apparently was aware of the potential costs; and, to the their credit, they recognized that information about Dartmouth’s deflated grades would need to be regularly publicly disseminated to address the first factor above.  However, the proposal is silent about the second and third factors — the likelihood that employers and admissions committees would invest the time and effort to map deflated Dartmouth GPAs against GPAs at other schools and would not be inclined to favor nominally higher but otherwise substantively equivalent grades from other schools.  The committee’s failure to research and analyze these factors is striking given that basic economic and behavioral principles, common sense and experience suggest that they may raise very real concerns.  For example, employers and admissions committees who do not repeatedly see a large volume of Dartmouth candidates likely will not invest in the information costs necessary to understand and map Dartmouth’s degree of grade deflation relative to other schools.  Moreover, even a moment’s reflection suggests that there are features that favor nominally higher but substantively equivalent grades from other schools.  For example, many graduate schools externally report the average GPAs of their admitted students.  These GPAs are generally computed without adjusting for relative grade inflation or deflation at the many undergraduate schools represented in a typical graduate school’s entering class.  The average GPAs are one of the factors that influence the perceived selectivity, and thus the reputations and rankings of graduate schools.  Consequently, graduate schools often have bias in favor of admitting students with higher nominal GPAs.  Even a minimal degree of research by the committee might have uncovered other such biases against deflated grades.

The committee’s failure to analyze the likelihood of mapping, or the extent of biases against deflated grades, is a critical omission which again suggests a Dartmouth bubble mentality – i.e., the idea that Dartmouth’s graduates are so in demand that we can assume that employers and graduate school admissions offices will do the work to evaluate Dartmouth candidates fairly and ignore existing biases against deflated Dartmouth grades.  One suspects that the failure also reflects the committee’s predetermined conclusions, haste to finish its work, and inclination to resort only to already readily available data.  A generation of Dartmouth students will pay the costs in countless small ways (e.g., a first job at a second-tier rather than first-tier firm) and large ways (e.g., abandoning a planned medical career altogether due to not being admitted to a quality medical school) if Dartmouth deflates and the committee’s untested assumptions that employers and admissions committees will not favor nominally higher but substantively equivalent GPAs prove incorrect.  Given the stakes, current and future Dartmouth students deserve a more fulsome investigation of these factors.

Third Rails and Sacred Cows

The proposal not only ignores the broader context of grade inflation outside of Dartmouth, but also steers clear of openly identifying and analyzing the cross-currents of the issue within Dartmouth.  This is a flaw in the proposal because grade inflation is not a universal phenomenon within Dartmouth.

On one end of the grade inflation continuum are required introductory courses in popular natural science and social science disciplines with B or B- medians term after term, year after year.  The proposal’s failure to acknowledge that grade inflation is a minor or non-existent problem in such courses insults the intelligence and experience of every Dartmouth student.  Moreover, contrary to the proposal’s recommendations, these courses are not graded by rigorous application of the Dartmouth Scholarship Ratings in the ORC to each student on an individual basis to determine whether that student’s performance was “excellent”, “good” or merely “adequate” in some universal and absolute sense.  On the contrary, these courses are graded by application of relatively inflexible curves designed to spread the students in each class between C-range grades and A-range grades based on relativistic comparisons of the students against each other. This is often based on relatively minor, and in some cases arbitrary, differences in grades received on assessments in those courses.   Such courses serve, among other things, to “weed out” lower performing students from popular majors and drive them into less popular majors.  Since such courses are disproportionately taken by freshmen, the courses also serve to terrify freshmen into making the jump to the college level, a considerable number of whom go through the emotional journey of dealing with the first B’s and C’s of their lives.  The courses also serve to check grade inflation and average GPAs overall at Dartmouth.  Finally, the courses also have the benefit, from the perspective of many departments, of allowing them to disproportionately give out relatively low grades to students who will not end up majoring or minoring in that discipline.

On the other end of the grade inflation continuum are virtually all courses in certain departments, and upper class electives and seminars scattered throughout various departments that have gained the reputation through A or A- medians and relatively light workloads of being “lay-ups”.  This is where grade inflation lives at Dartmouth. We all know it.

The proposal’s failure to call out the obvious can only be ascribed to fear of intra-departmental and intra-faculty third rails and sacred cows.  Apparently the committee fears that too many faculty members will oppose grade deflation measures if its proposal too clearly identifies the worst offenders, and challenges some of the tenuous comprises among the departments.   The most obvious such compromise is the 0.6 or more difference between average GPAs in the most challenging natural science and social science departments on the one hand, and a number of humanities departments, on the other.  The proposal not only does not directly criticize the departments on the high end of the range (many with median grades exceeding 3.7), but actually goes out of its way to argue that wide differences in median grades between departments have existed for a long time at Dartmouth, and that there is no need to alter this.  Committee members have repeatedly reiterated that they are not looking to eliminate these differentials in average grades between departments in interviews since they first published their recommendations.  It is analogous to saying that the way to curb excessive alcohol consumption at Dartmouth is for the student who consumes two beers once per week and the student who downs a six-pack four nights per week to each reduce their respective weekly intake by one beer.  Or saying that the way to shrink the federal government’s debt is to raise taxes the same amount on all citizens, not just the rich – what percentage of the Dartmouth faculty would support this approach?

Another obvious sacred cow in the proposal is the committee’s apparent belief that the autonomy of individual professors and departments in setting grading policies must be preserved.  Surely they must recognize that such autonomy, when combined with the incentives of individual professors and departments to inflate grades, is a central cause of grade inflation.

In any event, perhaps there are good reasons why overall grade inflation should be and can be addressed without eliminating the historic differences in average grades between departments.  And perhaps there are good arguments for how grade inflation can be addressed while preserving faculty and departmental autonomy.  But these topics and others that the proposal overlooked also need to be discussed more openly and thoroughly before we can move onto meaningful action.

We Don’t Have to Worry About GPAs No More. That’s Good. One Less Thing 

Another significant weakness in the proposal is the committee’s attempt to demonstrate statistically that GPAs really do not matter very much in graduate school admissions. As an initial matter, one must note the profound irony of such an argument emanating from a group of distinguished Ivy League academics, employed by a college that heavily weights high school GPAs in its undergraduate admissions and whose own journeys through graduate school and then the ranks of elite academia almost certainly belie the notion that undergraduate GPAs really do not matter.

At a more granular level, the data mustered by the committee to support its argument not only would not pass muster in any introductory statistics course, but also actually undercuts the committee’s argument. The committee bases its argument entirely on data on Dartmouth graduates who applied to medical schools for admission between 2002 and 2014. No data is presented for law schools, business schools, graduate arts and sciences programs, etc. The applicability of medical school data to other graduate schools obviously is unknown. But for argument’s sake, one can ignore this methodological weakness because the most serious weaknesses are in the two statistical observations actually made using this limited medical school data set.

First, the proposal notes that in each of 2002 and 2014, respectively, a student with a GPA of 2.4 was admitted to two medical schools and a student with a GPA of 3.7 or above was admitted to no medical schools. The holes in, and questions raised by this data are almost too numerous to list. These include the omission of all data regarding the applicants that may have been relevant to medical school admissions other than overall GPA (e.g., college major/minor, performance in pre-medical courses, MCAT score, race, gender, disability, economic background, medical-related work experience, demonstrated commitment to medicine and other significant extra-curricular achievements) as well as information about the number or ranking of the medical schools they applied to. For all the reader knows, the two non-admitted students with a 3.7 GPA were theater majors who had an easy course load and bombed the MCAT but decided nonetheless to submit an application just to Harvard Medical School on a lark. On the other hand, the two admitted students with a 2.4 GPA were students from economically disadvantaged backgrounds who double-majored in chemistry and physics, crushed the MCAT, indicated their commitment to practicing medicine in depressed rural areas, held repeated summer jobs in medical research, captained Ivy championship athletic teams and applied to a dozen medical schools and nonetheless were admitted only to one medical school in Mexico and another in Caribbean country (and not Cuba, mind you).

Second, the proposal plots the maximum, average and minimum GPA of all Dartmouth students who were admitted to at least one medical school and those who were admitted to no medical schools between 2002 and 2014. This data suffers from the same failure to control for variables other than overall GPA as described. However, given the much larger number of data points relative to the two individuals in each of 2002 and 2014 described above, one might assume for argument’s sake that these other variables tend to cancel out across the larger number of data points and thus that this plot provides a reasonable isolation of the effect of overall GPA on medical school admissions. In any event,  the data shows that average GPAs for those admitted and not admitted differ by about 0.3 each year (roughly 3.2/3.3 for non-admitted versus 3.5/3.6 for admitted). From this data, the committee concluded that while having a higher GPA is one contributing factor for successful admissions, it is far from the “determining” factor. In this context, “determining” apparently means the sole or overwhelmingly most important factor. Clearly GPA is not the sole factor in determining medical school admissions. Whether it is the overwhelming factor is debatable. However, this is not the relevant question. The question is, all other things being equal, how much of a deviation in overall GPA separates the average admitted and non-admitted medical school applicant. The answer from this data appears to be 0.3. This result does not suggest that GPAs do not matter, as the proposal asserts. Rather, this result suggests that differences in GPAs, and even relatively small differences, matter very much. This conclusion comports with the observation that grade inflation has led to both an overall increase in nominal GPAs and a compression in the range of nominal GPAs. Thus, 0.3 now represents a significant difference in GPAs and correspondingly the likelihood of medical school admission. Consequently, contrary to the conclusion advanced in the proposal, the data suggests that even relatively minor deflation in nominal GPAs at Dartmouth could have a negative effect on graduate school admissions for Dartmouth students.

NRO Must Go 

The proposal also recommends eliminating the non-recording option (NRO), arguing that the option is not being used as downside grade protection by otherwise hardworking students stretching themselves in courses outside their comfort zones as intended, but rather by students looking to get a passing grade and course credit with little effort. The committee may be somewhat correct in this assessment. But the student self-reported hours of work versus grade expectation data they relied on does not make the case adequately on its own. If they are serious about eliminating the NRO, they should gather data about actual work effort and grades that would have been received absent the NRO, and compare such data to that of students in the same classes choosing not to elect an NRO.

Are We Talking About Grade Inflation, Academic Rigor, Academic Effort, or All of the Above? 

The title of the proposal (“Proposal for Addressing Grade Inflation at Dartmouth College”) suggests that the committee’s concern is entirely with grade inflation. Nonetheless, the proposal drifts into discussion of academic rigor and students’ academic effort. To some degree, this is appropriate and unavoidable as the three topics are related. On the other hand, if the committee felt the need to address academic rigor and student academic effort as part of the consideration of grade inflation, they should have gathered considerably more data about the topics. The proposal clearly implies that the committee believes there is insufficient academic rigor in some corners the curriculum, but does not attempt to set forth the optimal amount of academic rigor, to quantify the prevailing level of rigor, or to identify the departments or courses in which sufficient rigor is lacking. No doubt, addressing and quantifying academic rigor is a difficult task. But if the committee feels the need to bring the topic into the conversation, they ought to undertake the task. In regard to student academic effort, the proposal contains some data regarding students’ self-reported hours of course work per week compared to the grade that students expected to receive in the course. This data is presented to support the case that the NRO should be abolished. Otherwise, however, the proposal contains no data about how hard Dartmouth students actually are working in their classes on average across the College, in particular departments, in particular courses, or broken down between freshman, sophomore, junior, and senior years. Instead, all the proposal offers is anecdotal viewpoints such as the one reproduced below.

We contend that many Dartmouth students overcommit in their extracurricular activities. This is understandable, since they had to compile a litany of activities to be admitted to Dartmouth in the first place. Nevertheless, too many Dartmouth students have too many activities outside of their classes. We wonder whether many students view Dartmouth primarily as a place to live and participate in activities – and, when time permits, to work on their courses. A number of students described to us their time commitments to extracurricular activities that are the equivalent of full-time jobs.

Not only are such viewpoints unsupported by data, (and seemingly indicative of a particular bias against athletes) but also contrary to the experience of the vast majority of Dartmouth students.

Conclusion

The committee’s proposal reflects the incessant carelessness and purpose-driven analysis that has come to characterize the implementation of many facets of the Moving Dartmouth Forward (MDF) program. Yet the fact that the committee’s recommendations have not yet been implemented is a cause for hope that Dartmouth may break the trend of heavily-flawed, hastily-crafted policy changes. We at The Dartmouth Review certainly hope this is the case. A faulty implementation of grade deflation and the other recommendations discussed in this article could be catastrophic, for both the College and the futures of the students that call it home. Without scrupulous, thorough, and unbiased analysis, the prospects of change for the better are grim, at best. Will the faculty and administration rise to the occasion?