Report on the Review of Grading Practices and the Determination of Honors Susan Pedersen, Jeffrey Wolcowitz, and John O'Keefe May 6, 2002 Many committees and departments of the Faculty of Arts and Sciences review their own grades and grading practices, but the Educational Policy Committee has also periodically scrutinized data for the FAS as a whole. It began such a review in the spring of 2001. This past fall, having noted that the data revealed a continuing upward trend in grades and a consequent compression of the effective grading scale, the EPC asked the Dean for Undergraduate Education to provide data on these trends to all members of the Faculty and to write to department chairs asking them to convene discussions about the pedagogical implications of our current practices. In November, faculty members were thus asked to discuss current grading practices, to articulate departmental views on grading standards, and to report back to the Dean for Undergraduate Education with any observations, suggestions, materials, or recommendations. Many departments reported by early spring, offering a number of helpful suggestions and reflections on these topics. A subcommittee of the Educational Policy Committee was established to consider those responses as well as other materials related to grading at Harvard, and met several times in February and March. As part of its work, the subcommittee found it important to articulate a set of principles that it believed should govern grading at Harvard. Having described those principles, it then put forward a set of recommendations for administrative action by the Office of the Dean for Undergraduate Education and for Faculty legislation. Those recommendations were discussed by the Educational Policy Committee and the Faculty Council in April and May. Noting that departmental discussions often revealed considerable concern about our current practices for awarding honors, the subcommittee also outlined for the EPC some possible means of reforming our honors system. In January 2002, the Dean for Undergraduate Education had already raised the question of whether the Faculty should consider eliminating the degree cum laude based solely on grade point average (the degree formerly known as cum laude in General Studies or CLGS). When this proposal was discussed in March, however, many faculty members expressed the view that the CLGS degree should be considered within the context of a wider review of our methods of determining honors. The EPC and Faculty Council thus also discussed possible amendments to our system of determining honors in April and May. The discussions in the EPC and Faculty Council resulted in two proposals that require legislation. It is proposed that the FAS substitute the more common four-point grading scale for its current 15-point scale; and that the FAS limit honors to a fixed percentage of the class. We should stress, however, that these proposals are part of a wider effort to make our grading practices and methods of determining honors more adequately support our pedagogical ideals--an effort that will depend as much on faculty and departmental efforts and administrative support as it will on legislation. This document is intended to make both these specific proposals and the wider issues clear. Grading Practices The discussion of grading practices begins by recognizing that Harvard undergraduates are among the most qualified and capable students in the nation. As a group they enter Harvard with excellent preparation, a deep desire to excel at their academic work, and the disciplined work habits necessary to do just that. It follows that they deserve an education tailored to their skills and talents. A Harvard education must be directed at Harvard undergraduates (and not at some other, notional group of students) who should be challenged and engaged by the academic work that the Harvard faculty requires of them. The grades and honors that we award should be calibrated to the work done by these students in the courses we teach, and should allow us to distinguish among that work, even with so many students performing at a very high level of accomplishment by national and international norms. The discussions that have gone forward in the departments have been thoughtful and constructive. Through those discussions, a number of common themes emerged. Many departments reported an overriding interest in teaching, and an interest in grading mostly insofar as it was part of a larger conversation about good pedagogical practices and our general educational mission. As important as grading is, many departments felt that other topics to which it is related were equally worthy of discussion, and feared that being too concerned about grading per se might interfere with what is best about a Harvard education. Several departments noted, for example, that while smaller courses and writing-intensive courses where faculty members work closely with students often have higher average grades, those grades may be evidence of the very high quality of instruction and learning in those settings. Yet, some aspects of our grading practices did emerge as concerns across the departments. Many faculty members agreed that the compression of effective grades limited our ability to differentiate among our students' work and to motivate our students adequately. Faculty members also expressed concern about our grading scale, about particular pedagogical practices (such as the move towards increasing reliance on papers rather than examinations), and about student evaluations. Grading practices vary significantly across the three divisions, and so perceptions of the nature of the problem differ as well. Very briefly, one can characterize these differences as follows. While the natural scientists use a wider range of grades and hence give a lower "average" grade, they also give proportionally more straight A grades: in the natural sciences, an A is the grade most often awarded. Natural scientists in particular, then, worried less about the range of grades available than about their diminishing ability to recognize truly exceptional work. The humanists, by contrast, employ a narrower range of grades (effectively, only A's and B's) while also having a relatively high modal grade: the A- is the grade most often awarded in the humanities. The social scientists also use a narrower range of grades than the natural scientists, but they tend to grade more stringently than the humanists: the B+ remains, by a slim margin, the most commonly awarded grade in the social sciences. Faculty members' concerns about grading practices should be understood in light of these rough differences. There is, of course, considerable variation across departments within these broad divisions and also among individual courses within any given department. Principles about Grading To address adequately faculty concerns about grading, we must begin by articulating why we grade our students. Many, if by no means all, faculty members tended to agree that grades have two main purposes. First, and in the view of many faculty members more importantly, they serve our pedagogical needs by indicating to students the relative strengths and weaknesses of their past work and by motivating them to do better work in the future. Second, they differentiate among the work students have done in a way that suits our internal administrative needs (for example, in determining honors) and also meets the needs of external audiences, including prospective employers, graduate schools, and professional schools. The EPC subcommittee agreed that these goals will be best met if we have as our principles the need to strive for transparency in grading, common standards, and an adequate range of grades. Whatever our grading standards, and whatever grading scale we use, all the grades in it should be clearly defined, and instructors should have clear understandings of the circumstances under which they will award each grade. The meaning of our grades should be apparent to instructors and students and to the external world as well; our grading practices should thus be transparent. Similarly, grading practices across the different parts of the curriculum should aim at some common standard. Thus, while grading is the responsibility and prerogative of the course instructor, departments and the College as a whole should make some effort to articulate collective standards and support faculty members in their efforts to adhere to those standards. Finally, the subcommittee agreed on the need to preserve an adequate range of grades. Pleas from particular faculty members to introduce an "AB" grade or an "A+" grade are a sign that our scale no longer differentiates adequately among levels of good work. Recommendations about Grading The principles enumerated above led the subcommittee to a number of recommendations for actions to be taken by individual faculty members, the several departments, the Office of the Dean for Undergraduate Education, and the Faculty as a whole. 1. The grading scale. Many faculty members argued, and the EPC and Faculty Council agreed, that our current grading scale (the 15-point scale) should be abandoned. The current scale skips a numerical step between A- and B+, between B- and C+, between C- and D+, and between D- and E. Faculty members in a number of departments increasingly mention the numerical gap between an A- and a B+ (and to a lesser extent the similar gaps further down the scale) as a factor that increases grade compression. It marks the difference between A-level work and B-level work, but in a way that exaggerates that distinction and penalizes students whose work receives B-range grades, when those grades are translated into a numerical grade point average. Faculty members increasingly report their reluctance to impose that penalty, given the upward pressures on grades, and this itself is a significant factor in the compression of grades in the A-range, and the increasing devaluation of B-range grades. In many areas of the curriculum (especially in the natural sciences) grading is done on a continuum that does not correspond to the current numerical scheme. Faculty members suggested, and the EPC considered, a range of alternative scales, some numerical and some letter-based. There were attractive elements to many of these, especially in that they offered the opportunity for a one-time recalibration in our distribution of grades. Yet, many faculty members were persuaded that no such recalibration could be enforced without a "curve," and there was wide opposition to mandatory curves. Even more persuasive, however, was the argument that to replace one idiosyncratic and unique Harvard scale with another idiosyncratic and unique Harvard scale would hardly promote the goal of "transparency." It was agreed, then, that we should take this moment to replace our 15-point scale with the standard 4.0 scale used by most other universities. This move would eliminate our current "gaps" and make our transcripts more useful. This proposal received wide support from the EPC and Faculty Council and will come before the Faculty for a vote on May 21st. In itself, of course, a new scale will not alter grading practices. Instead, faculty members must agree, individually and collectively, on the standards that we hold. Reports from departments and discussions within the EPC revealed some level of agreement on such matters. A good portion of the Faculty continues to feel that most grades should fall in the "B" range, with "A" reserved for work that is both excellent and exceptional. Unfortunately, some faculty members (and, still more, some teaching fellows) report feeling pressured by students into giving higher grades than they feel comfortable awarding; they report feeling a lack of adequate information about grading standards and administrative support in these matters. It is clear, then, that both the departments and the Office of the Dean for Undergraduate Education can do more to support faculty efforts to maintain high and rigorous grading standards. 2. Administrative measures. Given the degree of consensus that exists within the Faculty about the range and likely distribution of grades, the Office of the Dean for Undergraduate Education should do more to foster common standards and make those standards more widely known. The EPC has asked that the Office of the Dean for Undergraduate Education pursue this goal by taking the following steps: Review the language about grading in the Handbook for Students and Information for Faculty Offering Instruction in Arts and Sciences, and prepare new and fuller descriptions of our grading scale and our grading practices, as needed. Compile materials on grading standards from the various departments, and distribute those materials for the use of others, both to assist in their own evaluation of student work and to help faculty members become more aware of grading standards across the curriculum. Prepare materials on the College's grading practices and standards for orienting new faculty members; these materials should assist faculty members on an ongoing basis as well. Ask course heads to report the distribution of grades assigned in a course at the bottom of the grade sheet, as a way of raising awareness about actual grading practices. Provide additional grading information to individual instructors to supplement the Course Grade Index information currently distributed. Communicate with department chairs about instructors whose grading practices are significantly more generous or more restrictive than the norms across the College. Work with the Committee on Undergraduate Education to design effective methods for evaluating teaching that can contribute to the ongoing development of good pedagogical practices, including grading. Promote methods of evaluating student work that foster consistency and fairness in grading, such as sharing grading responsibilities among the teaching fellows so that section leaders are not consistently grading their own students, involving course heads directly in some specific grading activities, and including multiple methods of evaluation in a course to ensure that grades reflect both students' knowledge and their skills. 3. The transcript. Finally, the EPC considered the question of possible modifications to the transcript. Some of our peer institutions have made such changes, and there is a range of additional information that might be included for each course on a transcript (mean grade, median grade, percentile, or %A's in the class, as well as class size). Some departments indicated support for such a move here. The EPC subcommittee discussed possible alternatives and felt particularly positive about the option of including the percentage of A-range grades and the size of the course on the transcript. Such information, it was felt, would promote many of our grading goals. Such a system would leave faculty free to award as many A-range grades as they liked, but would make those practices transparent. It would make clear how exceptional or unexceptional any particular A-range grade was. It would provide a bit more information about students who received lower grades, but would not unduly penalize these students by introducing percentile rankings or some other such measure. The main argument against adding such information to the transcript is that it would, by implication, label as an "easy course" some exceptional courses that attract a particularly gifted undergraduate body and (correspondingly) have a high percentage of A's. (This has come to be known as "the Math 55 problem.") The subcommittee felt that this issue was something of a red herring. People who read transcripts are, usually, people accustomed to reading transcripts: such readers understand that, among two courses attracting large numbers of first-year students, one labeled "Math 55" is likely to be more challenging and more selective than one labeled "Math 1," and will judge accordingly; this is, of course, supported by the course titles that appear on our transcripts. Such readers would be untroubled to find that a course in medieval economic history attracting 2 students had 100% A's, and would not question the solidity of that "A" grade. They would, on the other hand, probably view more skeptically evidence of high percentages of "A" grades in large introductory courses. Such a transcript might thus generate some pressure to reduce the number of "A" grades in larger courses in particular. The full EPC was less persuaded than the subcommittee of the specific advantages of including this particular information on the transcript. Thus, the EPC intends further study of "enhanced transcripts" and may bring a proposal forward next year. Honors The discussion of honors at Harvard begins by acknowledging two unusual aspects of our system of honors. First, it is fair to say that we have historically assumed that honors are something of which all of our students are capable and to which all of them should aspire. We have tended, therefore, to differentiate among students more by the level of honors awarded than by the award of honors per se. Second, while we use the level of honors as a mark of distinction, we use honors in general more as a motivational tool: the prospect of a degree "with honors" motivates our students to attempt a more rigorous concentration program than they might otherwise undertake. These assumptions structure our curriculum at every level. That is, we both have distinct honors "tracks" in concentrations and distinct "honors-only" concentrations--and, while the designation "honors-only" is a description of the nature of the program required (and not of the outcome for any given student), both students and faculty members tend to assume that students undertaking an honors "track" will be awarded honors. There is some evidence, however, that faculty views on this question have shifted, with many now believing that honors per se (and not primarily the level of honors awarded) should be a means of distinguishing particularly outstanding work, as is the case at many of our peer institutions. As noted above, in their discussions of grading practices, many departments reported that faculty members were more troubled by the distribution of honors degrees than by the distribution of grades. Many departments worried particularly that the number of magna degrees granted is now so large that it includes too wide a range of student achievement. Correspondingly, the degree cum laude in a field has become less an honor earned by a student's work, and more a contractual arrangement where the degree follows almost automatically from a student's choice to pursue an honors track or to enter an honors concentration. The degree cum laude based solely on grade point average is meant to honor excellent work across the curriculum, but in fact serves as a way for students who choose not to pursue honors requirements in their concentration to receive honors. The fact that we no longer use the phrase "cum laude in General Studies" creates confusion about the distinction between a student who earns a degree cum laude in, say, Psychology, and a Psychology concentrator who graduates cum laude based solely on his or her grade point average. Worse, the existence of the degree cum laude based on grade point average devalues those cum laude degrees awarded on the basis of the completion of an honors track and on a concentration recommendation: faculty members report that students who undertake a more arduous course of study understandably resent receiving the same level of honors as their peers who do not, and that faculty members consequently feel great pressure to award a magna to any student who capably completes an honors track. Among the Faculty, there seems to be widespread agreement that our standards for honors need to be reexamined and articulated so as to defend the value of a degree with honors. The EPC and Faculty Council share this view, and reviewed some alternative methods for awarding honors. Principles about Honors The EPC and Faculty Council agreed that our current standards for honors were too low, and that honors should be used less to denote completion of a particular program of study than to distinguish that portion of the class who had completed their studies particularly well. The EPC also agreed that the structure of honors should be pyramidical: that is, more degrees should be awarded cum laude than magna cum laude, and more awarded magna cum laude than summa cum laude. The EPC also agreed that, as a rule, honors should be awarded on the basis of distinguished work in both the concentration, recognized by the concentration's recommendation, and in the "general education" and elective portions of the student's course of study, supported by College-wide standards. That said, following the Faculty discussion in March, the EPC also agreed that we should nevertheless preserve one type of honors (the degree cum laude based solely on GPA) for students who had not done an "honors track" but who have exceptionally high GPA's. Recommendations about Honors The EPC began by agreeing that the GPA cut-offs for honors were currently too low and should be raised. It thus examined various possible cut-offs for the magna and cum degrees in field, aimed at restoring the pyramidical structure to honors and at more realistically reflecting faculty views about the threshold of "honors" work. It proposed to raise the bar for honors to a B+ average for a cum degree in field and to close to an A- average for a magna. Given the concern of faculty to preserve a type of honors for those exceptional students who choose an unconventional program but perform exceptionally well, it proposed to preserve the degree cum laude based solely on GPA but to set the GPA bar at the level of the magna. Faculty Council examined, and was favorable to, the EPC proposal. Members were concerned, however, that the decision to set new, higher, GPA bars might itself be inflationary, and would lead students to calculate their GPA's and strategize about their courses even more carefully than they do now. Faculty Council thus favored preserving the rough distribution in honors achieved by the EPC's GPA cuts, but argued that this distribution should be expressed in terms of percentages. It recommended, then, that the degrees summa cum laude and magna cum laude together be awarded to 20% of the class (the summas to comprise, by existing Faculty legislation, between 4 and 5% of the class), and the degree cum laude in field to a further 30%. This proposal also received wide support from the Faculty Council and will come before the Faculty for a vote on May 21st. The following two tables illustrate the impact of these changes had they been applied to the past three graduating classes. (The percentages are not exact because some students have identical GPA's.) The second table also indicates the location of the GPA cuts for magna and cum laude in field, on both the existing 15-point scale, and the proposed 4.0 scale. Actual honors distributions Graduating Class 1999 2000 2001 Summa (actual) 5.17% 4.89% 4.55% Magna with Highest Honors (12.0) 3.48% 4.12% 4.19% Magna (12.0) 29.67% 30.29% 31.62% Cum in Field (10.5) 27.45% 24.69% 24.01% CLGS (11.0) 22.58% 23.98% 25.57% None (actual) 11.65% 12.02% 10.06% Proposed honors by percentages: summa as existing, summa + magna = 20%, summa + magna + cum in field = 50%, CLGS = those not recommended for honors by their concentration with GPAs at or above the magna cut but no more than 10% of the graduating class; magna with highest honors will continue to exist, but does not have a distinctive GPA cut. Graduating Class 1999 2000 2001 Summa (up to 5%) 5.17% 4.89% 4.55% Magna (15%) 14.93% 15.19% 15.51% Cum in Field (30%) 30.04% 30.08% 30.00% CLGS (lesser of magna or up to 10%) 2.53% 2.66% 2.57% None (effectively, at least 40%) 47.33% 47.16% 47.37% Magna Cut (15-point scale) 13.677 13.759 13.750 Cum in Field Cut (15-point scale) 12.690 12.733 12.767 Magna Cut (4.0 scale) 3.644 3.668 3.668 Cum in Field Cut (4.0 scale) 3.414 3.427 3.437 It is worth commenting briefly on the impact of these changes. First, under this system, the portion of the class graduating without honors would increase substantially, being never smaller than 40% and most likely around 47%. The number of non-honors graduates would thus increase by a factor of four or five. If such a system had been in effect in 2001, approximately 60% of that increase would come from the former "CLGS" category; almost 90% of the CLGS graduates would have graduated without honors. The remaining 40% of the increase would have been made up of students cascading down from cum laude; about half of those who graduated cum laude in field would, under this system, have cascaded down to no honors. The pool of students who graduated magna cum laude would have been similarly affected, with almost 60% of those degrees cascading down to the level of cum laude. This proposal would, then, accomplish two major goals: it would restore the distinctive and elevated status of the magna degree (about which the Faculty expressed considerable concern); and, by restricting the numbers of honors degrees based on GPA alone and by raising standards, would restore the honor of the degree cum laude in field. A few questions are likely to arise about this proposal. First, why would one wish to fix percentages in this way, rather than--as with GPA cuts--allow the proportion of the class graduating with honors to change over time? Faculty Council favored percentages for two main reasons: because the system would deter excessive calculation by students; and because it would be neutral on grading effects. Put simply, because the proportion of honors would be determined by percentage, it would not be affected by either grade inflation or deflation. It is worth adding, however, that percentages accord well with the view expressed by many members of the Faculty that honors should be used to distinguish the best students among those students in the College at any one time. We must also consider any unintended consequences or possible perverse incentives. Here, the main concern is the incentives we offer students for thesis writing. At present, most students attempting an honors track (which will often involve writing a thesis) can be fairly confident that they will graduate with honors. Had this system been in place last year, about one-fifth of those students graduating with honors in a particular field would have cascaded to no honors. If we were to assume about half of those students would have been in concentrations requiring a thesis for honors, about 100 students would have written a thesis and graduated without honors. It is possible, then, that students may elect not to write a thesis under this system, since the thesis would no longer be a near-guarantor of a degree with honors. Faculty members have mixed views about such an effect. A number of faculty members expressed the view that too many students write theses "for the wrong reasons" and that too many poor theses are written as a result. Insofar as this system would discourage less motivated or less gifted students from attempting a thesis, some of the Faculty would welcome those effects. On the other hand, some faculty members believe firmly in the value of the thesis as a learning experience for most or all students, and would regret to see numbers diminish in any way. It should be noted, however, that concentrations can discuss their views on this question, and can choose to support thesis writing in other ways (even, should they choose to do so, by requiring theses of all concentrators). Our current system of basing Latin honors on initial recommendations from the concentrations also allows concentrations to place more emphasis on that initial recommendation. While only those students recommended for highest honors but graduating magna have the higher concentration recommendation noted in the commencement program, concentration recommendations for English honors are noted on all transcripts. A student who graduates without honors but fulfills a concentration program in such a way as to earn an honors recommendation thus has that distinction noted on his or her transcript. As larger proportions of students cascade from magna to cum and from cum to no honors, it is safe to predict that both students and faculty members may wish to place more emphasis on the concentration's independent assessment of the student's work, and not just on the final Latin degree. A final point should be made about the possible effects on the interdisciplinary or honors-only concentrations. Under such a system, students with lower overall GPA's in these programs, like students with lower overall GPA's in honors tracks in the departmental concentrations, would graduate without honors. Neither students nor the Faculty should view that prospect with alarm. The distinction "honors-only" is meant as a description of the course of study offered by the concentration (that is, the concentration offers only a track, the completion of which will make the student eligible for honors) and not as a description of the outcome for any given student. Too often, however, both students and faculty members assume that such programs should be available only to those students committed not simply to completing such an honors program but to completing it in such a way as to achieve "honors". These assumptions have contributed to inflationary pressures, driving faculty and tutors to leave no stone unturned in seeking to make sure that all students in honors concentrations and honors tracks do indeed pass muster for a cum. In the future, since some portion of concentrators in all concentrations are likely to cascade and since no one will be able to know in advance the likely GPA cut, all concentrations should admit students based on their commitment to the program (including, when that is required, the thesis), and not their likely level of achievement. This proposed change is a large one. It significantly raises and clarifies our standards for honors. Its results should be watched closely. Particularly because we have historically used honors to motivate our students to attempt more rigorous programs (particularly the writing of theses), we should be vigilant in monitoring any possible adverse effects on the rigor of students' programs. Should such a result be apparent, and perhaps even if not, it may be worth asking whether we should reconsider the division between "honors" and "non-honors" tracks, and between "open" and "honors" concentrations, and simply craft a single set of requirements (whether or not that would include a thesis) for all students within a given concentration. Because this change is a large one, it is proposed to bring it into effect for the class of 2005 (that is, current freshmen). It should have no effect on concentration choice, but if some students feel that they would have chosen their concentration differently had they known they would fall under this new regime, it seems wise to begin implementation with those students who will begin work for their concentration in the fall.