Monday, September 10, 2012

The Curious History of "Take-Home Finals"

Much ink has been spilled about the significance of the cheating alleged to have occurred in Harvard's "Introduction to Congress" course. There has been such a run on high moral dudgeon that the supplies must be running short. What does this unprecedented scandal say about the ethics of youth, about their eroding sense of intellectual property in the Internet age, about the blurred line between collaboration and copying?

I suggested in a previous post that a deeper issue might be that large courses are sometimes run casually, or may even promote themselves as easy and fun in an attempt to swell enrollments. Students tend to take courses only as seriously as their instructors seem to be taking them. 

Over the weekend I was out of town and saw several alums at a party. They all asked the same question: What is this business of take-home exams anyway? There was a reason, they noted, that exams used to be proctored--it's called human nature. The greater the temptations, the easier the transgression, the more likely it is that people will sin.

I don't know what percentage of courses now administer take-home exams rather than sit-down exams or some other form of evaluation (seminars are often graded on the basis of research papers, for example). I suspect there are a great many more non-final-exam-evaluations today than even two years ago. They may well be the norm rather than the exception. And that is because NO EXAM is now the DEFAULT OPTION.

Until the fall term of 2010,courses gave final exams by default. Instructors could substitute a different form of evaluation if they made a timely request to the appropriate dean and explained what form of evaluation they wanted to use instead, and why. 

Starting four terms ago, the default was flipped, so that instructors are assumed to be using an alternate form of evaluation unless they make a request, by a certain date early in the term, to give a 3-hour, sit down final exam.

Defaults are well known to be powerful signaling devices, indicating what is normal and expected. For example, otherwise similar societies in which people are assumed to be willing to have their organs donated after death, unless they specify otherwise in advance, have far higher organ donation rates than places where the default is the opposite. Even though in both cases individuals have complete control over the decision, by checking a box at the time of renewing their drivers' licenses, for example, the donation rates vary drastically depending on whether you check to opt in or check to opt out. So while I don't know the numbers, I imagine the number of final exams being administered has shrunk rapidly since two years ago when the default changed.

And why was the default flipped? 

Because fall term exams were moved to before winter break. 


You see, professors were not very diligent about turning in those forms specifying what kind of exam they intended to give. The administration could not make up the exam schedule until they knew which courses were going to have final exams, so the the exam schedule tended to be posted late in the term. It was always a nuisance for students wanting to make their end-of-term plans not to know the date of their last exam. But the nuisance became a big cost item when fall term exams moved before Christmas, since airline ticket prices grow drastically as December 25 approaches. And Harvard was paying for a lot of those tickets, through its financial aid budget.  

So the solution was to change the default--no exam unless you asked for one, by a certain drop-dead date. No form to return any more, no email to respond to, unless you specifically want to given an exam.

One other wrinkle: Final exams may be given only during exam period, and other forms of assessment must be completed by the end of reading period, which precedes exam period. So courses that do not give final exams (and instead, for example, give "take home finals"), are finished earlier than courses that give final exams--certainly done for the students, and potentially for the professor too, depending on the exact timing of the alternative assessment and how long it takes to grade.

So altogether, there are pretty powerful incentives not to give final exams, and nobody is telling the faculty that they are educationally a good idea in spite of the disincentives.

I include below the memo that was sent to faculty in the spring of 2010 explaining the rationale for changing the default. I was on Faculty Council at the time and asked whether there were any educational considerations in the change. What I was thinking was not actually that students might cheat more on take home exams; I wondered whether anyone who had looked at the matter had come to the conclusion that take-home assessments are fairer, more accurate, more revealing, more educational in terms of what is learnt studying for and preparing the take home exam, and so on. As far as I can remember, educational questions did not enter into the discussion at all, essentially because it seemed the change was being forced upon us by other decisions already made.

To be sure, one thing sometimes leads willy-nilly to another to produce unexpected and unforeseen results. Still, I hope that somewhere along with the sanctimonious shock about the morals of the members of the Congress 125, someone asks some questions from 50,000 feet about what really happened here that could have been avoided. Of course mechanical changes to exam administration are not going to end cheating, but they will reduce the number of cases of students who thought they weren't cheating but learned to their sorrow that they were.

To: The Members of the Faculty
From: Jay Harris, Dean for Undergraduate Education
Date: May 7, 2010
Re: Change to default policy for Final Examinations given in courses in the FAS 
Because FAS policy requires that courses culminate in a scheduled three-hour final examination, the Office of the Registrar currently must schedule a final exam slot for each course unless the faculty member petitions for a substitution. At the start of each term, faculty are prompted to submit a Final Exam Request form to the Registrar. In order to produce an accurate schedule of final exams, the Office of the Registrar needs to hear from every faculty member before the Registrar’s Office can release a confirmed schedule for final exams, causing serious delays. As a consequence, our students and faculty are not able to plan for the end of the term until late into the semester. With the change of calendar, this problem becomes particularly acute in the fall semester when the exam period falls so close to the holidays. In the fall of 2009, our first term under the new calendar, many students who needed to purchase tickets to fly home were forced to wait until November because they needed to know when their exams would be scheduled before finalizing their plans. 
The solution to this problem is to post the final exam schedule much earlier in the term, but in the absence of accurate information from the faculty, and with the default requirement that we schedule a three-hour time slot for every course unless otherwise notified, any list of final exams produced early in the term would contain many courses that will not actually culminate in an exam. 
On behalf of the College and the Graduate School, I therefore move that, effective July 1, 2010, unless an instructor officially informs the Registrar by the end of the first week of the term of his or her intention to give a three-hour final examination for a specific course, the assumption shall be that the instructor will not be giving a three-hour final examination and no slot will be reserved for that course in the examination schedule. 
Should the Faculty vote to accept this motion, Information for Faculty Offering Instruction in Arts and Sciences would be updated accordingly. I enclose, for your review, a red-lined version of the changes that would be made to the handbook. It is important to note that faculty who wish to assign some alternative means of assessment for undergraduates would be allowed to do so, but the alternative assignment (even if it is an exam, such as a take-home exam) could not, according to current FAS policy, be scheduled to take place or be due during the Examination Period. For graduate students, Examination Period may be used for assignments when there is no final examination for the course.


  1. Final exams may be given only during exam period, and other forms of assessment must be completed by the end of reading period, which precedes exam period.

    My undergrad institution had the same rule, but at least in CS, instructors regularly the take-home final—excuse me, I mean the last homework, which is worth 50% of the semester grade—on the last day of instruction, making it due that afternoon at 5pm as required by the university, but giving an automatic 10-day no-penalty extension, which conveniently ended on the last day of finals week.

    any list of final exams produced early in the term would contain many courses that will not actually culminate in an exam

    Um... So?

  2. There are two exam administrations per day, at 9 and 2, for 3 hours. The number of course meeting times is greater than the number of exam slots. The allocation of course exams to exam slots is optimized, once it is know which courses are giving exams and which students are in those courses, to reduce the number of students having 2 exams in the same day. There are not enough exam slots to make it possible, a priori, to eliminate the possibility of a student having two exams simultaneously--we get by with a shorter exam schedule by waiting to draw up the specifics until all the information is in. So yes, that part makes sense--if you are not going to give an exam, it is helpful to know that before the schedule is drawn up, as it may make possible a better use of the exam slots.

  3. As a Harvard alum (CS, '02-'03), I recall that the rule about not having non-exam work due during the exam period was not particularly well-honored, though I wish it had been honored better. I distinctly remember a class being told that an assignment (a take-home exam in a class that had both a take-home and a 3-hour exam) was due on the last day of reading period, but if anybody wanted an extension to a specified later date, they could request one by raising their hand (at which point the whole class of 50-100 students raised their hands).

    I also liked the fall semester exams after Christmas, although I realize that preference put me in a minority of students. In the fall semester, I felt like I actually learned quite a bit studying for exams (particularly in the years when I didn't have as much work due during reading period), whereas the spring semester (with its tighter schedule with fewer holidays overall, and lack of the big gap before reading period) always felt too rushed and stressful straight up to the end to have a chance to study. (Glee Club spring tour taking up all of spring break certainly didn't help there, either.)

    1. David,
      You remember correctly. I think I will post a new item in response.

      I liked the old exam schedule better because it made end-of-term extracurricular theatrical and musical performances more celebratory. Now the actors and musicians have to transition very quickly from high-stress performance to exam-taking.

      The low point in the discussion of the calendar change was when the Dean of the College told the assembled faculty that one of the advantages of the new schedule was that students would have less time between lectures and exams to forget what they had been taught. If that was the problem, changing the exam date was not the solution!

  4. Harry,

    In addition to the financial issues, I remember the discussion at the faculty meeting focusing on a more prominent justification for the change: that the vast majority of faculty members were no longer giving final exams anyhow, and that the new default structure would therefore bring the rules into line with the current practice.

    I even think I remember (though I may be mistaken about this) that there was a kind of retroactive argument from educational effectiveness: If this is what the faculty are doing then it must be more effective; so let's make sure we're supporting it with our rules.

    Naturally, this retroactive argument is less than airtight. It is certainly possible, for instance, that the external disincentives were driving the trend. (Though notice that since the trend predated the calendar change, the relevant disincentives had long been in effect. We would have to think - cynically, but not necessarily for that reason wrongly - that the faculty had come increasingly to make educational decisions on the basis of non-educational factors.) But it's possible also that more and more faculty had come to think of final exams as educationally ineffective.

    I don't know what the research is on the pedagogical effectiveness of final exams. But I would be surprised if any uniform account across disciplines and levels is possible. Whether the rules should follow the practice or vice-versa, therefore, seems to depend on how cynical one is about the faculty practice.

    Sean Kelly

    1. Sean,

      I think you are right to think that the change was just going with the faculty flow, though I don't know that courses not giving exams were a majority at the time the change was made, much less a vast majority. (Of course, it depends on how you count courses--there are a lot of seminars.) Of course, we don't always go with the flow, do we? That is not responsible governance, self- or otherwise.

      I asked the educational question and got no answer, so it appeared that the deans did not have a view on whether final exams had advantages or not. With the deans offering no educational view and the no-exam option shortening the academic year, I think you actually don't have to be cynical about the faculty to think they saw no disadvantage and at least some marginal advantage to going without exams, and have made their decisions accordingly. I don't think you need to conclude that they were voting with their feet on educational grounds--it suffices to think they were simply responding to the lack of pushback on educational grounds and to the opportunity to get on with their non-teaching plans a little sooner.