If you give a major, timed assessment (test, exam, etc.) and nearly all of your students do poorly on it — as in, really poorly, 3/4-of-the-class-failed-it poorly — do you give a re-test and let them try it again? Or do you stick with the grades they got the first time? Do you invoke some kind of wigged-out grade curving scheme (no offense, Dave)? Or what?
Fortunately this hasn’t happened to me this semester, but it has happened to at least one of my colleagues, and we have an email discussion going on right now about what to do about it. Here are my thoughts on this. (Most of this post is verbatim from my contribution to the email discussion.)
For simplicity, I’m leaving the question of curving the grades out of this for now, and focus on whether you simply have a do-over for the exam or not. With that choice as the only one in play, I have been known on very rare occasions to give retests. I keep a couple of basic criteria in mind every time giving a re-test comes to mind:
(1) There has to be widespread failure of student learning in the class as demonstrated on the assessment. That is, I do not give retests to individuals or small groups, for fairness reasons, unless there is some incontrovertible reason for it.
And more importantly:
(2) There has to be evidence, the preponderance of which points to me as the main source of failure. It’s not a good idea, in other words, just to give a retest because the grades were bad. Without knowing WHY, exactly, the grades were so bad, retests can actually do far more harm than good, reinforcing among students the notion that if everybody blows it on an exam, it’s OK because the prof will just give a retest. (He can’t fail EVERYBODY, can he?)
If it’s practical, I’d have a colleague look at the exam for a second opinion on its design; talk with my students or give anonymous surveys about their preparation strategies; look honestly at my teaching and office hours work prior to the exam; etc. If the evidence points back to me or the exam, then I’d be the first person on board for a re-test. But if the evidence points to students — the exam and your instruction were reasonable, but they didn’t ask questions, come to office hours, do the reading, study appropriately, participate in class, or some linear combination — then students must bear the responsibility of their actions/inactions, and they must live with and learn from the grades they earned the first time.
It is not automatic that a class-wide failure on an exam means that I failed as a teacher and must therefore somehow make it right. Students (anybody, really) can just sometimes be irresponsible in large groups, and it takes a large-scale wakeup call to get them on track. The responsibility for student learning is shared, but just as students can learn with a bad professor, sometimes large groups of students can fail despite the best efforts of a great professor. Therefore I have to know WHY the grades were the way they were before you can make an informed decision. Just giving a retest without knowing the “why” will make grades go up and students happier, but it doesn’t really solve the root problem or prepare students for the next exam — and it doesn’t do much service to my college’s stated commitment to Responsibility either.
Like I said, I think I’ve re-done exams five times at the most in the last ten years. Once it was because there was at least one horrible typo on a problem, uncaught until grading time, that made the problem ten times harder/longer than it was supposed to have been — my fault, and I had no qualms giving a retest. Other times, I forget the details. (I’m old, you know.)
When I do this, I give the retest as a “pop” retest — students are not warned that I am doing this — using the exact same exam as the first time through. That way, students who really studied and prepared for the exam — and who therefore still retain the knowledge they had the first time they took it — benefit the most, and those who didn’t prepare as well — and who flush knowledge out of their brains immediately following an exam — don’t benefit as much. Since it’s the same exam, I will grade using the same rubric and then refund half the difference between the first and second takes. So a person who made a 30 the first time and a 100 the second time would end up with a 65. (Assuming that their failure was not some unambiguous, abject failure of teaching on my part; if it is, then they are entitled to a full refund of credit if they can repeat the task, such as was the case with the horrible typo I mentioned.)
That’s my take. What’s yours?
14 responses to “Do you re-test?”
As I started to read this post, I had my answer ready to type. But you have it there for me! If they all mess up, I can bet it’s probably my fault. Not always, but that’s when I start looking at my own practices.
Your method is somewhat faulty IF the test given had faulty details and therefore were faulty based on your test-design. If you believe that your original test on which too many students scored low was faulty, then re-giving the same test as a pop-test is not really a way to help students achieve better scores. Some miner redesign of the test should be performed and then this redesigned version given as the pop-test for score replacement. IF you are comfortable with the design of the test as originally given, then certainly, you could try just giving this same one again as the surprise unannounced pop-test.
Actually, you could have another alternative instead of giving a retest. Maybe you can just change the method of grading scale and use statistical score information to help reassign the meanings of your fixed-percent method (the one that goes as 90% for A, 80% for B, etc…)
You gave a discussion of sensible things to consider, but your last paragraph just did not seem reasonable. You should retest, if you choose, using an altered, preferably improved, test version.
I mean that the test should be corrected before readministered. It wouldn’t make much sense to give a test a second that was known to have a serious flaw.
Very good! That’s the way to do.
G, By “the same test” he surely means “the same problems, but corrected”.
For exams I have taken or graded, teachers would assume that the average student should score 50% of the points (or some historically stable µ). If the measured average is off, some blindly rescale, assuming the problems were too hard. Others look deeper and find out which problems were to fault in particular and wether there was some teaching/problem posing problem, taking appropriate measures.
I have never experienced a retest. You probably do not want to do that with 100+ students. On the rare occasion when problems were indeed faulty, they were taken out of the 100% but points achieved on them were awarded. This is not perfect since time lost on the infeasible problem is not considered but simple to execute.
I teach HS so my situation may be too different to merit, but
I retest Every test, ..if the student wants to . I offer an alternative test to students who come and spend two study sessions in my free period. I try to make the test as equal as is possible… My goal is to use testing to help drive instruction and learning rather than just for a grade… and to try and persuade them that the grade they get is based more on the effort they apply than what I do or who their dad is or …. Tests themselves seem to be one of the few things we know that aide memory/learning (something about that acid in the gut maybe)… so I am in favor of kids taking lots of them in to help both of us know where to direct their attention next.
I think your answer to this question (and how to implement a re-test) depends entirely on your view of the purpose of assessment. If you view assessments as opportunities for students to demonstrate their knowledge, and hopefully, mastery, you will never curve the grades on a test. And, you will offer re-tests. So, maybe the question is: Which matters more WHEN they demonstrate mastery (the first, and only time) or THAT they demonstrate mastery.
And, if they demonstrate mastery, why only give them 1/2 of their points back? Does that grading practice place the emphasis on WHEN they demonstrated mastery vs. THAT they demonstrated mastery?
Danielle, for me the answer is that both the “when” and the “that” are important. I tell students that the time constraint is part of the assessment of mastery. With mechanical skills such as one finds everywhere in calculus, what I am trying to assess is if they have attained a masterful level of fluency with using the tools, or whether they’re still fumbling with them when they are needed. If they didn’t do well the first time, there’s a good chance the fluency wasn’t there.
I’ve ended up taking a different approach—never giving timed tests, but only papers, projects, and programming assignments. Grading is more difficult, but the assessments are more meaningful. I generally allow students to re-do assignments, but warn them that I hold them to a higher standard if they redo them.
This is definitely the approach I’ve started taking in my 300+ level courses and both I and the students really like it for the reasons you mentioned. I don’t know if I am quite ready to do this in freshman calculus, though.
How specific do you get with students about your “higher standard”? Do you use a different rubric or is it just a judgment call?
My grading is all somewhat subjective. I’ve tried using a rubric, but it nearly always turns out to be useless. Students find ways to mess up that I never would have imagined and sometimes surprise me by doing something good that I didn’t expect. My classes are small enough (15-20 students) that it is more work to create a rubric than just to grade the stuff free form. If I do create a rubric, I nearly always end up having to modify it several times in the process of grading an assignment. If I have to go back over the assignments after changing the rubric, there isn’t much advantage to having one. I can go back over the grading to ensure consistency just as easily without a rubric.
(If I had graders, it might be a different story, but I do all my own grading.)
Pingback: Weekend Reading: Congrats to Bowling Green Edition - ProfHacker - The Chronicle of Higher Education
I prefer discounting the exam (making it worth a smaller percentage of the final grade) when assigning final grades to giving a retest. This can be a bit tricky, as I simultaneously treat the syllabus as inviolable. So what I ultimately do is discount the exam if they did worse than their personal average and leave it be if the individual student did better on it than their personal average. It’s not that hard to set that up in the grading spreadsheet.
Regarding the individual question that took much longer than anticipated – this is an important skill for the students to learn – don’t solve a problem to completion that is taking too long. Move on. I try to part longer/harder problems at the end of exams, but it’s not my fault as a test-writer if the student allows themself to get hung up in the middle of the exam.
Pingback: A Higher Education Grading Philosophy: Agree or Disagree? « The Communicator