An anonymous donor cashes in a $30 donation to ask:

Homework solutions from intro physics through grad school physics are available online, and while working through Jackson and Goldstein problems can be miserable without some guidance, the temptation is there to plagiarize. When you teach, do you use book-problems or write your own? Do you trust that those who are really interested in the subject will do the right thing and slog through homework like thousands before them?

An excellent question.

Homework is really a vexing issue. There's no way to really learn physics without doing physics, which means that students need to do practice problems if they're going to learn anything. There are lots of problems with this, though, starting with the fact that if you assign lots of homework problems, you need to grade lots of homework problems, and moving up through the wide variety of ways students have for making it look like they did the homework without actually doing the homework (from reverse engineering odd-numbered answers in the textbook to Googling up solutions).

How do I deal with the question of whether the students are doing the work themselves or Googling up old solutions? I deal with it by not caring about it.

I've spent an inordinate amount of time thinking about how to handle the homework in the intro classes, and have settled on a slightly complicated policy that I think works pretty well:

Homework and quizzes together make up either 20% or 30% of the grade, split evenly between the two. Homework is subdivided further, into two types of homework, nightly and weekly.

**Nightly Homework** is, as the name suggest, assigned every class, due the next class. A typical nightly assignment consists of 3-4 problems from the textbook, generally fairly straightforward plug-and-chug problems. These problems are not collected and graded, though-- students are expected to do them, but I don't check that they did.

As an incentive to do the assignment, though, roughly half of the quizzes in the term (I usually give 8-9 quizzes in the intro classes) are taken verbatim from nightly homework problems. The quiz consists of one problem, with about ten minutes allowed for solving it, and if students have done the homework, it should be trivial. I'll also go over nightly homework problems at the start of class, for those who have questions (and I have, on occasion, been asked to do the exact problem that was about the be on a quiz. Astonishingly, some people still missed it).

The idea of the nightly homework is to encourage students to work some problems immediately, while the material is still fresh. This should help them fix the basic ideas in their minds.

**Weekly Homework** is, as the name suggests, assigned once a week, usually on Friday, and due a week later. These assignments consist of 3-5 more complicated problems from the textbook, and these are collected and graded in detail. I collect homework on Friday, and grade it over the weekend, returning it (ideally) on the following Monday.

After I hand the homework back, I post detailed solutions, and students have the opportunity to re-do the assignment for half the credit they lost. Thus, a student who got an 80/100 can look at the solutions and re-write the answers to raise the grade to a 90/100. The idea here is to encourage students to take a second look at the problems, and see where they went wrong, before they completely forget what the assignment was about.

I explain all this to the students on the first day of class, including the rationale for the policies, and it generally works pretty well. Most of the students appear to be looking at the nightly assignments (quiz averages are usually pretty good), and everybody hands in the weekly homework. Probably half of the students take advantage of the re-write offer at some point in the term, usually the better students in the class.

As for the question of how the work is getting done, you'll notice that homework is 10-15% of the final grade, with quizzes making up another 10-15%. The exams (two mid-terms and a final) account for 50-60% of the final grade. The homework grade is just not that significant compared to the exams.

A student who is doing the homework and doing it well will generally do well on the quizzes and exams, and end up with a good grade. A student who isn't doing the homework will get crushed on the exams, and end up with a bad grade, whether or not they've copied homework solutions to obtain a good homework grade. A good homework grade will turn a B+ into an A-, but it's not going to make a C into an A.

As a result, I don't worry much about whether students are doing the homework by themselves or not. In fact, I make it easy for them to not do the work, if they so choose-- I encourage them to work in groups (which is an essential habit for upper-level classes), and I post the even-numbered answers on the class web page so they can check their work. I'll go over any problem at any time (though with the weekly assignments, I usually just set the problem up, and leave the details to them), and if they ask questions in class when the weekly homework is due, I'll allow them to hand it in later.

As a courtesy to my colleagues, I put the solutions and even-numbered answers on a password-protected page, but personally, I don't worry about how to enforce homework discipline. Homework is for practice, to help them learn the material, not a major factor used to determine the grade.

(Upper-level classes are a different beast, but I don' teach those as often, and the policies are less set.)

- Log in to post comments

Man, these blog topics sure are detailed! I just threw out some one-word TOPICS that I figured you would enjoy writing about.

But then your general readership is undoubtedly smarter than the average bear, and plus I am lazy.

Sounds like a far, far better homework policy than most that I encountered as a physics undergrad. As someone severely lacking in self-discipline, I could have done with some 'Nightly Homework' to cement the plugging and chugging.

I think that one should make homework a significant part of the final grade only for advanced, semi-exploratory courses in which the homework problems are set by the professor and are challenging enough that students' performances on the homework actually measure something that the course is trying to achieve, i.e. when the course is meant to push the limits of students' understanding instead of ensuring that they are competent in some basic field.

Do you practice the same devil-may-care attitude towards upper division classes as well? Presumably there's less funny business going on because the students care more about the material, but it still happens.

And what about year-to-year assignments, especially if you have repeat takers? In some of the harder classes (say, advanced quantum, E&M, stat mech, etc.) it's hard to come up with good problems (interesting, analytic, not too hard). Do you reuse assignments or come up with new stuff each year?

My husband is a stats professor and he takes a very similar view on homework as you do. The students need to do the homework to do well in the class, but it's easy to copy or "do" it with a friend. So, he has his class set up so you can't pass the course if you don't pass the exams. That way you can get perfect scores on your homeworks because you're copying them from your smart neighbor, but when you bomb the exam that you have to do on your own, you bomb the class. He's caught a couple people short that way.

[...]whether the students are doing the work themselves or Googling up old solutions?

It's pretty funny how often I see Google queries that lead to my own blog - largely focussed on microbiology - which are obviously homework questions. Sometimes, they even include the question number, where someone is obviously literally copying and pasting the entire question as-is into Google. ("2. Are acid-fast bacteria Gram-positive or Gram-negative?"...)

This sounds like a really good system- I wish I had had this while I was taking physics.

I use an on-line homework system where each student gets a different version of the problem (usually, but not always, different numbers, since they can also get different graphs or circuit diagrams). I encourage collaboration, since they can only copy the approach rather than the entire answer.

Problems come from the textbook and problem libraries, including my own library. They don't change much from year to year, but the exams do.

As you note, the exams are the real test. There one sees the perfect homeworks that turn into A grades, and the perfect homework that turns into a D or F. (It does not seem to make a difference whether they copy from a classmate or work from a textbook or Schaum's example, open book means closed mind.) One also sees poor homework turn into poor grades, of course.

I call lame on this system. Nightly homework, weekly homework, quizzes, midterms, and a final? This is the calling card of the professor who has forgotten that students have other classes too. And I really hope you're on semesters and not quarters--I have no respect for teachers who cram two midterms into a single quarter. There has to be a balance between student assessment and actual teaching, this sounds like a lot of assessment. And what happens to the poor kids who get stressed out taking tests? Sounds like they are SOL.

We spout the party line that "the students who spend time doing the problems do better", but I'm not convinced that's completely true. If we compare students who spend time doing problems and students who spend time playing Wii, sure, the students who worked the problems will do better. But what if we compare students who spent time working problems and students who spent an equal time just studying solution sets? Has anyone tried that experiment? Just because the typical physics proff learned physics by doing problem sets, that does not make problem sets the one true path to learning the subject.

I call lame on this system. Nightly homework, weekly homework, quizzes, midterms, and a final? This is the calling card of the professor who has forgotten that students have other classes too. And I really hope you're on semesters and not quarters--I have no respect for teachers who cram two midterms into a single quarter.

We're on trimesters, and the loss of your respect will keep me up nights, let me tell you.

It's interesting to note that, since moving to this system, I get fewer complaints about the workload than I did previously, in spite of the fact that the total number of assigned problems is slightly higher. Test scores, both on the exams that I write and on conceptual tests like the Force Concept Inventory, are also slightly better than they were under the previous system.

But what if we compare students who spent time working problems and students who spent an equal time just studying solution sets? Has anyone tried that experiment?

We do it all the time, though it's rare to find a student who will admit to just looking at solution sets. Students who do the homework problems consistently do better than students of similar ability who don't work the problems out themselves. It doesn't really matter whether they get the solutions from another student, from the textbook, or from the professor.

There are occasionally really bright students who can blow off the homework and still do well on native ability. This is particularly true in introductory college physics, where a good high school class will have covered a lot of the same material. In every case that I know of involving two students of similar backgrounds, though, the one who did the work scored signficantly better, while the slacker generally did worse than students with weaker backgrounds but better work ethic.

As a high school AP Physics teacher I have been using this technique for several years. I give many problems for HW and instead of the burden of collecting and grading I give a HW quiz just like you do every few weeks. It is very effective in seeing who is doing the work. I usually know anyway but the incentive of a good grade entices some students to do the work. I am glad my techniques are a good lead in for college science.

Regarding post number 9, I disagree with the premise that just seeing solution sets prepares you to solve problems more effectively than actually doing problems. It is rare that you see the same set up of a problem on a test or in real life. What we are trying to do I believe is help them incorporate the models of physics into their thinking process. This allows them to use those models to interpret the situation and apply the models and the techniques of solution they have developed in unique ways to the problem at hand. Although studying solutions in the text and in other places is valuable it must be accompanied by struggle.

I can attest that college science classes are indeed hard if you don't do homework or study. Or, um, attend class. Though a bit less difficult than math under the same conditions.

Your system sounds similar to a number of classes that I was in, though I think our physics classes had an attached lab section that had to be passed in parallel with the lecture section. I think it's a fine system; a good homework policy should create incentives, but not an undue burden.