Tuesday, March 09, 2010

HW policies: PROS and CONS

The last blog entry had lots of good comments about different HW policies. I enumerate them and say PROS and CONS
  1. Hard Deadline. PRO- uniform, no favoritism, can post HW Solutions or go over HW in class as soon as it is handed in. CON- there could be legitimate reasons for lateness that are short of a doctors note. CON- you want the student to DO the HW even if it will be late. CON- you need to be TOUGH to say NO.
  2. Moral Deadline (what I do, see last post). Same as Hard Deadline, but its a bit easier to say NO.
  3. Penalty for lateness. PRO- the students will still do the HW. CON- delay in posting solution. CON- slackers are still slackers. ODDITY- the penalty is supposed to discourage lateness. But it may encourage it (gee, 10% off if I hand it in one day late. OKAY, its a deal)
  4. Look at late HW only if they affect the final grade. PRO- less to look at likely, CON- Don't really want to keep track of these things. CON- student may not be discouraged from handing things in late.
  5. Only count (say) 10 of the 12 HWs, and have HARD DEADLINES. PRO- same as HARD DEADLINE. CON- students will blow off 2 HW's, possibly the last two which may be important for the final. CAVEAT- raises the much bigger question of whether to treat students like adults or like ...students.
  6. Students get x number of late days (this one was new to me). PRO- well defined rule, flexible but no favoritism. CON- delay in posting solutions. CON- keeping track of it.
  7. If you miss a HW then the others will count more (up to some limit). PRO- uniform. CON- students may still miss some HW they should do.
  8. HW are OPTIONAL. PRO- they sink or swim on their own. CON- they sink or swim on their own.
Diff topic- how much to COUNT HW? I often count it low (like 10-20 percent) so that I don't' have to worry too much about cheating. Actually I think its GOOD if students help each other but BAD if students copy each other, but it can be hard to tell.

Which of these work best? Depends alot on the school and the course and even the profs willingness to say NO.


  1. What is the definition of "late"? Intuitively, there is a deadline. But, the institution can provide levels of excuse (death in family, house burns down, accident leading to hospitalization, athletic events) that supersede my definition. These are rare is small classes; par for the course if there are several hundred students. Some students expect to see solutions and grading quickly; yet others linger in submitting, due to these validated excuses.

  2. YES- documented excuses supersede
    the rules and I have no problem with that.
    And YES- what counts may differ from school to school. The policies in the post were more along the lines of excuses
    that are BELOW that level- like your dog dying.

  3. For about two decades, I give the students option to be graded completely on
    the homework. The alternative is a more conventional grading system.
    Deal is they do more homework than if they take exams,
    they have to come in for help so I can discourage and catch academic

    I give double credit for homework that is on time. (I believe deadlines
    are to help the student achieve and not hurt those who are
    having difficulty. So I informally tell them
    I will be lenient on deadline if they come in a few days
    before deadline for help and come in each day after that.)


    "Contract Grading for Computer Science" Journal for Computer Science Education, Vol 2, No 3,4, pages 16-22, December 98.
    (by myself)

  4. You missed one.

    Students get one get-out-of-jail-free card but you tell them they are playing Russian Roulette with the system.

    If the solution gets discussed in class before they hand it in, tough! They took the roll-of-the-dice and they lost.

    And you put in a different policy for "catastrophe" (= death of parents, grandparents, etc.) which is a rare event and doesn't exactly count as "favoritism".

    Tough love is underappreciated. What you want is a "transparent" policy.

  5. Think of the worst thing that ever happened to your best student, and create a simple policy that would handle that situation. In general, I would vote for a simple, no-nonsense policy that caters to good students. You want poor students to become good students.

  6. My current policy (which seems to be working out) is to apply an exponentially increasing lateness penalty: 2^(n/6) percent off for an assignment that is n hours late.

  7. How much should HW count?


    (Most) College students are adults -- they can vote in most countries. It is reasonable to expect them to actually do the work and compare their solutions to a (detailed) solution key, and figure out for themselves if their understanding is good. In any case, there are office hours, teaching assistants, etc. to help (before and after the HW is "due").

    I am not a Professor, but have been one. I tell my own kids that I "grade" them not on accomplishment but on effort. As a Professor, though, I'd prefer to grade on accomplishment and not on effort. In fact, I prefer not to grade at all, perhaps just give an Satisfactory/Unsatisfactory grade. I don't believe Universities should be in the business of "certifying" students' learning -- if a prospective employer wants to know how good a student is, they should test it for themselves.

    To me, it is more important to help the students learn, and to do it in a professional way. For example, I think it's important for the Professor to gauge precisely how many hrs/week a student needs to put in for their course, keep it reasonably low (after all, students have other classes and a life), keep the variance of this low across weeks, and assign work that can be realistically completed in this time. This will teach them things like time management, which are very important in real life (e.g., how to break a project into units, each of which can be accomplished in 2-3 hrs of effort).

  8. My wife gives students hard deadlines, but offers a 3 day extension to any student that emails her to ask for an extension at least one day before the due date.

    This rewards students for proactive behavior, while still tightly bounding how long assignments are floating around after the deadline. The main downside is that it would be a bit labor intensive for large courses.

  9. I'm #7 ("If you miss a HW then the others will count more (up to some limit"), except that my limit is zero. If they don't want to do the homework, I'm quite happy to let students base their grade entirely on the final exam. They're adults.

  10. I don't know when MrChips was last a professor. I completely agree with the feeling, but you must also adapt to your audience, nothing else will do.

    Where I was an undergrad (Europe), we didn't even *have* homework. One final: if you pass, you're home free. One midterm, to help you in case your final is weak. (Also, no GPA, that helps).

    It was up to us to pick up books, do problems, and figure out how much was enough.

    These days, I'm just happy if my students look at the homework at any point. I don't count it for much because of that, but I'll take any lateness, without justification.

    Why? Because I tried things like assigning homework and not picking it up (basing quizzes off the homework and so on) and no one did *anything*. So I'll take the stuff copied off the back of the book (sometimes off the wrong chapter!--hilarious) and the stuff copied over the classmates if it means that the exams will be slightly less weak.

    Again, to me it's mostly an issue of audience. I would proceed differently somewhere else.

  11. "you must also adapt to your audience, nothing else will do"


    Does the legal profession go and say "eh, bar exam, whatever, if they show up let them pass"? No. Of course not.

    Why are so many academics so willing to throw the value of an education at their institution under the bus?

  12. In my class, hw counts for 30 units of final grade (out of a hundred). But: if a student gets only x of these 30 units, the remaining 30-x are "moved" to the exam.

    So, for example: If a student gets perfect score on all hw, his exam is worth 70 units. If a student skips all hw, his exam is worth 100 units. Any behavior between these two extremes is possible. Overall, hw affects only how _risky_ your exam is.

    I first knew of this system in grad school and I have used it several times since. Most of the time, most students "bite": they do as much hw as they can.

    Plus it is surprisingly easy for me to post a lot of hw with hard deadlines (whoever can't keep up, they just miss opportunities, they do not lose points) and subtract points for lateness (now `subtract' means just `postpone').

  13. "you must also adapt to your audience, nothing else will do"

    Well. Why not? I can ignore my audience and make sure they have learned *nothing* from my class. Or I can adapt.

    I have taught at many fine places, and many not so fine places. The stuff I routinely did at the nicer institution would simply blow the mind of the other guys, so why even bother to try.

    And of course, Calculus credit from Harvard and Calculus credit from the local community college are simply not worth the same thing.
    You'd be hard pressed to convince an employer otherwise.

    And this is in no contradiction with the "bar exam" argument. The unwavering(-ish) standards to pass the bar are the *destination*, making sure your students are learning something from a class means allowing them to *start* the journey. If you're way over their head, they do not even have a fighting chance to begin.