Remember, I expect your final turnins today

[Jay McCarthy]

--
Jay McCarthy <j...@cs.byu.edu>
Assistant Professor / Brigham Young University
http://faculty.cs.byu.edu/~jay

"The glory of God is Intelligence" - D&C 93

[Daniel Haskin]

Pardon, but by what time? say, 1pm?

[Jay McCarthy]

11:59pm :)

> --
> You received this message because you are subscribed to the Google Groups
> "byu-cs-430-winter-2013" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to byu-cs-430-winter...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.

[Jay McCarthy]

I'll be doing the grading and then fixing the script so people get the
grades they deserve :)

[Jay McCarthy]

In the end, I decided to remove the lateness penalty, because I don't
feel that theorems every get easier. After that, we had the following
scores-ish:

24xx
18xx
14xx
14xx
14xx

I'm happy with everyone's work, especially given that most students
have just about doubled their score since half way through the class,
even though the bottom three are significantly less than the rest. My
desire to be satisfied makes me want to give the bottom three A-s and
the top two As, to recognize the difference. My desire to be moderate
and balancing makes me want to give the top two As and the bottom
three Bs or B+s.

I don't like to change rules generally, especially not to make things
worse for anyone. Using the original rules, the grades would be A- B C
C C-, which in some sense is pretty good and clearly what the students
who did extra work intended.

In general, I don't want to punish you for being in an experimental
class. But, if I change the scoring function to give people the grades
I want, it's not a good scoring function for future semesters, imho.

Another approach is to base the points off of the high estimates of
time (2, 10, 30, 120, 480) and not the low estimates (1, 5, 15, 60,
240). Changing this changes the points people go to [lateness included
on the left]

48xx -> 41xx
36xx -> 35xx
29xx -> 29xx
29xx -> 26xx
28xx -> 23xx

This is exciting because the function from half-way through the class
gives everyone As, like I want.

I think everyone can agree that the assignments were harder than we
anticipated and that lateness shouldn't be significantly punished, as
explained before.

I'd like your opinion on whether you feel this is fair and reasonable.
Please reply publicly or privately by tomorrow.

Jay

[Andrew Kent]

Ugh - this is tough. BTW I'm writing this from the assumption I'm 18xx (which could be wrong, just throwing that out so all my mental cards are on the table and because I don't care), so any potential bias or lack of understanding comes from this thought in the back of my head probably. Also I haven't chatted with anyone recently about the class so I don't feel I have a good idea of other peoples thoughts... but here goes:

I'll be happy with whatever happens. I worked hard most of the time and at the end have shifted focus more towards other things (finals, research, etc...).

I totally agree the assignments were generally harder than anticipated, and that lateness is best left unpunished (it's kind of self punishing by the very nature of not having it done yet even without it losing points because you're falling behind). However, without the decreasing points from lateness spurring people to move on from the old and into the new I would worry students might get bogged down / overwhelmed with old, undone problems and never catch up (whereas I would think it best to cut your losses and try and maintain pace from the current position) - but maybe that's just something students in a 400 level class have to realize and make work for themselves.

I don't want the grading to feel *too* generous. Good grief, whoever got 24xx points did almost double the work of half the class. It would seem a crime for that to be an A and half of that to be an A- (unless it's a log grading scale...) - However, as you mentioned this is an *experimental* class and there was an ongoing theme of "Hey, put your shoulder to the wheel and we'll make the grades work out" - So I could see an argument from a 14xx-er that they were doing just that and it's not their fault some crazy lunatic (ha ha, j/k) got 24xx points.

In short, I guess I would vote for the A's and B's spread.

As for a long term grading curve... I think ours was pretty close to a good long term target (other than maybe the point reqs being a little high). I do think something that encourages a 'pace' or encourages students to do 'some' proofs from each chapter might be nice. Like, a minimum required points from each individual chapter (or different based on chapter importance/size), or using the idea of 'required' and 'optional' proofs to ensure some minimum, standard coverage. I dunno, I can't help but worry about this general idea (of the class being so flexible as far as what work you must do that students get lost trying to decide when to attack what if they start to miss some stuff, etc...)

Yuck, that's a lot of thought that probably doesn't help much. Well, hopefully that mixed w/ other peoples comments is useful.

TL;DR Whatever you choose will be fine, I'm confident. 1 vote for A's and B's.

<3 Andrew

[Denna Lawrence]

I think the high estimates of time more accurately reflect how long the exercises took me, in general.  However, this would give me an A, and based on the work I did, I think that's too high of a grade.  Of course I want an A :) but I think students should be required to put in more time throughout the semester than I did in order to get an A in the class.  The work I did qualifies better for a B-range grade, in my opinion.  So for a scoring function for future classes, I think the very last one (high time estimates) is too generous.

I also don't think removing lateness penalties is a good idea because students will procrastinate doing any work and be very overwhelmed and unsuccessful by the end of the semester.  The skills and ideas in this class need some time to "marinate", so it's important to be working the whole time.  I would like to see an additional policy of some sort that further encourages students to do some work each week.  Perhaps different-starred exercises could have different lateness penalties.  For example, 1- or 2-star exercises could have a 100% lateness penalty (no credit given) after the week they're assigned, encouraging students to at least do those problems every week.  Then, 3-star exercises could be subject to the lateness penalty we used of 10% off each week after the due date.  Finally, 5-star exercises could have a 0% lateness penalty so students could go back and do these all throughout the semester for the full amount of points.  This (hopefully) satisfies the concern that the theorems don't get any easier so it's unfair to have a strict lateness policy on everything.  The 4-star exercises could fall in either of the latter two categories.  If a student does all (or most) of the 1- or 2-star exercises on time each week, I think that warrants a C. On top of the 1- and 2-stars, doing most of the 3-star exercises on time warrants a high B or so, and doing a handful of the 5-star exercises could push the grade up to an A.  Perhaps it would be possible for a student to get an A without the 3-star exercises if they do most of the 5-star exercises; this could act as some sort of redemption option for students wanting to get a better grade at the last minute.

From my gauge of how much work students did in the class, I think we would deserve the A, B, and 3 C's distribution if this weren't the first run of the class.  But as this was an experimental course, I feel fine with the more lenient A/B spread for this semester.  For future classes, I think the high-point estimates are too generous even with the 10% lateness policy in place.  I'm more in favor of a different lateness policy for different-starred exercises that encourages students to do work consistently throughout the semester.

[Jared Forsyth]

I think that makes a lot of sense. Although the problems did take longer than the "point values" would suggest, I think the grades-to-points is still reasonable. In future you could both double the points, and double the point scale -- so that 1 point would more closely reflect one minute.

I also agree with Denna's idea for varied lateness based on star-value. That would really encourage everyone to at least keep up with the class.

[Andrew Kent]

I think Denna's suggestion for the lateness problem is an interesting, rather elegant solution. *golf clap*

On Friday, April 26, 2013 7:46:01 AM UTC-6, Denna Lawrence wrote:

[Jay McCarthy]

I think this idea is brilliant and I will definitely do it next time.

As far as whether to use the high or low end of the estimates, I think
I will at least use the middle. The other thing I plan on doing is
collecting data from the students on how long it took to do something,
so I can evaluate the accuracy of the estimates.

[Jay McCarthy]

I'm going to go with the high point estimate, no late penalty,
original grading function system. I've just sent each of you an email
with your grade. I'll wait until Monday for you to let me know if
there's a problem and then post them.

The email contains a list of problems that I manually graded to be
incorrect or incomplete. You should make sure this list is accurate. I
would have sent you a reason on the day you turned it in (which is
listed in the email.)

Jay

[Jared Forsyth]

thought on "how much time do you spend" -- if you ask them to use git, and commit after completing each problem...

not totally air-tight, but might be enough to improve estimates.

>> email to byu-cs-430-winter-2013+unsub...@googlegroups.com.