bobgian@psuvax.UUCP (Robert S. Giansiracusa) (01/01/84)
This past fall I taught Penn State's first undergraduate Artificial Intelligence course. Interest was higher than average, but lack of motivation was still a big problem (not unusual at large state schools). I considered several "unusual" grading policies, and I post them here in hopes that this kicks off a discussion about the function of grades in education in general. 1. Keep guaranteed minimum grades (based on exams so far), allowing some upward mobility but no downward (this is what I actually did). Pro: those with high grades content those with low grades have SOME hope of improvement Con: those with high grades have little motivation to do anything those with low grades feel little hope attendance low, interest low 2. Keep current grades as TENTATIVE only, subject to upward as well as downward mobility (this is the usual default). Pro: lows like it motivation improves Con: highs hate it 3. Give all Cs. Pro: grade competition and concern removed Con: anger and hatred (toward me!) 4. Give all As. Pro: popularity Con: lack of incentive possible administrative repercussions 5. Announce all As given, then give REAL grades based on effort and initiative demonstrated AFTER that announcement. Pro: filters out those who really want to learn from deadwood Con: resentment and lack of credibility when students find out the true "scoop" 6. Base EVERYTHING on a single final exam. Pro: uniform standard no arbitrariness in overall policy Con: possible arbitrariness in how final is graded "All eggs in one basket" 7. Base EVERYTHING on a project. Pro: those who do it learn a lot Con: those who do it spend tremendous time on it those who can't afford the time fail 8. Purely subjective evaluation. Pro: might be fairest scheme, in reality Con: gives APPEARANCE of arbitrariness students have difficulty internalizing criteria for success Reactions or comments, anyone?? -- Bob -- Spoken: Bob Giansiracusa Bell: 814-865-9507 Bitnet: bobgian@PSUVAX1.BITNET Arpa: bobgian%psuvax1.bitnet@Berkeley CSnet: bobgian@penn-state.csnet UUCP: allegra!psuvax!bobgian USnail: Dept of Comp Sci, Penn State Univ, University Park, PA 16802
spaf@gatech.UUCP (Gene Spafford) (01/01/84)
Newsgroups: net.cse
Subject: Re: Grades -- do they help or hurt??
References: <387@psuvax.UUCP>
There is an interesting discussion of grades and motivation buried
somewhere in the middle of "Zen and the Art of Motorcycle Maintenance"
by Robert Pirsig. It might be worth finding.
Let me ramble on a bit about grades in general, and then I'll relate my
own experiences so far. There may be something worth discussion in all
of this. As background, I'm finishing up my PhD in computer science
here and I hope to find a teaching/research position once I graduate
(sometime this year). I've taught 6 classes (3 each of 2 courses, one
which I designed myself), and I've taken many, many more.
Grades are supposed to serve one basic purpose -- indicate mastery of
the material in the course relative to other students and relative to
the "average." That is, a grade of "C" is usually taken to indicate
that someone has mastered the material at the level of an "average" or
satifactory but minimal level. It is not supposed to be a measure of
how hard someone works, unless that is a specific requirement for the
course.
Supposing we know what we wish to measure, how do we get that
information? Some people may master the material very well, but be
unable to express it. At GaTech, we have a significant population of
non-native speakers of English in the graduate program. These students
are at a disadvantage when asked to demonstrate their knowledge through
something like an essay exam. Then again, a number of undergraduates
I've taught can't compose their thoughts in correct English, either.
Therefore, something like an essay exam or a phrase completion exam may
not accurately reflect mastery of course material.
Some students don't function well on tests. They are methodical
thinkers, and they may do quality lab work, but they just cannot work
quickly when confronted with a test in a short time period. Then there
are people who go into an actual panic when confronted with an exam.
I've seen cases of students actually breaking out into tears in the
middle of an exam due to stress. And everyone has their off days where
some things just don't go right at all. How do we measure their
knowledge without exams? In fact, what kind of exams do we give? We
teach that it is important to research an answer before trying to
implement it, it is important to design a program on paper before
coding it, yet we give exams where we expect students to come up with
answers under a deadline and without the usual resources. Tests aren't
always an appropriate tool for measurement.
As instructors, we also have to deal with the pressure of grades and
grade inflation. For instance, at Georgia Tech, a grad student with
less than a 3.0 average (our of 4.0) in any quarter is put on academic
warning for the following quarter. Thus, if a student takes only one
course, and gets a "C" in that course which indicates that they did a
satisfactory job, they are put on warning for a quarter! Problems with
scholarships, work-study programs, grad school entrance requirements,
and many other pressures all effect the professor and the student in
adverse ways. I've had students drop a course I was teaching in which
they were very interested (and needed), but they could not finish it
because the knew they could only get a B or C as a final grade.
There is another kind of pressure, too. If you tend to grade severely,
you may attract fewer students. This kind of reputation is possibly
undesireable, especially if the FTE count of your students has an
effect on the funding of your department. At the same time, giving
many good grades results in a different kind of reputation which
eventually cheapens the very grades you give. You also need to worry
about consistent grading with respect to other instructors teaching the
same course at different times.
Finally, I don't think it is necessary or possible to motivate the vast
majority of students. The best we can hope to do is increase the
motivation of the best students, and attempt to light a fire under the
students with the potential to benefit from it. Many people go to
college because Mom and Dad want it, because they think they'll make
big bucks from it, because they'll find a spouse there, or any of a
number of other reasons. They are paying tuition to take up space.
Some of them may eventually develop a love of learning, but we, as
instructors, cannot expect to instill such a desire in the hearts of
all of our students. Especially not in the students who are just
taking our course for 3 elective credits and all they want to do is the
minimum to pass.
Okay, so I've outlined some of the problems (as I see them). Now let
me say what I've done for grading. My grades are based in the
philosophy that there should be 3 or 4 grades -- satisfactory,
unsatisfactory, excellent, and perhaps something between satisfactory
and excellent. These correspond, roughly, to "C," "D," "A," and "B."
"F"s are for students who don't show up, or fail to turn in some of the
work required for the course.
When I taught an undergraduate course in computer systems, I had
average enrollments of 40-90 students. The course was the last
required course in the major, and was a prerequisite for most of the
interesting senior-level courses. The course was to be an major
introduction to operating systems, compilers, language systems, virtual
memory, and other odds and ends. It was 2 quarters worth of work in 1
quarter, and I got quite a reputation as a slave-driver. However, out
of about 160 students I only failed one, and I suspect that I gave
about 40 or 50 A's. I based grades on a quarter-long project which was
done in 4 parts (each graded separately), homework, 2 in-class tests
and 1 or 2 pop quizes, a final exam during finals week, and extra
credit work. The tests counted for 45%, the project for 45%, and the
homeworks and the rest for 10%. Final grades were somewhat flexible
depending on my judgment. That is, someone doing a spectacular job on
the project and in-class exams, but messing up the final, would have
the final discounted in my final averaging procedure. Usually, final
grades were pretty obvious once the body of work was examined.
Another course I taught was a special topics course in computer
organization and architecture. The course was cross-listed as
undergrad and graduate, and enrollment was with instructor's permission
only -- I wanted to make sure they were warned about the amount of work
involved. I have been told that the course gained the reputation as
one of the three toughest the department offered, yet I always had a
full enrollment. Therefore, I don't believe that any of the students
who took the course had motivation problems. I told them at the start
of the course that everybody was starting with a "B" and what they did
from there determined their final grade. That is, I expected average
performance to be at a "B" level compared to any "random" group of grad
students who might be selected to do the course. I never assigned a
letter grade to anything I passed back. I figured the numerical
scores, and posted the averages and ranges on the board. My usual
breakdown was "very good," "good," "try harder," and "come talk to
me." On one exam, I put the numeric scores and little foil stars (like
back in first grade) as comments. Final grades were, again, pretty
obvious. That is, there were some very obvious break points between
grade levels. People who put in extra effort and did extra work got
A's, while those who just did enough to get by ended up with C's.
A common point to both courses was that I offered many opportunities
for reasonably challenging extra-credit assignments. Individuals with
poor scores could raise their grades, and the advanced students could
challenge themselves (and me!) by doing extra work. I think almost
every course should encourage that. I also made it known that I was
open to student suggestions for extra credit projects, or redesigning
the required project to fit their own particular styles (within reason,
and with consultation). Teaching is supposed to encourage growth, not
force everyone into the same mold.
In the end, it comes down to the fact that you are teaching the course
instead of some machine, because you have judgment. You have observed
your students (hopefully), and can make adjustments to whatever numbers
you generate as grades. We should never lose sight of the fact that
each student is an individual with different strengths and abilitites.
If we are to be effective as teachers, we need to encourage mastery of
the material, and not mastery of the system. Let your students know
that they are paying for the course in time and money, and they are the
ones responsible for what they learn. You are there as a guide with
experience and knowledge. They can use that resource, and gain from
it, but it won't be forced on them. The ones who gain get the better
grades. The ones who let the opportunity pass them by get the poorer
grades.
--
Off the Wall of Gene Spafford
School of ICS, Georgia Tech, Atlanta GA 30332
CSNet: Spaf @ GATech ARPA: Spaf.GATech @ CSNet-Relay
uucp: ...!{akgua,allegra,rlgvax,sb1,unmvax,ulysses,ut-sally}!gatech!spafnorskog@fortune.UUCP (01/02/84)
#R:psuvax:-38700:fortune:2900001:000:449
fortune!norskog Jan 1 14:27:00 1984
I went to U.C. at Santa Cruz, where we didn't have grades.*
I would advocate this approach: everybody gets the same grade,
unless they flunk. This should encourage a little co-operation,
among grade-vipers.
Lance Norskog
Fortune Systems
{hpda,harpo,sri-unix,amd70,ihnp4,allegra}!fortune!norskog
*(Or frats/soros, pro sports teams, very much grant-oriented graduate
work, or any other of the disgusting evils of higher education today)derek@sask.UUCP (01/02/84)
My personal favourite grading technique was one offered in one of my classes. The prof offered the students a choice as to what marking scheme they wished to be graded with. o 100% of final o 100% on labs and assignments o 100% on papers or any breakdown of the above. Each student was graded according to individual choice. It was also possible to ammend your selection at any time (up to the last class). sask!derek
riddle@ut-sally.UUCP (Prentiss Riddle) (01/02/84)
I found this summary of different grading schemes very interesting. I
would like to add an additional argument against scheme #6, "Base EVERY-
THING on a final exam": this is extremely prejudicial to people who, for
one reason or another, know the material but perform poorly on tests. Some
students suffer from what is known as "test anxiety", and have severe
emotional and even physical reactions to test-taking which prevent them
from doing their best on exams. (I had a mild case of this in high school,
and I know what a vicious circle it can be: I was a straight A student,
but whenever I had to write an in-class essay, my hands would shake so much
that I couldn't even read my own handwriting. Soon the fear of the fear
itself was enough to trigger the reaction.) Other students may not be made
particularly nervous about taking exams, but still don't do as well when
expected to shoot from the hip on an exam as they would on a paper, a
project or a take-home. As anyone who has prepped for an SAT or a GRE can
tell you, test-taking is itself a skill, and grading based entirely on
exams gives unfair advantage to those who may not have an outstanding
knowledge of the material, but who have mastered the art of taking tests.
Last year I had a bit of exposure to a very different grading scheme when I
spent two semesters as an exchange student in Munich. The German university
system seems bizarre from the American point of view. First of all, there
is really nothing which corresponds to the American undergraduate program at
German universities: students are expected to attend graduate-style seminars,
do research, and listen to lectures on the intricacies of their professors'
own research from day one. This is partly justified by the fact that German
students are a year older than Americans when they begin at the university, and
German universities accept only the cream of the school system (trade schools
and apprenticeships are the usual route to many careers which call for a
university degree in the U.S.). Nevertheless, I found that many German
students I talk to spent their first two years or more just trying to figure
out their place in the chaos around them before they could begin to study
seriously.
In Germany, there is no such thing as failing a course. Students are required
every semester to sign up for more courses than it is humanly possible to
attend (the equivalent, I would guess, of 25 or 30 semester-hours at an
American university). Out of those, they pick a few classes to go to
regularly, the number varying a great deal according to how serious the
students are. Many of the courses are simply lectures, for which no credit is
available. Others are seminars, in which the students may do anything from 0%
to 100% of the talking, depending on how the professor wants to organize
things. In some departments, seminars are chummy little sessions in which
students and professors wrangle over problems of mutual interest, but I once
tried to attend an introductory philosophy "seminar" for which over 100 people
showed up! Attendance typically drops over the course of a semester, until a
lecture which was packed during the first week may have only half a dozen
students at the end.
At some point, students decide which classes they want to try for a grade in.
The number varies a great deal here, too, but among my acquaintances, two or
three successfully completed courses in a semester was considered a hefty
load. There is no minimum requirement, although a student who goes several
semesters without making any progress toward a degree may be kicked out.
Getting a grade in a course usually involves writing a single, long paper,
sometimes an exam, and more rarely may require that the student gave one of
the lectures in a seminar earlier in the semester. Homework, quizzes and
smaller projects are extremely rare (although I am told they are common in
"lab" courses, which are of secondary importance to the courses described
here and which are rather sparsely offered by American standards). Thus
grades are based almost entirely on a single spurt of activity at the end of
the semester.
Not only that, but the real measure of a student's performance is delayed
even longer: grades don't matter as much as comprehensive exams, which
come up once every few years in a student's career. There are three major
sets of exams: a qualifying exam after two or three years, an exam for the
master's degree a couple of years later, and a doctoral exam several years
after that. Each student must have a major and two minors through the
master's level, and a major and a minor for the doctorate. Latin is still
a requirement at all levels, although students may successfully absolve it
early on. And, of course, a thesis or a dissertation is also required for
a degree.
The near total lack of pressure most of the time, punctuated by deadlines when
all the knowledge the student is supposed to have been accumulating suddenly
comes due, has exactly the effect you would expect: procrastination followed
by furious activity and/or despair. Suicide is a serious problem among
German students; I lived in a highrise dorm in Munich from which, regular as
clockwork, students would fling themselves at exam time each semester. Some
students learn to handle it, to pace themselves so that they are ready when
deadlines arrive without having to panic. Such students are rather shocked
if they come to the U.S. and see the day in, day out little pressures of
our undergraduate student life. They don't like being treated like children
and told exactly when and how they must learn a given set of material, and
I understand their point. Still, I wonder about the others, those who sink
rather than swim; are their losses worth it? As someone who has serious
procrastination problems himself, I am glad to have grown up in a university
system which provided me with enough structure to get things done.
I hope that you'll forgive my rambling (and any inaccuracies of detail which
the Germans out there may be able to correct). My point, I suppose, is that
there are models out there for the "all the eggs in one basket" school of
grading, and that I don't particularly like their results.
----
Prentiss Riddle
("Aprendiz de todo, maestro de nada.")
{ihnp4,seismo,ctvax}!ut-sally!riddlebobgian@psuvax.UUCP (Robert S. Giansiracusa) (01/02/84)
Thanks for a very interesting posting (..gatech!spaf). I have read the discussion in "Zen and the Art of Motorcycle Maintenance", if fact, one of my students gave a xerox of it to me! Fascinating approach! I may try it next time I teach a large lecture class. Our backgrounds sound very similar: I'm also finishing up a PhD in CS at MIT while working as an instructor at Penn State (my RA funding at MIT ran out and I couldn't afford tuition!). Also expect to finish in next 6 months. Have only taught 4 classes (3 different courses) so far, last two of my own design. This spring I'm teaching another two originals, the one I described in a posting to net.ai and net.cse yesterday. (I described it as one but it is listed as two here, one "theory" section as a grad seminar and one "lab" as an undergrad project class. Purely administrivia.) > Grades are supposed to serve one basic purpose -- indicate mastery of > the material in the course relative to other students and relative to > the "average." That is, a grade of "C" is usually taken to indicate > that someone has mastered the material at the level of an "average" or > satifactory but minimal level. It is not supposed to be a measure of > how hard someone works, unless that is a specific requirement for the > course. I "sort of" agree, but not fully. In fields where most of what one "masters" is obsolete in 5 years, how important is "mastery"? Actually, a better question is "mastery of what?". I feel that grades should reflect a student's capability to deal with future problems in the domain in a mature and intelligent manner. Yes, "mastery" of technical details of the moment helps, but more important is inculcation of an attitude of reverence for learning and the ability to "master" FUTURE information. IE: self-teaching. I firmly believe that NO TEACHER CAN "TEACH" -- ONLY STUDENTS CAN LEARN. You (..gatech!spaf) correctly point out many of the problems students have in expressing what they have learned (poor English [be they native speakers of it or not!!], poor "mathematical maturity", poor interest, exam panic, etc.). For these reasons I try to grade on my subjective assessment of many factors, rating motivation and ability to learn on one's own among the highest. I try to give students AMPLE opportunity to demonstrate these characteristics. Like you, that includes optional projects (of student's own design), different kinds of exams, MUCH personal discussion ("my office hours runneth over"), and so on. The technique I used in my last class (the undergrad AI course also described in a posting about a week ago to net.ai) was to guarantee minimum grades at mid-semester based on exams to that point. Students knew exactly what grade (minimum) they would get, and I offered many options for increasing them. Those satisfied could "rest on their laurels", and some did so. Interesting fact: MOST of the students with guaranteed 'A's at that point continued participating (discussing in class, writing papers, doing projects) very energetically, despite the fact that doing so could have NO effect on their grades! One advantage of this system: I could with a very clear conscience fail about 20% and give 'C's to about 30% (I gave no 'D's at all - I regard it as nothing but a "consolation prize".) They knew their progress and had 6 weeks to act on it -- if they failed to do so, I won't take the blame! Also, my exams were somewhat wierd. (I posted the exams themselves and a commentary about them about a week ago to net.ai.) They were designed to "disarm" exam panic by stimulating interesting discussions rather than to "measure performance DO-OR-DIE". Some liked them and said that was the best part of the course; others hated them AND ME with a passion!! > You also need to worry about consistent grading with respect to other > instructors teaching the same course at different times. Fortunately, that AI course had only one section! > Finally, I don't think it is necessary or possible to motivate the vast > majority of students. The best we can hope to do is increase the > motivation of the best students, and attempt to light a fire under the > students with the potential to benefit from it. Many people go to > college because Mom and Dad want it, because they think they'll make > big bucks from it, because they'll find a spouse there, or any of a > number of other reasons. They are paying tuition to take up space. > Some of them may eventually develop a love of learning, but we, as > instructors, cannot expect to instill such a desire in the hearts of > all of our students. Especially not in the students who are just > taking our course for 3 elective credits and all they want to do is the > minimum to pass. Sad, but oh so true! Especially at ESU (Enormous State U - see Doonesbury). I think the real problem is not grading per se, it is (1) finding and attracting the brightest and best students and getting them to WANT to take our courses, (2) making it worth their while to do so, (3) lighting fires under them, and (4) hoping that we can be good enough entertainers [this takes real work] that even the poorly motivated among our students can at least enjoy the course! -- Bob Giansiracusa (Dept of Computer Science, Penn State Univ, 814-865-9507) Arpa: bobgian%psuvax1.bitnet@Berkeley Bitnet: bobgian@PSUVAX1.BITNET CSnet: bobgian@penn-state.csnet UUCP: allegra!psuvax!bobgian USnail: 333 Whitmore Lab, Penn State Univ, University Park, PA 16802
neal@denelcor.UUCP (Neal Weidenhofer) (01/09/84)
************************************************************************** >I feel that grades should >reflect a student's capability to deal with future problems in the domain >in a mature and intelligent manner. Yes, "mastery" of technical details >of the moment helps, but more important is inculcation of an attitude of >reverence for learning and the ability to "master" FUTURE information. I agree totally. I maintain however, that the student's capability to deal with present problems is the best predictor we have available of his future capability. Likewise with mastery of current vs future information. My own grading schemes tend to be fairly simplistic--about 50-75% projects & about 25-50% exams--throw out the lowest one or two items--and apply 90-80-70-60 to the average of the rest. I don't normally apply any kind of a "curve" but neither do I ignore feedback about a test question or a project being too hard or too easy. I realize that there's some potential unfairness there and that I might not recognize an Einstein. If I see an obvious problem, I reserve the right to correct it (I tell my students this at the beginning of the course). It also helps to realize that Einstein wasn't recognized as "an Einstein" right away and it didn't seem to slow him down much. Regards, Neal Weidenhofer Denelcor, Inc. <hao|csu-cs|brl-bmd>!denelcor!neal