15

An answer to a recent question about the importance of looking at a student's whole transcript got me thinking. Is there some other way of summing up a student's academic performance, other than the GPA? Something that would distinguish among the three types of 3.5 students described in Jake's answer:

  1. This student is pretty steady, but not great, and tends to get an even mix of As and Bs across all of their classes, obtaining a 3.5 GPA.
  2. This student always takes the class with the easiest grading policy whenever they can, and tends to get mostly As in the "easy" classes. When they cannot evade a "hard" class, they get mostly Cs, resulting in a 3.5 GPA.
  3. This student is very strong and gets mostly As, but there was one semester when family problems disrupted their life, and missed assignments caused them to get Bs and two Fs, also ending up with a 3.5 GPA.

In other words, is there a way of summarizing a student's transcript, that doesn't ignore so many important details?

I suspect that incorporating just class rank doesn't help much (an anecdotal, subjective contribution): my high school allowed one to choose to take classes with credit/no credit (C/NC). I chose to be graded with letter grades for a handful of classes, and received A's in all of those. I suspect this may have been why I graduated with a laughably high class rank.

  • 2
    I thought this is what letters of recommendation were for. Are you looking for some sort of other statistic? – chipbuster Dec 26 '16 at 2:39
  • @chipbuster - Yes. I wonder if anyone has come up with some other statistic. As Jake showed, three completely different students can present with an identical gpa. (Well, there's always class rank, but I don't think that helps much.) – aparente001 Dec 26 '16 at 2:42
  • 10
    Why should there be? Any statistic will necessarily conflate students with very different academic profiles. You cannot summarize 20 numbers with one number without losing information. In particular, there is no reason to believe in a ground-truth toral ordering of transcripts. – JeffE Dec 26 '16 at 5:03
  • @JeffE - I don't mind using a group of statistical measures. There is a lot of ground between a full transcript and single-valued gpa. – aparente001 Dec 26 '16 at 5:05
  • 3
    Better for what? Also, I'd appreciate it if you'd clarify what level of academia you're talking about (and making sure it's on-topic for this site). – Pete L. Clark Jan 1 '17 at 22:35
16

I believe that there cannot be any single numerical statistic or small set of such statistics that summarizes the nature of a student well.

The first reason for my assessment is that there are too many different ways one may wish to use information about a student, and too many ways that a student can have an unusual background that will not fit into those boxes well. The second reason is that any metric, once established, can (and will) be optimized against, thereby undermining its value.

Summarizing a student on one or a few metrics is no better than the much-bemoaned practices of judging faculty for tenure by a few metrics.

2

Class rank "doesn't help much": do you have an empirical basis for saying so? Let's say you are trying to see how well high school performance predicts college grades. I think the research shows that class rank and quality of high school program will enhance by a considerable margin whatever predictions you would obtain from GPA alone.

But that presumes that you are looking for a means to predict college grades. You haven't said why you want to sum up each student's high school performance. There is, probably fortunately, no single best way to "sum up" anything. That would turn something arbitrary into something universally right or wrong. Instead, systematic assessment hinges on the idea that we choose certain indicators because they are useful to predict or explain specific other things that are of primary interest. Perhaps you want to know how far a student will travel in school; how much s/he will participate in the civic process; or how happy s/he will be at your school. Each of these outcomes figures to have its own best set of indicators, i.e., its own set of factors that will furnish the best explanations or predictions.

Another way to say this is that the validity of an instrument or indicator is, strictly speaking, always described with respect to a specific use that will be made of it.

  • My question is phrased to encompass university studies. (I started thinking about it in connection with special education legal arguments -- but you can forget I said that. Really, if you review Jake's answer, that's enough motivation for the question.) – aparente001 Jan 1 '17 at 20:49
  • Disputes about eligibility for special education often center around the degree to which the student's disability has an "adverse effect" on his or her academic performance. Then you may see the parents trying to point to the two Fs, and a general decline from A's to B's and C's, and the school district defending its non-eligible decision on the basis of a gpa that doesn't clearly show the decline. Example: Two PE grades of 100 and two music ensembles of 100 can dilute the effect of the two F's. (You can forget I said this, too. I'm just explaining how I came to start thinking about the gpa.) – aparente001 Jan 2 '17 at 5:22
2

Jake Beal's answer says "there cannot be any single numerical statistic or small set of such statistics that summarizes the nature of a student well." But numerical measures of scholastic achievement are not going away, and it's better to use better measures rather than poorer ones.

This paper describes a very thoroughly worked out proposal for reforming the calculation of GPAs: Valen E. Johnson, "An alternative to traditional GPA for evaluating student performance," Statist. Sci. Volume 12, Number 4 (1997), 251-278, http://projecteuclid.org/euclid.ss/1030037959 . It's designed to eliminate the penalty that students suffer for choosing majors where GPAs are low, and to eliminate incentives for faculty to inflate grades in return for higher student evaluations. (Student evaluations of teaching have been shown to be inversely correlated with learning.) Johnson was writing in 1997, which was before the existence of web sites such as myedu that allow students to so easily choose instructors based on their lenient grading -- so, if anything, the need for this type of reform is much more acute now.

Another way that IMO grade point averages should be reformed is that there should not be such a heavy incentive for students to take a course once, drop a few weeks before the final with a W, and then take it again in order to get a higher grade. I teach at a community college in California, and the incentives for this behavior are very strong and very effective. It's perverse, because the state legislature is very concerned with encouraging students to graduate more quickly, but these incentives produce the opposite effect -- and college administrators love it, because if a student takes a course n times, the school gets their revenue multiplied by n. Students may be told that a W on their transcript looks horrible and will make them more likely to be rejected when they apply to transfer, but I don't think the system necessarily works that way. Popular majors at some schools (e.g., kinesiology at Cal State Fullerton) have extremely high GPA requirements for transfer students, and students know that if their GPA doesn't make the cut-off, their application won't even be considered. And for unpopular majors at low-status schools, weak students know that all they need to do is get an AST degree, and they're guaranteed admission, so all they need is a 2.0 GPA. My suggestion for reform would be to put some kind of cap on W grades, such as 2 per student, statewide, in our community college system.

  • Interesting exposition of a problem I was not aware of. Good analysis, interesting proposal. Seems a bit tangential here, but certainly worthy of a separate question. – aparente001 Feb 3 '17 at 18:34
  • there should not be such a heavy incentive for students to take a course once, drop a few weeks before the final with a W, and then take it again in order to have more time to master the course material? Why would you want to discourage that? – JeffE Feb 4 '17 at 15:35
-1

GPA as it is applied at most universities is not a good indicator .There are some things that could make it a better indicator .It will never be a perfect indicator .If there was less emphasis on assignments and more on Exams there would be less scope for cheating .If there were more little Exams the stats would even out more .If the Exams were written in such a way that it was really hard to get an A and really hard to get an E .When you see a class where more than half of the students get A passes how can an employer take things seriously .They should only give A passes to say 5% .Then an A is something .These days it also depends greatly on what university .If there was more moderation between universities this could help things .It has also been said that a lower grade in a harder exam is a better metric .This was the case in the New Zealand Junior scholarship exams .

  • 1
    How does this answer the question? – Tobias Kildetoft Jan 2 '17 at 20:35
  • To do a better method would require radical change which probably would not happen .This is why I suggested tweaks to the entrenched system. – Autistic Jan 2 '17 at 20:46
-5

Obviously, and I mean really blatantly obviously, getting published in a top-ranked journal, getting one (or more) patents and then starting your own company, or getting a relevant award, prize, or honor (NAS, Noble, and a couple of hundred others in various disciplines) would serve you far better. Which would you rather have, a Nobel Prize (relevant to your area of academics) and a 1.0 GPA, or a 4.0 GPA and no Nobel? The apex of the academic world is accomplishment, this is either in ideas or leadership. Just so obvious that if you demonstrate mastery of either, preferably both, then nothing else (aside from criminal or 'problematic' behavior) matters.

  • 1
    Those all seem like zero one variables. All or nothing. Either you have one of these spectacular results to report, or you don't. I'm wondering if there's a better way to boil down the grades. Something in between just reading the gpa, and poring over the whole transcript. – aparente001 Jan 2 '17 at 7:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.