Declining Standards in Higher Education

In a paper entitled, “Leisure College, USA”  Philip Babcock and Mindy Marks have documented dramatic declines in study effort since 1961, from 24 down to 14 hours per week.  This decline occurred at all different sorts of colleges and is not a result of students working for pay.

At the same time, colleges are handing out better gradesIn other work, Babcock presents strongly suggestive evidence that the two phenomena are related.  That is, lower grading standards lead to less studying.  They also lead students to give better course evaluations.

To me this looks like evidence of big problems in higher education, though I’d love someone to convince me otherwise.

Andrew Perrin has been a leader in developing an institutional response to concerns about grading.  See his original scatterplot post on the topic, “grades: inflation, compression, and systematic inequalities.” as well as the more recent scatterplot discussion.

ADDED 5/4:

Fabio at Orgtheory considers four possible explanations.  I’ll quote him:

  1. Student body composition – there are more colleges than before and even the most elite ones have larger class sizes.
  2. Technology – the Internet + word processing makes assignments much easier to do.
  3. Vocationalism – If the only reason you are in college is for a job, and this has been true for the modal freshman for decades now, you do the minimum.
  4. Grade inflation – ’nuff said.

To address them in reverse order.  Fabio thinks he can rule out grade inflation because even students in hard majors report studying less… I gather he’s arguing that have really tough (uninflated?) grading are studying less, then it seems arbitrary to posit one unnamed cause in those disciplines, and a separate cause (grade inflation) in the other discplines.  I’m not sure if that argument with that data are strong enough to convince me.  I’m not saying that grade inflation explains 100% of the change.  My guess is that it explains some of it, but that both phenomena have common and distinct causes.

Fabio’s favored explanations are vocationalism and technology.  I don’t really like either of them.  First, I don’t know that it’s true that those seeking more career oriented education do the minimum.  Second, as Fabio mentioned, they claim the dropoff is similar across courses of study (though I’m not sure how fine grained that data is).  As for the idea that technology makes studying more efficient, most of the decline in studying had already occurred by the mid-eighties, before email and the web.

A priori I would have predicted the effect was mostly explained by change in the composition of colleges and college students, but the authors claim that the trend was similar among highly competitive colleges.

Any other theories?

ADDED 5/5:

I should have mentioned this before.  The authors are analyzing different surveys with somewhat different methodologies and then attempting to make them comparable.  They lean pretty heavily on the 1961 Project Talent survey.  If that is, for whatever reason, an overestimate, the decline might be far less dramatic.  Ungated version of the paper here.

ADDED 5/6:

After a closer look at the paper, I don’t think the data is fine grained enough to show that today’s students that are similar to those who attended in 1961 (ie. privileged students at top schools) are studying less, or at least not much less.  Therefore one cannot rule out the theory that much/most of the decline is due to compositional change. I wish the authors had made their agreement/disagreement with my assessment more clear because I think it is of fundamental importance in interpreting the trend.

&rfrPhillip Babcock & Mindy Marks (2010). The Falling Time Cost of College: Evidence from Half a Century of Time Use Data NBER Working Paper (April) Other: 15954

Advertisements

16 Responses to Declining Standards in Higher Education

  1. eli says:

    It’s a great question. I think you should have kept your own explanation from the top as a possible explanation along with Fabio’s 4 — i.e. the purely internal/organizational explanation: that a major facet of this dynamic has to do with the rise of student evaluations, which leads to faculty incentives to relax standards, leading in turn to students who do less work. It would be very interesting to see if there were comparative data from US institutions that don’t use course evaluations or to see whether study effort varies depending on evaluation practices across institutions.

    I think it’s at least possible that a cultural explanation, though impossible to quantify (sorry), plays a major role here. It could be the case that education just used to be taken more seriously and was more highly valued in the U.S., and hence had much more power to enforce a local norm where students did more work. If we hypothesize that education’s cultural status has declined in the postwar period (as it has become more commonplace), that alone could have had a major effect on the kinds of study norms that teachers can maintain. It seems like it would be really interesting to compare with, say, Korea or someplace else known for making education a major cultural value.

    One last thought on analyzing these data. When we’re faced with a global decline, do we necessarily have to look for a global explanation that accounts for all the subgroups in the affected population in the same way at the same time? Or is it possible that some smaller subgroup began to study less *and then the rest of the population followed more or less through a process of simple imitation*? For instance, say that more and more students worked and hence were studying less, and then all the non-working students discovered that they could get away doing less too. It seems like a mechanism like that would be consistent with the data but would remain invisible within it.

    I’d like to see more detailed data, too: surely there are smaller subgroups who study 20 hours a week or more? I basically spent half my waking hours on my senior thesis, my last semester of college; surely there has to be more internal variation than these presentations of means indicate.

  2. Michael Bishop says:

    You say:

    I think it’s at least possible that a cultural explanation, though impossible to quantify (sorry), plays a major role here. It could be the case that education just used to be taken more seriously and was more highly valued in the U.S., and hence had much more power to enforce a local norm where students did more work.

    Perhaps there is a cultural force at work. Surely there are some things we could, in theory, measure that we’d consider evidence for that argument. If we dig deeper we might even explain why the culture changed.

    When we’re faced with a global decline, do we necessarily have to look for a global explanation that accounts for all the subgroups in the affected population in the same way at the same time? Or is it possible that some smaller subgroup began to study less *and then the rest of the population followed more or less through a process of simple imitation*?

    While I’m sure a detailed examination would reveal many different processes operating which influence study effort, I have not conducted such an examination. Though there are exceptions, more often that not similarities are caused by other similarities and differences are caused by differences. Still, I encourage speculation, especially, but not only, if its accompanied by predictions which are testable.

  3. eli says:

    Apropos of cultural explanations, I should have said “difficult” to quantify rather than “impossible.”

    On the other hand, I’m not quite sure what you mean that “more often than not similarities are caused by other similarities and differences are caused by differences”?

  4. Anonymous says:

    You say:

    A priori I would have predicted the effect was mostly explained by change in the composition of colleges and college students, but the authors claim that the trend was similar among highly competitive colleges.

    Can you explain for me why what comes after the “but” in any way contradicts what comes before it?

  5. Michael Bishop says:

    Eli and Anonymous, I think you are asking related questions (though tell me if you disagree).

    Let me take a step back. When I said I would have predicted that an average decline in studying was explained by changing composition of colleges and college students, what I was really hypothesizing was that among today’s college students, there is a group of them who are similar to the college students of 1961, and that these students study a similar amount to the students of 1961. But today’s college students also include a large number who are dissimilar from the students of 1961, and these students study far less, bringing the overall average down.

    However, the authors claim that the decline in studying is apparent at the same competitive colleges, and among students from more advantaged backgrounds. In other words, the subset of today’s students that are similar to the students of 1961 are also studying less.* This means that a change in the composition of students and colleges cannot explain all of the decline.

    Now, it is possible that a similar decline across groups has different reasons for occurring for each group, but this is theory is more complex than the theory that the same effect has the same causes. We have no choice but to apply Occam’s razor.

    Hopefully that explains what I meant by “more often than not, similarities are caused by other similarities, and differences are caused by differences.”

    Can anyone offer a counter-example?

  6. Michael Bishop says:

    *I forgot to write my footnote: after a closer look at the paper, I don’t think the data is fine grained enough to show that today’s students that are similar to those who attended in 1961 (ie. privileged students at top schools) are studying less, or at least not much less. Therefore one cannot rule out the theory that much/most of the decline is due to compositional change.

    • Philip Babcock says:

      Regarding the footnote: We wanted to make sure that Project Talent made sense. So I wouldn’t say we lean *that* heavily on Project Talent. We replicate the trend in 8 other datasets, three of which predate 1961 and are remarkably consistent with Project Talent. We, too, were very concerned with compositional changes in the college population. We concur that there’s no way to account for all unobservables. That being said, there are simply no observable changes in demographic characteristics that make this decline go away. It occurs, in fact, at selective colleges that were becoming even more selective over the time interval in question. (In the NBER paper, we did not show the 1961-81 decline for these schools because we don’t have their SAT scores in 1961. But if you simply define highly selective schools as those that were highly selective in 1981, you get very large declines for those schools between between 1961 and 1981, exactly as in the main sample.) Large declines also occurred at colleges that were becoming less selective. The decline occurred everywhere for every group. We even have cognitive ability scores in some of the datasets (including Project Talent) and the decline persists when holding constant the students’ percentile in the cognitive ability distribution. We may not be able to capture perfectly some other notion of “privilege,” but it would seem to us a subtle and very peculiar variety of privilege that proves invisible every related measure.

      • Michael Bishop says:

        Prof. Babcock, Thank you so much for commenting!  I wrote a new short post to call people’s attention to your comment and to make clear my great admiration for this work. 

        What I was trying to get at with my footnote was that it would be nice to see the study-time changes broken down by multiple categories simultaneously with special attention to the most elite students.  It is hard to imagine the engineering students at top schools ever studied much harder than they do now.  Another way of looking at it would be to see whether the distribution of study times is more skewed in recent years.  e.g. 95 percentile in1961 vs 95 percentile in 2003.  That said, your point is well taken that the decline in study-time cuts across many groups.

        Other random thoughts:
        The meaning of a given level of father’s education changes a lot over
        time.

        I take it the data don’t allow you to characterize the change within individual schools?  Could you aggregate to well-known groups of schools, e.g. the Ivies?  I’m imagining a figure with school names on the left and confidence intervals for each data time point horizontally on the right.  You could present that with and without adjustments for changes in school composition.  The same style of figure could be used to present much of the data aggregated in Table 6. 

  7. […] Declining Standards in Higher Education – via Permut – In a paper entitled, “Leisure College, USA“ Philip Babcock and Mindy Marks have documented dramatic declines in study effort since 1961, from 24 down to 14 hours per week. This decline occurred at all different sorts of colleges and is not a result of students working for pay. […]

  8. […] replies on College Slackers Philip Babcock was kind enough to reply to my previous post about his research.  This is the second time a scholar I don’t know […]

  9. […] course.  The surprising thing is that this is negatively correlated with later performance.  In my post on Babcock’s and Marks’ research, I touched on the possible unintended consequences of student evaluations of professors.  This […]

  10. […] of the American Sociological Association“), provides some more detail in his analysis: In my post on Babcock’s and Marks’ research, I touched on the possible unintended consequences of student evaluations of professors.  This […]

  11. […] Grade inflation: why weren’t the instructors all giving A’s already?? by Andrew Gelman. See my old post on grade inflation here. […]

  12. Megan says:

    It’s really a comment about this
    http://scatter.wordpress.com/2009/12/15/grades-inflation-compression-systematic-inequalities/

    How about this as an adjustment for GPAs.

    Rank the kids in the class within each grade level. Take away (rank-1)/max rank in grade level from the old GPA value.

    So if one person was top in class and 5 peple got As then their GPA would be
    4 – (1-1)/5 = 4

    The second person would get
    4- (2-1)/5 = 3.8

    The lowest rank A would get
    4 – (5-1)/5 = 3.2

    the second ranked person on a B out of 10 Bs would be
    3- (2-1)/10 = 2.9

    And if you want to make it unbiased with respect to previous GPAs then add 0.5.

    (It still doesn’t (completely) get around interdepartmental grading differences.)

    • Michael Bishop says:

      Hey Megan, thanks for the comment. Your proposal seems to convey the similar information as class rank but I think class rank (or sharing each classes grade distribution) is better because under your proposal people only get a sense of the grade distribution within a particular letter grade range. To say it another way, if you get a 3.2, they know how you compare to other As, but that means a lot less if they don’t know how many Bs and Cs were given out.

      As you point out you still face the difficulty of differences between departments, classes, and profs. I also worry that with class-rank or grading on a curve, it may discourage students from helping each other but hopefully that effect is small… especially in a large class.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: