College students are not as intelligent

I made a comment earlier that college students, and by inference college graduates, are not as intelligent as they used to be on average. I made that comment based on what I'd seen in the General Social Survey. What I had seen was a decline in average WORDSUM score over the years (WORDSUM being a variable which records how many correct responses individuals received on a vocab test). But I'll lay out the data here.

I limited the sample to whites between the ages of 22-35. That way I get a snapshot of those who graduate from university in a particular time period, and I need to limit the sample because of the nature of my hypothesis as to why college students are less intelligent.

Below is a chart which shows the mean WORDSUM scores of those with college educations or higher, and those without college educations, with 95% confidence intervals on the bars.


As you can see, those without college educations don't change much, but college graduates seem to be getting duller, on average. Why? I think it has to do with the fact that the less intelligent are going to college. We know that the number of college graduates has been increasing, at least until the most recent age cohort. Here are the trends in the GSS:


Now, let's do a scatterplot of the mean WORDSUM score vs. % of those with college degrees. I had to omit some years in the case of the latter, since we had more data points for that.


As you can see, there's a modest fit between the two trends. 40% of the variance in the WORDSUM score can be explained by % of college graduates. Perhaps there's nothing to it, but I think there are strong theoretical grounds to expect this. As the proportion of Americans who go to college converges upon those who graduate from high school, universities will naturally begin to resemble private high schools.


More like this

My post below elicited a lot of response. One thing to point out though, which I want to emphasize: a higher proportion of smart people go to college now than in the past. How can this be? First, let's review the change in distributions of intelligence of those with college degrees (or higher) and…
The Audacious Epigone has an interesting post up, Burden of boredom borne by blockheads: This isn't just me speaking from personal experience--the data confirm it. The GSS asked respondents in 1982 and again in 2004 how often they have time on their hands that they don't know what to do with. Using…
One "urban legend" which is in common circulation among my friends is that liberals are smarter than conservatives. From my own personal experience this seems plausible, and I doubt I'm the only one as evidenced by the furious speed at which the "Bush voting states have lower IQs" meme spread…
I watched this rather bizarre Kirk Cameron video today. He's promoting a plan to distribute Origin of Species with what seems to be a scurrilous preface to college students. His argument is that college students need to know the truth about evolution. I don't know how far he'll get, but I suspect…

Are you aware that your finding directly contradicts the Bell Curve finding that the IQs of college and non-college samples have been diverging (in spite of more people going to college)? I for one think it is important whether society is becoming divided along IQ lines or not. At least one of you is wrong. Chris Brand once told me he thinks M&H are wrong on this question. I think this issue needs more scrutiny.

If your theory is the main explanation, then wordsum scores for all whites should be stable. If wordsum scores for all whites are declining, then that would imply an overall downward trend, not just a shift of some low-scoring people from the "not college" to the "college" group.

Classical education inflation. It is another proof that education does not increase IQ.

Then how do you explain the consistent scores in the no-college group? The new people going to college are coming from somewhere. Wouldn't you expect that the new people going to college would skim off the cream from the former non-collegians?

This is an interesting, but perhaps trivial, result of enrollment trends. In my own 37 year career, I thought that students were getting progressively lazier: wouldn't do assigned readings or tasks (even accepting lower grades); were inattentive in clase; and generally surprised these behaviors were "punished" by lower than expected grades.

It might be interesting to see if this is a general phenomenon or merely one aging man's fantasy. The general collapse of primary and secondary education for the great majority of working and middle class children (but not ruling class children) might be related, also.

By Bob Sykes (not verified) on 25 Sep 2009 #permalink

What? No Flynn Effect in the general population? ;-)

Razib, you have reproduced the results of some famous Stanford professors, supporting the sorting and signaling model of higher ed (as opposed to the human-capital-building one)!…

Education and Verbal Ability over Time: Evidence from Three Multi-Time Sources

Nie, Golde and Butler

Abstract: During the 20th century, there was an unprecedented expansion in the level of educational attainment in America. Using three separate measures, this paper investigates whether there was a concurrent increase in verbal ability and skills. Changes in verbal ability in the general population as well as changes in the verbal ability of graduates of different levels of education are investigated. An additional investigation of how changes in the differences between males' and females' educational attainment are associated with changes in differences between their respective verbal abilities follows. The main finding is that there is little evidence that the large increase in educational attainment has resulted in an increase in any of the measures of verbal abilities and skills.

Bull. First of all the error bars practically overlap for the WORDSUM graph of white college graduates. Nothing can seriously be taken from the graph besides non-College educated people tend to have a lower WORDSUM. Your scatter plot doesn't doesn't imply causation (R^2 = 0.38?). Measuring things like this is just really hard and saying students are just getting dumber is not a good way to correlate causation. Less professors have tenure, could that be why? School costs way more, could that be why? There are even reports contrary to yours! Sorry, but your analysis serves no need and is just needless snobbishness really. Who fucking cares about WORDSUM anyway.

Razib, due to your demonstated expertise and your historically demonstrated accuracy in presenting statistical information, I will accept your finding on it's face (Translation: You do a good job, Razib).

Is it not to be expected that as the system of higher education admits an increasingly larger portion of the overall population of citizens that the overall level of assesed intellectual functioning for such institutions would gradually decrease (marginally) over time? This is particularly the case since vocabulary assesment is quite sensitive to familial and social class variables.

The practical question is: Is this not (still) what we want to accomplish in a democratic society?

As a retired psychologist, I find nothing surprising or discomforting in this finding. I somewhat expected it. I find nothing wrong with providing higher education to more of our citizens. I see no major deliterious affect on graduate and post-graduate levels of higher education. The more intelligent cohort will continue their climb up the ladder of educational attainment and an increasing larger number of our citizens will get a better education.

A win-win situation, in my opinion.

Thanks for the information!

It'd be interesting to compare this to a vocab exam on high school seniors to see if that has changed. My suspicion is that High School has become "dumbed down" slightly as well. Which isn't to deny the point you make which I also think is at play.

As a college degree has become a requirement for a well paid job the selectivity of colleges has gone down. Graduating college is seen as a norm and something any reasonably intelligent and motivated student can accomplish. Used to be a college graduate was something special and it was expected that any who made it would be significantly above average.

The norming of a college degree, and lowering of the difficulty of the work to accommodate this norm, goes a long way to explaining the results. Add to this more and more legacy and 'buy-in' students come in without regard to qualification and the effort to maintain standards can be seen to be up hill.

The other aspect is that maintaining tough standards is difficult. Saying no to a rich and entitled kid and his family has consequences. They can negatively influence contributions to the school and can cost the school money in a law suit. Administrators, sometimes even professors, who piss off the wrong person can be fired or experience a lot of pain. So the tendency is to fold and quietly give Junior his 'A' even when he deserves a 'D'.

Of course one good grades are seen as an entitlement by virtue of paying the school it become even harder to crack down and the students themselves start to fail to see the point of putting in any effort.

In the end even prestigious universities can become virtual diploma mills where you pretty much can just send in a check and get your sheepskin. Which would be fine except for the fact that the diploma mills are not even handed in who can buy a degree and who has to work for it.

"Bull. First of all the error bars practically overlap for the WORDSUM graph of white college graduates."

Whats that gotta do with anything?

Of course this graph proves Razibs point. By the way the decline in IQ ( or G) of college graduates must be even greater than this as - at least one would hope - that more education would push the wordsum scores higher. Nurture overcoming nature.

It would be interesting to see if the stil remaining tough degrees ( engineering, law, medicine etc.) had this convergence with non-graduates. Since they have to select from the best - probably not.

Is there an online WORDSUM test available?

By Veronica.abbass (not verified) on 25 Sep 2009 #permalink

re: selectivity. many of the ivy league schools got really selective in the 1960s in comparison to the past. so what is happening is a two-tier system.

It seems to me that these data may well say something about vocabulary, but it does not follow that they are a meaningful measure of intelligence. What a person learns is very different from a person's capacity to learn.

By Catharine (not verified) on 25 Sep 2009 #permalink

Inductivist June 2008.

Charles Murray predicted the average IQ of modern college students -- within 1 point -- by reasoning from the % of the population that now goes to college.

By Jason Malloy (not verified) on 25 Sep 2009 #permalink

I was just at my 50th HS reunion. A number of the people there did quite well with a HS education or less. I talked to one of them with less, and she was happy with her life but she made sure her kids got as much school as they could. Times have really changed. (Jobs I trained on the job for in 1975 now require three months or more of full time schooling.)

there are two major points of view about this: the negative one is that credentialization is being used to make entry into jobs more difficult while fattening up the schools; a second point of view is that junior colleges and colleges are being forced to compensate for the weaknesses of the HSs; and the third is that the skill set needed for even ordinary jobs is greater and more school-like than it was in 1910. For example, bottom-end jobs these days include low-level hospital work (aide, custodial), and that requires some specific training where pick and shovel work didn't. (Yes, custodial. Untrained custodians are a big infection-control risk.)

By John Emerson (not verified) on 25 Sep 2009 #permalink

Actually Murray extrapolated that college applicants should now have an IQ of 104, while Inductivist found that college graduates now have an IQ of 105. So he may have been spot on.

By Jason Malloy (not verified) on 25 Sep 2009 #permalink

I'm not too surprised. I went to grad school at a large state university. Part of my job was grading all 300 of the Religion 101 essay tests. I would say about 10% of the tests were barely literate. What was supposed to be two paragraphs would be two fragment sentences (amazingly most of these weren't athletes). Only about 20-30% of the tests would be what I considered C or above (basically literate with general ideas being sound), with 3-5% being A level papers (Well argued and factually correct ideas, few if any grammatical and spelling errors). Mind you, these weren't hard questions; stuff like "Describe the differences and similarities between the three main branches of Christianity" In any case the profs then went and graded on a curve. When I finally got a chance to teach my own curriculum, I decidedly chose not to grade on a curve. I tend to believe it's a major part of why others got picked for the openings at the Community College level and I didn't.

"Your scatter plot doesn't doesn't imply causation (R^2 = 0.38?)."

It kinda does though. I mean it's not as if causation could be operating in reverse (college students getting dumber means there are more college students?). And there could be a third factor at work, but if so, what? Overall population IQ is increasing (Flynn effect) or at worst not changing (if you don't believe in the Flynn effect).

As for who cares about WORDSUM - people with a high WORDSUM, dude.

Is there an online WORDSUM test available?

With the appropriate google search terms you can find it. To protect the integrity of the test, I won't link to it here.

To make the conclusion you have assigned as a title you would have to do several things. First, you would have to show that this test is indeed a reliable proxy for IQ, one way being to correlate the Stanford-Binet and WORDSUM scores of a sample of people. You would need to analyze the mean and standard deviation of the WORDSUM test(assuming it's standardized).Perhaps one can get wildly different scores based on the question difficulty. We don't know the nature of questions.Secondly, it appears from the graph that the 95% confidence intervals overlap a great deal. That is, the true mean is captured by several points that you may have decided as different. Further, you would have to do statistical analysis to see if the results are indeed different. You'cant just correlate like crazy and draw pretty graphs and expect us to swallow your conclusions whole.