Sunday, June 7, 2009

Gender of College Students vs Time

What an interesting coincidence.

On Thursday, Dean Dad blogged about a claim concerning the "rapid increase" in college grads producing a drop in wages, leading to a discussion where Dictyranger posted a link to census data on educational attainment including tables on education levels and income broken down by many relevant categories. I'll be looking specifically at Table A-2.

On Sunday, the BBC News reported on a study in the UK documenting that women are ahead of men in almost every measure of academic achievement. (Aspects of this are also relevant to a separate discussion of The Gender Knot at Zuska's place because the one area where women remain far behind is in math, physics, and engineering.) One intriguing comment from that article is quoted at the bottom of my post, but my emphasis here is on numbers relevant to the discussion at Dean Dad's.

The following graph, showing data that can be interpreted as indicating a running average of the college graduation rate of a 5-year cohort of young adults, has been constructed from Table A-2 of the census data:

Totally sexist colors have been used to code the gender data.

The fact that the data go from 1940 to 2008, including the GI Bill students as well as the children and grandchildren of the 'Greatest Generation', makes these data really interesting for a variety of reasons. Note that data for a given year (such as the jump in 1995) reflect the graduation rate of "normal age" students about five years earlier (circa 1990).

Here I will start by pointing out that there is no evidence of a "rapid increase" in college graduation rates other than a 5% step jump circa 1995. If they were smoothed a bit more, as one sees in the graph below, one could argue that the college graduation rate for MEN has not changed at all in over thirty years. I spent too much time figuring out a weird feature of the file to construct a graph of the derivative, but you can see that the most rapid increase was from about 1965 to 1976 (for both men and women, but men started from a higher point) and that the recent increase (for the Echo generation) is due to a continuing steady increase of female graduates. Men are, indeed, lagging significantly behind.

If there is a correlation such as the author claims, it should show up in salary data for the GI Bill group (where the growth of graduates was spectacular, doubling in just a few years) and for the Baby Boom cohort to a greater extent than at present. After all, the rate at which men are graduating college today is not high enough to replace the men who are retiring. I wouldn't be surprised to see such a correlation, by the way. There was a major recession in 1975, one that heavily impacted the glut of Baby Boomers entering the work force out of college. That drop for men after 1975 no doubt reflects the oversupply of college grads along with the end of the Vietnam draft (and the need for a college deferment).

The grad rates for women caught up with men around 1985 (that is, for students graduating around 1980) and the rates for men and women tracked together from 1985 to just after the 1995 jump (that is, for students graduating circa 1980 to 1990). At that point, the rate for men settled in at around 26% while that for women continued to rise above 33% with a huge jump in the last few years.

So ... are the data talked about by David Frum due to a glut of college grads, or are they a result of women being paid less than men for entry-level positions? Since Frum did not comment on the fact that the rapid increase is mostly among women, perhaps it is relevant that he is a "conservative" commentator. (That also shows up in his failure to include CCs in his discussion of cost, and his comments about HS exit exams. He seems unaware that the reason many jobs require Some College is precisely because CCs provide that benchmark with our "prep" classes.)

There are wage data in those census tables, but they need to be corrected for inflation to have any relevance here. But, combined with the tables available at Oregon State for the CPI going back into pre-history (literally, in some cases), the data do exist to look at some of those questions as well.

A broader view is shown in this graph, which shows the fraction of the entire working-age (and retired-age) population, again from Table A-2:

I should have put some straight lines on there to show the average slope, but I trust you can see that the fraction of the entire population with a college degree has grown pretty steadily over the years. That for women has increased at pretty much the same rate since 1970, while that for men (and hence the total) is growing more slowly in the last 20 years than it did in the 60s.

What is striking about this graph is that it shows that we are, RIGHT NOW, at the point where the fraction of men and women with a college degree is about the same. Jobs that require at least four years of college should (other things, like child care and area of specialization, being equal) have similar numbers of men and women in them. To someone who started work when that was definitely not the case, this helps explain why I have the distinct impression that a 'phase transition' has taken place. One has taken place. And when you include the disproportionate number of men in some (but not all) technical fields, that also explains why other areas are now majority female.

Technical Note

It took me several tries before I figured out why Excel was refusing to make sense out of the "year" column in the tables. The rocket scientist who created them entered the year information as ..1980 with the dots made white so they were invisible in the table. Trimming them seemed to make the years disappear completely because then the year inherited that color and became white - but the years were still there and could be made visible by changing the font color to black. At that point I could tell it to treat the year as a "date" and get the kind of graph I wanted.


This is raw Uncle Al bait if there ever was any. From the BBC article:

A science test taken by 11 and 12-year-olds in the mid-1970s had been successfully passed by 54% of boys and 27% of girls.

When the same test was taken in 2003, the scores for both boys and girls had fallen to 17% - a much more rapid decline for boys.

What an absurd way to achieve equity, by making the entire system worse rather than parts of it better! (Yes, I know that the UK is not the only place that might have taken this approach.)

1 comment:

Sherman Dorn said...

Good analysis, great questions of the data. The one caveat, esp. for the past 15 years, is the fact that a part of the advantage for women in earning bachelors is the disproportionate college attendance over the age of 22. That means that part of the 25-29 BA attainment (and part of the gender gap) is coming after "traditional" college ages, conflating some of the issues. Not entirely, because you're looking at a sensible slice of each cohort experience, and the stagnation in male BA attainment is real and much more noticeable in the first graph than I've seen elsewhere.

There are ways to look at college attainment at smaller and more varied age intervals, but the overall picture is not that different -- a gender gap grew in the postwar era, shrank, and then reversed.

Moreover, your two hypotheses about the stagnation in male BA attainment are good. One important one to add is the peak of the baby boom, which hit high school graduation in 1973-74. Part of the wage stagnation since '73 is a cohort effect. NOT ALL, by any means, but some of it.

So one more question is, to the extent that a cohort effect exists, was it gendered?

And yet another question/alt. hypothesis: the 1970s was in the middle of the 15-year rise in proportional divorce rates (divorces per thousand married women), as well as increased (and probably more visible/everyday) feminism and the enactment of Title IX and associated regulations. To what extent did the increased visibility of divorce (from real changes in population behavior), feminism, and Title IX keep pushing the college-going and -completion rates for women at a time when the completion rates were stagnating for men?

And probably others... but this is a thought-provoking entry. Thanks!!