A bogus statistical framework: When the New York Times covers public schools, you can always expect major errors.
So it was last week, when the famous but bungle-prone paper published a front-page report about the enrollment of black and Hispanic students in the nation's most competitive colleges.
As we noted on Tuesday, blacks and Hispanics are "under-represented" in the student bodies of the 100 colleges the New York Times chose to study. More specifically:
In 2015, 22 percent of the nation's college-age population was Hispanic. But only 13 percent of freshmen at the 100 schools were Hispanic.
For black students, the degree of underrepresentation seems to have been a bit more substantial, though in truth its hard to tell from the data the Times presents. In 2015, 15 percent of the nation's college-age population was black. But only 6 percent of freshmen at the 100 schools were black.
In that sense, blacks and Hispanics were "underrepresented" at those hundred colleges. But the Times made an even gloomier claim in its high-profile, front-page report. It claimed the degree of under-representation was even worse today than it was in 1980! Headlines included, here's how the report began:
Affirmative Action Yields Little Progress on Campus For Blacks and HispanicsThe Times' conclusion does sound very gloomy. That's always the desired state of affairs in reports of this type.
After 35 Years, Racial Gaps Widen at 100 U.S. Colleges, an Analysis Shows
This article is by Jeremy Ashkenas, Haeyoun Park and Adam Pearce
Even after decades of affirmative action, black and Hispanic students are more underrepresented at the nation’s top colleges and universities than they were 35 years ago, according to a New York Times analysis.
The Times' conclusion was wonderfully gloomy—but did its analysis make sense? We'd have to say it didn't. It seems to us that the New York Times, as is its wont, made several major errors, including its use of a bungled statistical framework.
What led the Times to conclude that blacks and Hispanics were more underrepresented in 2015 than they were in 1980? Let's look at the reasoning the Times employed in the case of Hispanics.
We'll use the data the Times employed. We'll assume the data are accurate. (In classic fashion, the paper failed to identify or link to the sources of its various data, and there seem to be substantial problems with some of the data employed.)
In 1980, Hispanics constituted six percent of the nation's college-age population, according to the Times. But at that time, just three percent of students at the 100 schools were Hispanic.
Subtracting three from six, the Times describes a three-point enrollment shortfall in 1980. Then, the paper moves to the data for 2015:
In 2015, Hispanics constituted 22 percent of the nation's college-age population. But at that time, just 13 percent of students at the 100 schools were Hispanic.
Subtracting 13 from 22, the Times describes a nine-point enrollment shortfall in 2015. Since nine points (2015) is larger than three points (1980), the Times concludes that the enrollment gap has grown—that Hispanic kids are even more underrepresented.
That statistical framework doesn't seem to make sense. To illustrate the nature of the problem, we'll imagine an alternate case where the logical flaw is more apparent:
(Imagined) Hispanic youth, 1980:In that imagined situation, Hispanics would have had extremely marginal representation back in 1980. Their share of the college enrollment would have been only one-eighth their share of the college-age population.
Share of college age population: 8 percent
Share of college enrollment: 1 percent
Enrollment gap: 7 points
(Imagined) Hispanic youth, 2015
Share of college age population: 24 percent
Share of college enrollment: 16 percent
Enrollment gap: 8 points
In that imagined situation, Hispanics would have been much more strongly represented by 2015. There a substantial shortfall would have remained, their share of the college enrollment would now be two-thirds their share of the college-age population.
In this imagined situation, Hispanics would still be "underrepresented" in these colleges, as compared to their share of the college-age population. But they would have come a long way from 1980, when they were badly underrepresented as compared to their share of the population.
Still: according to the statistical framework used by the Times, Hispanics would have been more underrepresented in 2015 in this imaginary situation. In 1980, the enrollment gap was only seven points. Now it would stand at eight!
In short, the Times adopted a puzzling statistical framework as it compared the degree of representation in 1980 to that in 2015. How ridiculous is that framework? Consider this more extreme imaginary case:
(Imagined) Hispanic youth, 1980:According to the Times' statistical framework, Hispanic kids would have been more underrepresented in 2015 in that imagined case!
Share of college age population: 5 percent
Share of college enrollment: 0 percent
(Imagined) Hispanic youth, 2015
Share of college age population: 24 percent
Share of college enrollment: 18 percent
Can we talk? The New York Times never fails to bungle in reports of this type. The newspaper's instinct for error is almost unerring all matters of the type.
In this case, the problems extend well beyond this puzzling statistical framework. The fine print at the end of the Times report suggests that the Times is comparing a type of statistical apples in 1980 to a type of statistical oranges in 2015. Meanwhile, here's another uncheckable possible problem:
The Times is using highly imprecise numbers in its report. As noted above, it reports these actual numbers for the nation's Hispanic youth:
Hispanic youth, 1980:It looks like Hispanic representation in the top 100 colleges that year was exactly half its representation in the college-age population. But uh-oh:
Share of college age population: 6 percent
Share of college enrollment: 3 percent
In the real world, that "6 percent" could be anything from 5.51 percent up to 6.49 percent. Meanwhile, that "3 percent" could be anything from 2.51 percent up to 3.49 percent.
That lack of precision makes a difference. Scanning the data as the Times reports them, it looks like Hispanic enrollment in 1980 was exactly half its proportion of the college-age population. But it could have been substantially more or less than half. Consider this actual possibility:
(Possible) Hispanic youth, 1980:If that's what the actual numbers were, Hispanic enrollment was actually 38.7 percent as compared to the Hispanic proportion of the college-age population. As is routinely the case at the Times in education reports of this type, the work here was a mess from top to bottom.
Share of college age population: 6.49 percent
Share of college enrollment: 2.51 percent
Don't worry—other problems seem to lurk in this high-profile front-page report. And of course, the biggest problem is this:
The important question we have to solve as a nation involves the substantial "achievement gaps" which contribute to these enrollment gaps. The Times devoted exactly one paragraph to that matter in this slipshod report. (Their work there is highly selective.)
Black and Hispanic youth are indeed "underrepresented" at these 100 colleges. How can we create a future world in which that won't be the case?
Down through the years, the Times has never shown the slightest sign of caring about that question. More significantly, it has never shown the slightest sign of possessing the intellectual skills that would let it play a role in the search for answers.
The Times produced a gloomy conclusion when it produced this report. In this age of billionaire-funded "education reform," this is always the desired outcome in reports of this type.
Absolutely nothing has worked! That is the mandated mantra.
In the process of reaching its gloomy conclusion, the Times displayed, as it constantly does, a highly peculiar state of affairs. On average, black and Hispanic kids still lag behind their white and Asian-American peers in basic academic skills. But no one lacks basic intellectual skills like the deeply puzzling people who take up space, year after year, at the hapless Times.
One final point:
You'll never see these matters addressed on your favorite "cable news" shows. They're too busy serving us liberals the porridge we very much like.
Dearest darlings, use your heads! Rating would drop like a stone if your favorite cable stars discussed the actual lives of the nation's low-income kids! That's why our darling, Rachel Maddow, has never done it in the past and won't be doing it now.
If a kid gets shot, we pretend to care and pretend to get mad. Otherwise, all is silent.
Where did we get the 3 and the 6: Where did we get the 3 percent and the 6 percent for Hispanics in 1980?
The numbers appear, though only barely, in the Times' hard-copy report. On line, where data presentation should be easier, those basic numbers don't appear. They just didn't bother to post them.
This is the way the New York Times rolls. As we've shown you down through the years, they do this every time.