A letter in the Washington Post!

TUESDAY, MAY 7, 2019

The role of basic facts:
In his column this Sunday, Nicholas Kristof reminded us of the type of issue we simply never hear discussed on our favorite "cable news" channels.

Those channels exist to sell viewers The Chase. Other topics can just go hang in the yard.

On Saturday morning, the Washington Post ran an intriguing letter about our public school "achievement gaps." The letter helps illustrate how poorly we citizens and news consumers may understand basic facts concerning major topics.

The letter came from a DC-area professor. It went exactly like this:
LETTER TO THE WASHINGTON POST (5/4/19): Achievement gaps between groups in the United States remain a major educational problem. Unfortunately, the April 20 Metro article “Report shows disparity in success,” about the achievement gaps in Montgomery County Public Schools, obfuscated this important issue by focusing on ethnic group differences in achievement and not simultaneously considering the role of socioeconomic status. Such research shows that (on average) students from poorer backgrounds do less well in school than wealthier students.  A higher percentage of Latino and African American children are poor compared with their white counterparts. The article focused on Latino, African American and “disadvantaged” students as discrete groups, rather than presenting the complex interplay of ethnicity and economics.

Comparisons across ethnic groups should be made only when socioeconomic status also is considered or controlled statistically. Not doing so results in erroneous conclusions about the causes of performance differences, which can reinforce stereotypes that certain ethnic groups just don’t do well in school. Such stereotypes are harmful to members of the stereotyped group themselves, as psychologist Claude Steele’s research on “stereotype threat” clearly indicates.
The letter suggests that prevailing gaps between different "racial" and ethnic groups are mainly an artifact of family income. Within prevailing pseudo-liberal understanding, that counts as happy talk.

That said, to what extent are achievement gaps an artifact of family income? Here are some chastening basic data from the most recent administration of the National Assessment of Educational Progress:
Average scores, Grade 8 math
Nationwide public schools, 2017 Naep

Higher-income white students: 299.74
Lower-income white students: 275.28

Higher-income black students: 272.71
Lower-income black students: 255.02

Higher-income Hispanic students: 280.22
Lower-income Hispanic students: 264.22

Higher-income Asian-American students: 320.52
Lower-income Asian-American students: 289.26
For all Naep data, start here.

Perhaps some better data exist somewhere. But as you can see, lower-income white students slightly outperformed higher-income black students on this Naep math test. Based on a conventional though very rough rule of thumb, lower-income Asian-American students outperformed higher-income black kids by well more than one academic year.

In Naep data, as in most standard educational data, the income levels to which we refer are based on eligibility for the federal free and reduced price lunch program. The "lower-income" kids to whom we refer are those who are eligible for this program. "Higher-income" kids are not.

That letter in the Washington Post seemed to feed a favorite liberal narrative. Basic data from the most recent Naep seem to say something more gloomy.

On the more uplifting side, the Washington Post and the New York Times will never report on a gloomy topic like this. It simply isn't done! Meanwhile, you'll see public schools discussed by Rachel after she's banked her first billion.

On cable, it's all about the thrill of The Chase. No other topic need apply. The thrill of The Chase really sells, and low-income children do not, unless they're packaged in a brain-dead version of the familiar mainstream classic, The Bad News Bears Knock It Right Out of the Park.

Low-income children don't sell. So too with the topic of Kristof's column. Michael Cohen draped in chains is much more delicious, more fun.

We're just showing you how it works. You can decide how you feel.

21 comments:

  1. Dividing kids into income groups by race is not what is meant by statistically controlling for income level. Somerby cannot address this question using the method he has chosen and he doesn't realize it because he is ignorant about statistics.

    There may be a far larger proportion of black kids in poverty than white kids. Despite using the same cut-offs to determine poverty, there may be far more black kids in deep poverty compared to white kids. The impact of severe poverty may be considerably different than that of mild poverty, and being urban versus rural matters. So Somerby has not tested his theory and has no business telling liberals off.

    ReplyDelete
    Replies
    1. The high-income black students (above the cutoff) do not score as well as the low-income white students (below the cutoff) so this demonstrates that income level is not the sole cause of the overall difference, which is Somerby's main point in this post (he has not revealed what his theory is). Factoring in income level would not remove the difference. But other cultural differences must obviously influence the test results. For example number of books in the home has been shown to have some effect, but this is not something that is measured for all students who take the NAEP test. There is little incentive for kids to study if the main economic opportunities in the neighborhood are seen to be drug dealing and athletics. This is not something that can be easily measured numerically and factored in to test results. Only after all environmental factors have been identified and measured - an impossible task - could the difference in these tests be ascribed to genetics.

      Delete
    2. Your first paragraph is well stated indeed, particularly when you tangentially refer to his “gloomy” characteristics of the NAEP scores. Lower-achieving Asians seem to perform better than any higher level group except the white one. Perhaps that’s the gloomy aspect of this, to which you refer: A rich family environment can make a whole lot of difference, which I think should be obvious, tiger moms aside.

      Leroy

      Delete
    3. You can't figure this stuff out using means alone. You need to know something about variability of these groups. The Asian mean could be much higher not because the kids are smarter but because there are fewer poor kids among them. The poor Asian kids could have a higher mean because the depth of their poverty or the number of kids living in the worst conditions is smaller than in other groups. This has nothing to do with the intelligence or academic proficiency of most of the kids but is determined by how poor the poorest kids are (extreme values) and the proportion of very poor kids compared to those closer to the cutoff. Unless you know this about the distribution, you cannot draw these conclusions about Asian kids compared to white or black kids (or black kids compared to white kids either).

      You are wasting energy thinking about it without having actual facts to consider. Means without standard deviations are useless.

      Delete
    4. Those high scoring Asians certainly seem to cheat more post-high school. Society is not served by pushing kids to over-achieve. Wealth-attainment is an empty goal.

      Delete
  2. "The letter suggests that prevailing gaps between different "racial" and ethnic groups are mainly an artifact of family income."

    That is not true, dear Bob. The letter talks about "socioeconomic status" and "poorer backgrounds".

    It's you, Bob, who're performing a zombie reductionist trick by equating "socioeconomic status" with "family income".

    The family of an impoverished university professor, and the family of a ghetto dweller in line of several generations of ghetto dwellers, may have the same family income, but very different socioeconomic status.

    ReplyDelete
  3. Do a test of NAEP means for white and black kids with income as a covariate, not as a second IV (race x income) as Somerby presents here. That controls for income. There may be other factors that similarly need to be controlled to get a fairer picture of the impact of race alone (Somerby's preferred explanation for the gap in scores).

    ReplyDelete
    Replies
    1. Somerby doesn't talk about the interaction either. There would presumably be one.

      Delete
    2. When you have multiple factors that you think might be influencing academic performance, it might be better to do a regression analysis. That way you can not only find out which factors influence NAEP scores (or control for the ones you wish to exclude), but you can also measure how much or how strongly each hypothesized factor affects those NAEP scores.

      Those who study this stuff would never just group kids and compare the means the way Somerby does. This is more of a form of propaganda, in my opinion, than any attempt to get at the truth of what affects academic performance.

      Delete
  4. The figures show that test scores do correlate with income to a considerable degree. But, that doesn't mean that low income causes low scores. It doesn't mean that giving people more income will raise the test scores.

    To answer these questions, one needs to know why lower income children are generally worse students.

    ReplyDelete
    Replies
    1. ******
      David in Cal March 23, 2019 at 10:07 PM

      This is the new Democratic talking point, but it won't last long. First of all, just about nobody opposes releasing the Mueller Report - not even Trump.
      *****

      White House invokes executive privilege to bar former counsel from turning over documents to Congress


      Hey David, you fucking treasonous bastard genius, tell us, who pays for Donald J Chickenshit's insane tariff war with the world?


      Stocks fell sharply Tuesday after a top U.S. trade official indicated that higher tariffs on Chinese goods are coming later this week, disappointing traders who hoped President Donald Trump’s weekend tweet threat was just a negotiation tactic.

      Delete
  5. mm -- despite a big drop in the stock market today, stocks are up 45% since Trump was elected. That's a great result.

    ReplyDelete
    Replies
    1. That's a great result for corporations and rich people.

      Delete
    2. ******
      David in Cal March 23, 2019 at 10:07 PM

      This is the new Democratic talking point, but it won't last long. First of all, just about nobody opposes releasing the Mueller Report - not even Trump.
      *****

      WASHINGTON — President Trump asserted executive privilege on Wednesday in an effort to shield hidden portions of Robert S. Mueller III’s unredacted report and the evidence he collected from Congress.

      Hey, fuckface, remember how you were telling me how wonderful Donald J Chickenshit was for not asserting executive privilege? Now you will be saying the exact opposite, and you won't even be embarrassed. DinC, Fucking treasonous bastard.

      Delete
    3. Turns out two things about that Mueller Report.
      1) It doesn't "totally exonerate" Trump at all.
      2) Collusion, collusion, collusion.

      Delete
  6. “In Naep data, as in most standard educational data, the income levels to which we refer are based on eligibility for the federal free and reduced price lunch program. The "lower-income" kids to whom we refer are those who are eligible for this program. "Higher-income" kids are not.”

    This is quite bogus. “Eligible for federal free lunch” and “not eligible for free lunch” tell us only very generally about a family’s income. Eligible is generally below a certain dollar value, but “not eligible” is literally any income higher than the cutoff. It is too broad. In other words, you have not controlled for income by simply calling your data points “lower income” and “higher income.”

    ReplyDelete
  7. Somerby is such a tease. Obviously, he knows the real reason for the achievement gaps. Of course he does. He just chooses not to tell us what it is in his opinion. It isn’t test prep, it’s not segregation, it’s not income. He has, via convincing and exhaustive proof, shown us this. And he shows it by his own ad hoc number-pulling from the only source that matters, the online NAEP data explorer, and without referencing a single academic study to back up his claims. Somerby isn’t even a statistician. But who needs to be when dealing with those pesky statistics after all?

    One of these days, he will finally reveal the answer. I just know he will. Then he will tease us for twenty more years as we try to figure out the solution that he alone knows.

    ReplyDelete
  8. “the April 20 Metro article “Report shows disparity in success,” about the achievement gaps in Montgomery County Public Schools”

    Where was the discussion of this article in TDH? It’s about achievement gaps, something that TDH has told us never gets discussed at our elite media outlets. For that matter, where is the link to the article so that we can put the professor’s letter in proper context?

    “The letter suggests that prevailing gaps between different "racial" and ethnic groups are mainly an artifact of family income.”

    And just five minutes ago, in Somerby’s previous post today, Somerby was complaining about inaccurate paraphrase..!

    ReplyDelete
    Replies
    1. For that matter, where is the link to the article so that we can put the professor's letter in proper context?

      Had you clicked on Somerby's link to the letter, itself, you would have found in its second sentence a link to the article you were so going to read to "put the professor's letter in proper context." But, of course, you're not here to discuss the subjects Somerby raises, you're here to lodge complaints about Somerby.

      Delete
  9. This comment has been removed by the author.

    ReplyDelete