Statistics in woke service?

You’ve heard the proverbs. “There are three kinds of lies: lies, d____d lies, and statistics.” “Figures don’t lie, but liars figure.” Two days ago, Twitter offered proof positive of both those maxims. It also offered further proof, in case we needed any, that the Ivy League is not what it once was. John Harvard – and separately, Elihu Yale – must be spinning in their graves to see what the institutions that bear their names have now become.

A statistics student passes off woke illogic for sound science

Meet Kareem Carr – whose Twitter profile belies his account title. He calls himself a “data scientist” but is really a student working on his Ph.D. at Harvard.

I tweet about data science: what it is, how to do it better, why it matters for the rest of society.

His profile at Harvard says he has become a consultant in data science at the Institute of Quantitative Social Science. If the thread he dropped yesterday is any indicator, Harvard has very low standards for making someone a consultant.

Fellow user End Wokeness became his target when he published this bar chart.

The title: “Interracial violent crime incidents in 2018.” It defines three groups – white, black and Hispanic. The United States Census defines the term white to include not of Hispanic origin. Hispanic origin trumps anything. The Bureau of Justice Statistics, the source of the chart, no doubt uses the same definition as the Census.

That chart shows two things: that when members of one “race” attack another:

1. Whites are disproportionately the targets of such attacks by either of the other two groups, and
2. Whites do not attack others nearly as frequently as those others attack whites.

The numbers in the bar chart are absolute numbers, not per capita numbers. Whites still are a plurality if not “the majority,” so per capita numbers would show whites being even less offensive – and blacks and Hispanics being more.

Enter Kareem Carr, ready to give those other groups a “scientific” excuse. To do that, he manipulates statistics.

Two threads, showing how to manipulate statistics

Mr. Carr’s anchor tweet has a screencap of End Wokeness’ chart, and Elon Musk’s response to it. End Wokeness’ point was that the media focus on white-on-black crime though it is the least frequent category even in absolute terms. (Elon Musk, for his part, wonders why the media should have such a flawed focus.)

Herewith the thread:

Of course he doesn’t want answers to the questions he asks at the end. In his mind he knows the answer. Blacks attack whites because whites have all the money, and whites attack blacks because whites are racist. (Here we go again: racist ≡ white.) Furthermore he bases his “moral principles” on envy – let him who has more, suffer more. A thief is an irregular wealth-redistribution agent, and a murderer is taking justifiable revenge.

None of these attitudes properly befits a scientist. We see here a shameless manipulation of statistics to serve a political end – and a projection on his targets.

Yesterday, another user answered Carr as he needed answering:

In reaction, Michael Nayna, whom “Wokal Distance” quoted for the concept hermeneutical tribe, left this:

Two other users left this insightful dialog:

(Note: VIGRE stands for Vertical Integration of Research and Education in the Mathematical Sciences. It’s a scholarship program to encourage more Americans to take math.)

One other person commented, as we have, on the apparently low academic standards at Harvard. Why else, he seems to ask, would Harvard even think to reward this person with a Ph.D.?

These two users described a further consequence:

Analysis

Are data missing from the original Justice Statistics graph? Yes. Again, we see no denominators, so the problems of interracial crime and violence might be better in some ways – and worse in others – than that graph represents. Nor do we see any same-race crime numbers that would allow a comparison – the one valuable question Carr asks.Furthermore we see only a single snapshot. Where are the data that would show a trend? Any law-enforcement officer knows that criminals attack those closest to them. Common street crime, which these statistics are supposed to measure, is an opportunistic phenomenon. No one goes out of their way to attack someone of a different race.

Or they didn’t. Are they doing it now? Again, it would help to know where these incidents are happening, to see where the perpetrators are finding their opportunities.

The problem is: no one else seems to be asking these questions. Kareem Carr asks only one of them – but the legacy media are not asking any of them. Instead they’re drumming up resentment – as CNAV noted yesterday.

Your editor lived with statistics – in medical school and as a Fellow in Medical Information Science. The fallacies in Carr’s presentation are truly cringeworthy – and so are the gaping holes in legacy media coverage.

The real problem

Once again, we have lies, d____d lies, and statistics. But it’s too facile to suggest, as one user did, that Mr. Carr exemplifies the Dunning-Kreuger Effect. David Dunning and Justin Kreuger, writing from Cornell University in 1999, described how those who know the least think they know the most – and those with the least intellectual ability simply cannot recognize their own incompetence.

That can explain ignorance, and blithe overconfidence, but it cannot explain malice. Kareem Carr deliberately and maliciously misuses statistics to make the moral case for punishing whites. Whether for the irregular fine system we call theft, or for the equally irregular summary execution we call murder, he offers the same excuse.

That does not mean that all who are not white, even remotely conceive of making the same excuse. Nor can we forget that many whites also offer the excuses Carr offers. Race isn’t the problem; ideology is. Or, it’s one problem. The other problem are institutions of higher learning who lower their standards and allow such thinking to pass muster.

Happily, Twitter has gotten better than it was. Time was when Twitter would not have allowed anyone to dispute Carr. That time has passed, and for that CNAV is grateful.