Statistics

Racism Eats Telomeres. An Exciting New Paper-Generating Research Area

What do you call racism which isn't?

What do you call racism which isn’t?

A crack team of researchers led by David Chae has “discovered” that racism, the severest of all crimes, eats the telomeres of black men. Or maybe “eat” is the wrong word. How about “dissolves”? And it wasn’t really racism, as we’ll see, but racism sounds scarier than anything else.

What’s a telomere? Think of it as a cap on the ends of your chromosomes which keeps them from unraveling. Telomeres have a length which is considered a rough measure of age, with shorter older. Which is to say, on average older people have been found to have shorter telomeres.

The exciting new peer-reviewed paper is “Discrimination, Racial Bias, and Telomere Length in African-American Men” in the American Journal of Preventive Medicine.

The press translates this as “Racism May Accelerate Aging In African-American Men“, where by “may” they mean “does”.

Accelerated aging and a greater likelihood of suffering from an age-related illness at a younger age are two consequences being linked to African-American men who have experienced high-levels of racism throughout their lives.

Ah, “linked to” that wonderfully non-committal but dread-filled phrase loved by reporters and researchers the world over. More sins are covered by “linked to” than any other. But never mind.

The idea here is that racism invades the body like a virus (I’m extrapolating from their ideas) and nibbles away at telomeres, which in turn cause people to become older than they are, chronologically speaking, and to suffer earlier from age-related illnesses. How side-long glances of suspicion and unfriendly n-words seep into the blood stream is a mystery the authors leave mysterious. But racism is “linked to” telomere shortening.

Except it isn’t exactly racism. Our crew used the “Implicit Association Test“, which is supposed to measure racism, that most vile of all isms. But it isn’t. Instead, it’s a score of how fast you hit the computer keys when presented with images like the one above.

Now whatever this score it, people do not attain the same one on repetitions of the test. There is thus uncertainty built into the IAT score which is never (that I have seen) carried forward in any analysis. That means the wee p-values reported in studies which use the IAT are always too low. They are also too low considering that IAT is not measuring racism. What is said to be “significant” isn’t.

How far from actual racism is the IAT score? Nobody in the world knows. What makes it science is that this problem is ignored and the IAT is defined as a “racism measure.” The IAT is in such widespread use because academics have an impossible time finding real racism, especially on college campuses where most studies take place, so they have to make some up, no matter how subtle. Else what would they study?

Anyway, Chae and gang gathered 90 black men, gave them the IAT, asked some questions about perceived racism, peered in their genes, and then set free with a “$70 gift card” (to what store we never learn). Make sure you get this: Chae did ask questions about real racism.

Two guys’ values were tossed because their telomere values did not fit the authors’ preconceptions. This is a common strategy used by researchers to ensure conclusions match expectations.

Chae and gang did not do anything as simple as showing the IAT scores against telomere length, which would have been revealing but not especially publication worthy. Instead, the stuffed all the data they had into a regression model as a function of telomere length misreading it as nearly everybody does.

The real “Racial discrimination” questions did not give wee p-values in any of the models they reported, which means “non-significant.”1 Meaning actual racism was not “linked to” telomere length. The IAT score, which they relabeled as “Implicit racial bias”, did give a wee p-value, which might be why its label changed. The effect was small and the model only weakly predictive.

Which is why the authors brought up “‘stop-and-frisk’ policies”, “profiling”, and something about “mortgage markets” in their conclusion. None of this stuff had any bearing on what the authors did, of course, but still: we can never discuss racism enough.

The authors never no recognition their results could be spurious: e.g. IAT score would have been confounded with chronological age. Older men probably scored higher on the IAT, which measures speed, plus older men had more opportunity to experience real racism. And older men would, on average, have shorter telomeres. Nor did they say how the IAT scores could have shortened telomeres. But they did admit “holding a pro-black bias may serve as a buffer against racial stressors.”

———————————————————

Thanks to Al Perrella who not only found the paper, but dug it up from the deep recesses of the web.

1Technical note: the authors included a real racism-IAT score”interaction”, which gave wee p-values; but this sort of trick (interactions with insignificant main effects) is frowned upon by your better sort of statistician.

Categories: Statistics

2 replies »

Leave a Reply

Your email address will not be published. Required fields are marked *