Thread by @epkaufm

1/ Authors of a pro-DEI paper published in Nature and reported in Science made critical errors which nullify their findings. Not only did they bury inconvenient findings, they reported junk results.

science.org/content/articl…

2/ First, they found that Black and Hispanic scholars have far fewer cited published works than White and Asian scholars but did not report this.

3/ Women are also less productive than men. What we see is that the race gap is found mainly among men (decline slope of blue line), with little difference between races among female scholars (flat red line). This model controls for discipline and years in academia.

4/ Their other striking finding is that White and Asian candidates appear to be discriminated against in being promoted to full professor. They are 10-20 points less likely than Blacks/Hispanics to get promoted to professor. This gets a brief mention but is largely hidden.

5/ The finding they tout, that among less productive scholars Blacks and Hispanics are discriminated against, is based on flawed data. The authors used a dataset in which 1571 individuals were inflated to 9032 by multiplying by the number of academic letters written for them.

6/ Using only 1571 individuals results in the statistical significance of their headline finding disappearing. Had they done the analysis properly, the paper would not have been published. But researchers and publications lapped it all up without question since it accorded with their priors.

7/ Read the whole story at:

erickaufmann.substack.com/p/junk-science…

This is why we need to start supporting new heterodox social science initiatives like my heterodoxcentre.com @UniOfBuckingham