As a methods prof, a lot of the social science done or cited - especially by journalists - to "prove" contemporary Wokish talking points seems almost intentionally badly done.
(2) An example would be the claim single parents do as well as married couples at raising kids, IF YOU ADJUST FOR INCOME. Maybe. But, 2 people make more $$$ than 1, and men out-earn women. You'd have to make ~220% what the mean single mom does to be as stable as the mean couple.
(3) On a different note, 3+ complex "Bayesian" analyses of police shootings look at B:W risks of getting shot by cops, and find ratios of 2-3:1. However, none adjust for the B:W difference in victim-reported crime rate, which is 2.4:1, and totally explains the gap.
(4) More broadly, any model which doesn't include multiple independent and control variables is ~worthless. Black people get hit more by cars? Does this hold after you adjust for driver/pedestrian behavior, age, population density, etc? If not, there likely is no scandal at all.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
