As a general position statement here, the basic issue is that the Conscious Kid captured video contained statements like "children as young as two are already using race to reason about people's behaviors". Johnson and Project Veritas claim that this is ridiculous, and that it is wokist and racist lies.
I haven't read the research to say whether or not it is true. I knew about the Clark doll test which was on children from 3 to 6, so it seemed at least possible to me that children at age two could use race to reason about behaviors. I haven't read material on earlier ages like the Sapolsky book Pat is referencing, but that apparently agrees with the training on this point.
I'm not endorsing either the training or Johnson in general. But I think on this point, the training seems right and Johnson is wrong. Johnson and Project Veritas seem to be pushing that young kids naturally don't see race, so parents don't need to do anything and their kids will naturally grow up with no racism. I think that isn't true. I think a stronger point could be made arguing about *how* to teach children about race, rather than *whether* to teach children about race.
From Pat's description of Sapolsky, he is similarly skeptical of IATs. However, his tests use neurological signals in addition to reaction time, which gives greater detail on the mental activity going on.
Not "his" tests. Sapolsky's book is a survey of contemporary understanding, not original research.
Sorry, Pat. You had said that, and I knew it at one point but slipped up.
So they did investigate whether it was just contrast.
I'm well aware there are certain people who actually do study these types of things with a more critical eye. My criticism isn't targeted at Sapolsky but toward the broader field of academics and charlatans using IAT and not bothering trying to challenge their own assumptions. Heck, most of the time they aren't even qualified to parse their own data due to general statistics innumeracy among the "social science" crowd.
I wasn't able to find a funnel plot specifically for IAT publication bias, but I suspect that a plot for IAT power would look similar to this plot done for another commonly-abused topic, Stereotype Threat.
(Ref: https://russellwarne.com/2021/02/17/the-gremlins-that-strengthen-stereotype-threat/)
I generally agree here. I've had a fairly in-depth look into this, since I first got a PhD in physics - and later got a Master's in Education. Statistics handling in education was awful compared to physics, which isn't at all surprising. The sort of people who are talented in education tend to have worse math skills than physicists. Two things I would add:
(1) From my reading, statistical handling in education used to be even worse. There was a push in the 1990s to firm up data handling with pre and post tests. In earlier times, education was even more driven by unproven trends, like the often-mocked "New Math" of the 1960s.
(2) Publication bias happens in nearly all scientific fields to at least some degree. It's not a function of innumeracy per se, but comes from a higher-level pressure against publishing negative results because they don't seem significant. I saw it regularly in nuclear and heavy ion results, for example, and less so in high-energy physics (though it still happened). I had a roommate whose physics PhD I considered nearly complete crap because of statistical handling.
The point being - I take psychology and social sciences especially with a big grain of salt -- but I am even more skeptical about random people making assertions about those fields.