Lee Jussim is absolutely infuriating to read. Most of this argument hangs on what as far as I can tell is a semantic twist. He argues that if DKE exists, it s

Lee Jussim is absolutely infuriating to read. Most of this argument hangs on what as far as I can tell is a semantic twist. He argues that if DKE exists, it should evidence itself in a non-linearity of "self-assessed IQ" vs. "tested IQ". But... I don't see how or why the interpretation demands that. He explains, sort of, it in terms of higher IQ self-estimations being necessarily more accurate, which is itself an untested hypothesis! The DKE, as typically expressed, isn't about linearity of the fit curve, it's that the slope doesn't go through the origin. The plot he shows right in the article makes it clear that median 60-IQ responders self-assess at 100, and 140-IQ responders get it exactly right. That's the DKE, as clear as day. Yet his text immediately afterwards is "Bottom line: No evidence of DKE in a very large sample test, when tested most rigorously." But what he's disproven is some linearity correlary that wasn't in the original paper and isn't cited, nor defined by him. Basically this is a fancy strawman, with pretty data. That's not to say there isn't room for reasoned argument about what the mechanism behind the DKE effect is or in what direction the causality goes. But this isn't it.