This is really funny because Kaufmann claims these are "significant" correlations between all these variables but if these are Pearson's r (standard correlation coefficient) then these are moderate-to-weak correlations. Did he think no one with any statistics knowledge would read his "analysis"?

Oh, that's even more egregious than I originally thought. Kaufmann seems to be engaging in a textbook instance of cherry-picking to fit a preconceived narrative.

Suggestion: polling on ballot measures (e.g. CA Prop 50) and some local/downballot races (i.e. mayoral, DA/state AG, etc)

Also worth noting that all the surveys depicted focused on college students at elite universities; the graph also excludes two of the surveys included in the review but not included in this graph (HERI and CCES), which surveyed a broader pool of respondents and didn't find a significant change