Very interesting, thank you for sharing that!

The patterns in subpopulations seem to match general expectations. Very cool.

Almost no responses in the $100k+ incomes. I wonder if that's a general problem with the method?

It could be a general problem. For collecting responses from high-income individuals the method faces the double-whammy of: (1) smaller number of potential respondents; (2) respondents are least willing to spend their time on such surveys (or the sites the survey-wall blocks).