K-means clustering: US polarization overwhelmingly due to Left moving leftward

I was far right for 20yrs but now identify as a recovering conservative. It was weird to shift left while not changing.

When it comes to politicians it is the opposite

https://www.pewresearch.org/short-reads/2022/03/10/the-polar...

And I believe that result because I’ve studied DW-NOMINATE in depth. I’ve run into problems trying to do dimensional reduction on preference data and the literature on DW-NOMINATE is distinct and geometrical being written by political scientists on the quantitative side as opposed to data science or ML.

I can very much believe that left-leaning civilians have moved left at the same time right-leaning politicians have moved right, I mean politicians don’t care what ordinary people think very much anyway, see

https://www.cambridge.org/core/journals/perspectives-on-poli...

Note wherever they try DW-NOMINATE there is a clear left-right axis. Once in a while when a country is in a transitional state like 1960s America where the Republicans and Democrats were changing places in the south you see a small secondary axis. You might want to believe in that Libertarian political test with two axes but it is a lot of hooey and it does not point to a position any politician could inhabit.