It almost always gives one left-right dimension except in really unusual cases of a political system in the throes of change, such as the US in the late 1960s when segregationism was coming apart in the South. I've thought that it oughta be possible to do the same for ordinary citizens who sometimes answer a panel of 50 or so questions on a public opinion poll. I've got the data to make it work but the dimensional reduction methods that I've used for continuous variables don't work well for categorical variables where you need a different loss function.
thanatos519 · 9h ago
I gave different answers to these questions but you predicted the same answer ... does that mean most people give the same answer for these? :)
> In general, I believe government spending should be decreased to reduce the burden on taxpayers.
> In general, I believe government spending should be increased to fund more public services.
incomingpain · 9h ago
It asks me my country(Not usa), then asks loads of US politics questions.
This factor analysis for legislator behavior is pretty interesting
https://en.wikipedia.org/wiki/NOMINATE_(scaling_method)
It almost always gives one left-right dimension except in really unusual cases of a political system in the throes of change, such as the US in the late 1960s when segregationism was coming apart in the South. I've thought that it oughta be possible to do the same for ordinary citizens who sometimes answer a panel of 50 or so questions on a public opinion poll. I've got the data to make it work but the dimensional reduction methods that I've used for continuous variables don't work well for categorical variables where you need a different loss function.
> In general, I believe government spending should be decreased to reduce the burden on taxpayers.
> In general, I believe government spending should be increased to fund more public services.
70% accuracy, 21 out of 30.