@jwhandley17 I've been working on a building up a comprehensive set of @BESResearch demographic variables, so I can run dredge-regressions without "attitudinal/behavioural/political" variables displacing everything.

/1

On this chart <--Left is "supports govt redist to make incomes more equal", -->Right is "opposes ...".

(Would like to categorise all of the variables ... but it's hard to do automatically or manually - might be possible to do some of it via the answer sets)

/2

Re: Building a coalition to support more redistribution rests on

(1) people who support it because they think it means more money for them
(2) people who support it even if it means more taxes from them because they think it's ultimately a good thing for abstract reasons

/3

You can't afford to neglect either side of that - and neither side is super stable.

(1) group is easily bribed/negotiated with separately and prone to focus desire for "redistribution from" to other groups ("soak the rich" will always be in competition with "soak foreigners")
/4

(2) group can likewise be bribed by parties offering socially liberal alternatives (redistribution is attractive ... but so is a party offering less humiliation for foreigners)

/5

The issues with HE/education variables does underscore an issue - your necessary coalition is "people with no qualifications" and "people with high qualifications" - meanwhile, opponents have a coalition of "all the people in the middle".

/6