Socioeconomic status, ideology & party

Andrew Gelman has a post up, Who are the liberal Democrats and the conservative Republicans?, which shows that conservative Republicans tend toward higher incomes, while conservative Democrats tend toward lower incomes. I decided to see if something similar was discernible in the General Social Survey. I used the PARTYID, POLVIEWS and SEI variables to explore the question, and limited the sample to whites and the years 1998-2008 (so as to have contemporary relevance and control for ethnic confounds). I clustered all Republicans & lean Republicans into one category, and did the same for Democrats. I also clustered all who were liberal and conservative into one category (extremely to slightly). Finally, socioeconomic status ranges from 17 to 98, and I broke it into three categories of "low," "middle" and "high," 17-47, 48-77 and 78-98. The percentages in the population for these three categories were 54%, 36% and 11%, so "low" is really lower to lower middle class. In any case, the charts below....

i-c1cee4cfa80e1fe49f306861685a351d-whitedem.png

i-c492a18fe979b6b37dc8b087622bea86-whiteind.png

i-0170131314143329fd75345481030759-whiterepu.png

White Democrats



Liberal Moderate Conservative
Low 38.4 57.9 62.8
Middle 43.6 33.1 31.8
High 18.1 9.1 5.4




White Independents



Liberal Moderate Conservative
Low 62.6 62 66.4
Middle 32 30.3 26.8
High 5.4 7.7 6.8




White Republicans



Liberal Moderate Conservative
Low 50.1 50.9 40.2
Middle 38.8 38.1 44.8
High 11.1 11 15

The same general trend as in the data Gelman reports; conservative Democrats tend to be lower socioeconomic status, while conservative Republicans are higher status. Since the intelligent and educated are more ideological this makes sense. Over the past generation more Americans routinely have labelled themselves Democrats than Republicans and conservatives than liberal, but both Democrats and conservatives have been confused as to why they can't always translate their numbers in polls into permanent dominance in politics. I suspect there's some sort sentimental preference in people labeling themselves Democrats and conservatives which doesn't really reflect much in terms of their voting or the realized preferences contingent upon their voting.

More like this

In the post below I combined some of the Census Regions for reasons of sample size. But I decided to do this again without combining, but removing some of the questions because of small sample sizes. Again, I also limited the sample to whites between 1998-2008. But, I added another category:…
Question below about the details of what conservative Democrats or liberal Republicans might believe, etc. I decided to look for a few questions. I removed Independents because their sample sizes are a bit smaller. I clustered all those with socioeconomic status 17-47 as "Low" and those from 47-98…
A question below: I'm curious about the demographics of this category, specifically their geographic distribution, religion and ethnicity. First, I limited the sample to whites to remove confounds of ethnicity. Interestingly, in the GSS in the period between 1998-2008 24% of black Democrats/lean…
John Hawks points me to a "He said, she said," piece which wonders whether there is an inverse relationship between belief in the paranormal and religion. The basic thesis is that the mind abhors a vacuum so without institutionally guided supernatural beliefs people simply revert to "default"…

Tantalizing. Does a good job of showing the internal structure of each party, but it's also frustrating. For example, a conservative Democrat is not much like a conservative Republican and might possibly be like a moderate or liberal Republican. And are the L-M-C groups within each party defined as equal in size -- i.e., are the lines drawn to define terciles (if such a word exists)? And knowing the sizes of the three party groups relative to one another would help too.

By John Emerson (not verified) on 20 Aug 2009 #permalink