I have a new paper (together with Alan Jacobs and Scott Matthews) that looks at the interaction between income inequality and vote choice in comparative perspective. This was presented at the APSA conference (Chicago, Aug/Sep 2013) at rather a good panel.
This is ongoing research, but the conference draft is available via SSRN. Comments are extremely welcome.
Here’s the abstract:
A growing literature has inquired into the political consequences of rising income inequality in the United States. Scholars have identified a number of mechanisms through which American democracy has become more responsive to the interests of the very rich than to the those of lower- and middle-class citizens. Among the patterns of unequal influence that analysts have observed is a strong “class bias in economic voting” identified by Bartels (2008). Specifically, Bartels finds that lower- and middle-class voters are far more responsive to election-year income growth among the richest Americans than they are to overall economic growth or to growth within their own income brackets. In this paper, we examine this troubling feature of U.S. electoral politics in comparative perspective, asking (i.) how widespread class biases in economic voting are in advanced democracies and (ii.) what generates them. Analyzing electoral behavior in three OECD countries (Canada, Sweden, and the United Kingdom), we find clear evidence of class-biased economic voting with substantively important electoral consequences outside the United States. Most surprisingly, we find that the class bias is not limited to national contexts characterized by market-liberal norms and institutions. We then propose two possible mechanisms that might contribute to the class bias — an informational mechanism and an ideological mechanism — and test for their operation in the United States and Sweden. The results are highly consistent with the operation of both mechanisms in the United States and weakly suggestive of an informational effect in Sweden.
Timothy Hicks 
Great research. I hope that you will release the Stata code somewhere — it would make it very easy to build on your work by replicating and porting the analysis to other countries.
Glad you like the paper. The Stata code is the easy part. It goes something like:
probit incumbent_vote mean_growth percentile_growth incumbent_tenure incumbent_pid if resp_inc_low == 1, vce(cluster year)
The time-consuming bit is collecting all the election surveys for a country, coding up the variables in a poolable/comparable way, and adding in the various income quintile data series so that you have values for those variables. The more elections that are added, the better, as it’s the election-N that is relevant for the inferences (rather than the actual N).
Anyway, we’re still working on this project, so there should be plenty more to come…