SocImages has an excellent post today about perceptions of the economy based on an April, 2011 Gallup poll. Some of the interesting findings:

[M]ore than half of respondents said the U.S. is in either a recession or a depression, that’s down significantly from the 69% who said so in late 2008, while the 27% who said the economy is growing is an enormous jump compared to the mere 3% who thought it was in September 2008 …

[T]hose making less than $30,000 a year had a notably more negative outlook on the economy than those with higher incomes. Not only were they more likely to think the economy is doing poorly, but nearly half thought we’re experiencing a depression — twice as high as the proportion of those making $75,000/year who thought so.

The most interesting finding, to me, concerns the partisan split in perceptions of the economy. Here’s the results table, borrowed from SocImages:

So, the main finding is that three times as many Dems as Reps think the economy is growing (42% vs. 14%) while 68% of Reps think we’re in a recession/depression vs. 43% of Dems. With access to cross-tabs, we could figure out quickly if the gap in perceptions by income and by party ID gap are connected – Republicans not actually being the party of the rich (in terms of voters, at least), it’s possible that the two are the same thing. But the gap by party is actually larger than the one by income, so that seems unlikely to explain everything. Suppose there remains a consistent partisan gap. Why?

I wonder if part of the answer has to do with the context of the modern public opinion poll, especially as conducted by Gallup. For more than 70 years, Gallup has been associated with presidential polls (see Sarah Igo’s The Averaged American on the history of the Gallup poll). Polling and politics have become almost synonymous – academics refer to their tools as surveys, politicos deal in polls, and there is even linguistic conflation between elections and surveys (the Iowa Straw Poll vs. the Gallup Poll). All this to say, I wonder if the polling context is one that is already politically primed, and specifically primed to presidential politics. If that were the case, we would predict that Democrats and Republicans would give much more similar answers if asked in face to face contexts, or in say a psychology experiment, when not primed with any overtly partisan information, than they would in a poll, even if the poll never asked about politics (which of course these polls did, though I don’t have the question ordering in front of me).

That would of course still be a fascinating finding, but it would mostly be a finding about the politicization of public opinion, and not as much as one about the way Americans “really” see the economy. An equally plausible, and interesting, finding would be that the economy itself has such a political valence that all our feelings towards it are tinted by the current political climate – because I’m a Dem and a Dem is president, the economy must be doing better than I would otherwise think (or vice versa). A way to test that explanation would be to compare questions about the assessment of the economy as a whole to more targeted questions – how is your family doing? How hard would it be for you to get a new job? Etc.

By the way, if anyone wants to hunt for the data and run these analyses, I’ll gladly post them here. Or if anyone can think of other consequences of the thesis that polls are always already politicized that could be easily tested in the data, or knows of good existing research on the subject, please write a comment!