Andrew questions David Brady’s claim that “partisan preferences actually drove the perception of the economy” under George W. Bush for “the first time . . . in all of our polling data.” My analysis Andrew mentions of data from 1988 focuses on responses to a seldom-repeated set of questions about how specific economic and social conditions changed over the eight years of the Reagan administration. However, it is easy enough to track the impact of partisanship on perceptions of economic conditions using the standard ANES question about whether the national economy has gotten better or worse over the past year, which has been asked in every presidential election year since 1980. Here are year-by-year estimates:
(Ordered probit coefficients because the categorical distribution of perceptions is quite skewed in some years. The 2012 data point is based on May-July data from the Cooperative Campaign Analysis Project, which employs a somewhat different measure of economic perceptions, so is only approximately comparable.)
Clearly the tendency to see the economy through a partisan lens is not new. (The t-statistics for these coefficients range from 6.8 to 20.5 in the ANES data; the t-statistic for the larger CCAP sample is 46.5. So these are “statistically significant differences” by conventional standards.) Indeed, it is not even clear that the tendency is increasing. The partisan bias in economic perceptions was large in 2004, but not much larger than in 1984; and it was actually relatively small in 2008 (perhaps because even Republicans had a hard time denying that the economy was getting worse in that year).
It is also easy enough to track the impact of economic perceptions on presidential voting behavior, net of partisanship. Again, there is a lot of year-to-year variation but no clear trend:
Making sense of these fluctuations would be a considerable contribution to our understanding of presidential elections. Al Gore’s failure to run on the economy is often blamed for the small coefficient in 2000, but there was a lot of talk about the economy in 2008 and the coefficient was even smaller. An occular comparison with Lynn Vavreck’s data on economic content in campaign ads and media coverage suggests no clear overall relationship. If the coefficient in 2008 is small because economic perceptions were so skewed, why are the coefficients in 1980 and 1992—years in which economic perceptions were almost as skewed—so much bigger? For all of our collective interest in “priming,” I don’t think we have this nailed down by any means. (However, thanks to the existence of the ANES project, we do have much of the data we would need to learn something.)