The Ecological Fallacy Revisited: Aggregate- versus Individual-level Findings on Economics and Elections, and Sociotropic Voting

Abstract

Several aggregate-level studies have found a relationship between macroeconomic conditions and election outcomes, operating in intuitively plausible directions. More recent survey-based studies, however, have been unable to detect any comparable relationship operating at the individual-voter level. This persistent discrepancy is puzzling. One recently proposed explanation for it is that voters actually behave in an altruistic or “sociotropic” fashion, responding to economic events only as they affect the general welfare, rather than in terms of self-interested “pocketbook” considerations.It is argued here that the discrepancies between the macro- and microlevel studies are a statistical artifact, arising from the fact that observable changes in individual welfare actually consist of two unobservable components, a government-induced (and politically relevant) component, and an exogenous component caused by life-cycle and other politically irrelevant factors. Because of this, individual level cross-sectional estimates of the effects of welfare changes on voting are badly biased and are essentially unrelated to the true values of the behavioral parameters of interest: they will generally be considerable underestimates and may even be of the wrong sign. An aggregate-level time-series analysis, on the other hand, will often yield reasonably good (if somewhat attenuated) estimates of the underlying individual-level effects of interest. Therefore, in this case, individual behavior is best investigated with aggregate- rather than individual-level data.It is also shown that the evidence for sociotropic voting is artifactual, in the sense that the various findings and evidence which ostensibly show sociotropic behavior are all perfectly compatible with the null hypothesis of self-interested, pocketbook voting.

The Differential Effects of Economic Conditions and Racial Attitudes in the Election of Donald Trump

“…Referring back to concerns raised by Kramer (1983), we also acknowledge that job loss and local unemployment can both represent a mixed bag of economic phenomena -some of which voters tie to politics, and some of which they do not -and this could lead our models to underestimate the extent to which economic distress is associated with voting behavior among those who do attribute such distress to decisions made by political leaders. Table 1 In substantive terms, holding all else constant, a one unit movement in the racially conservative direction (disagreement) on the first FIRE battery item, anger that racism exists, is associated with nearly a 36% increase in the relative likelihood that a white votevalidated respondent reported voting for Trump over Clinton, a 24% increase in the relative likelihood that they didn't vote relative to voting for Clinton, and a 27% increase in the relative likelihood that they reported voting for a minor party candidate as opposed to Clinton.…”

“…Some scholars have argued that community-level economic conditions will be more strongly associated with voting behavior (Feldman 1982;Kinder, Adams, and Gronke 1989) than individual economic distress. However, Kramer (1983) points out that changes in an individual's financial circumstances can be affected by politically relevant factors such as government policy as well as politically irrelevant factors such as exogenous shocks or life cycle considerations (retirement, for example) -and that for this reason one should not necessarily expect economic voting to be purely sociotropic. While our dataset does not allow us to fully address the aggregation problems laid out in Kramer's article, we are able to address some of these concerns using the data we have available.…”

Voting and the Macroeconomy

“…The contribution to understanding instability of macroeconomic voting-particularly cross-national instability-has been substantial, yet mainly empirical, without reference to an explicit theoretical foundation. The absence of a theoretical referent is odd because a compelling framework, which might have been used to practical advantage in empirical work on instability of economic voting, had emerged during the first part of the 1970Sin the unobserved errors-in-variables and latent variables models of Goldberger (1972aGoldberger ( , 1972b, Griliches (1974), Joreskog (1973), Zellner (1970), and others, and those models had been applied to a wide variety of problems in economics, psychology, and sociology during the following twenty years.31Moreover, the errors-in-variables specification error model was applied directly to the problem of unstable economic voting a full decade before the appearance of Powell and Whitten (1993) in a brilliant paper by Gerald Kramer (1983),which was targeted mainly on the debate launched by Kinder and Kiewiet (1979) concerning the degree to which voting behavior is motivated by personal economic experiences ("egocentric" or "pocketbook" voting),32 rather than by evaluations of government's management of the national economy ("sociotropic" or "macroeconomic" voting). 33 Kramer's argument, which subsumed the responsibility hypothesis, was that voters rationally respond to the "politically relevant" component of macroeconomic performance,where, as in the subsequent empirical work of Powell and Whitten and others, politicalrelevance was defined by the policy capacities of elected authorities.…”

Real-Time Associations Between Engaging in Leisure and Daily Health and Well-Being

“…Although not all examples of within-and between-person approaches produce seemingly "opposite" relationships (recall that they test separate questions and thus may either agree or disagree), this example highlights the ecological fallacy [17,18]. The ecological fallacy states that relationships between variables at one level (e.g., between individuals) cannot be assumed to exist at the same magnitude and direction at another level (e.g., within individuals) (for discussion of the different applications of between-and within-person models, see [19]).…”