Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I even say who wrote them. If you have more recent summaries to add to this collection, send them my way I guess. Sorry for the ads; they cover the costs of keeping this online.
Opp. 2001. Why do people vote? The cognitive-illusion proposition and its test. Kyklos 54: 355-378.
Opp puts forth the "cognitive-illusion proposition," the idea that people mistakenly believe they are able to influence the outcome, to explain why voters actually participate in elections. He uses 1998 German Allbus survey to test this hypothesis.
Opp focuses on the P and D terms of the rational choice PB - C + D model. He argues that voters have an illusion that their P, the probability of being decisive, is much higher than it actually is (which is all but zero). Thus, people actually have basically no influence but misperceive this fact, and this misperception determines electoral participation; voters mistakenly view voting as instrumental, that it can achieve a goal or outcome. He breaks down D into two separate components, a voting norm and social rewards. He defines the voting norm as the obligation a voter feels to participate, while social rewards include recognition or status for voting or social sanctioning for not voting.
The voting had already been performed by the time the survey was enacted, so the causal chain is backward but Opp argues that preferences and beliefs are stable enough to use this data. First, Opp looks at the Perceived influence variable and finds that only 5.7% of respondents say they have no influence but 63.8% have "strong" influence, which shows evidence for his hypothesis that voters suffer from a cognitive illusion
In model 1, limiting the independent variables first to discontent and Perceived influence, discontent decreases likelihood of voting while perceived influence has the expected positive impact, although surprisingly not that strong. In model 2, he examines these two plus an interaction between discontent and influence; the interaction term is the most significant and of largest magnitude. Bringing in the D terms in model 4 explains a little more of the variance, with both the norm and politically active friends positively correlated with voting as expected. The control variables added in model 5 further increase the explanatory power of the model, with all the independent variables still significant and in the hypothesized direction.
The age variables had the highest standardized regression coefficient of any variables, which might need explanation in terms of the model, if this is so important. Furthermore, the cognitive illusion variable, so crucial in his discussion, had a relatively pedestrian effect.
Opp is especially strong in emphasizing that individuals are irrationally assuming they have a role in the election. What may be individually rational for an individual can, however, lead to non-Pareto optimal outcomes, with the obvious example being the prisoner's dilemma. Transcending a prisoner's dilemma by whatever method and receiving the dual co-operate outcome is better than the equilibrium dual defect, and a move towards this direction could easily be considered rational. If a candidate's or party's voters can coordinate their action to ensure that enough vote so that they achieve electoral victory, I would argue that even those who voted will have behaved rationally. The party of purely maximizing rational voters (who almost always abstain) would always lose and would always receive the suboptimal outcome. I would further argue that individual voters are having an impact on the election by voting and coordinating; although individually their vote is not decisive, the pattern of coordination is necessary to elect their candidate. Perfect coordination is not necessary, only "good enough."
Or, perhaps putting it in less rational terms, given that it is clear that people do in fact vote in large numbers from our past experience, it is going to be necessary for rational actors to vote if they have any hope of influencing the outcome. The bounded rationality or altruism of other voters may thus encourage the rational actors to vote because voting is a complementary strategy, rather than substitution. Also, Opp's rejection of the instrumentality of voting also seems harsh. If the D term is all that matters, as a voter my choice of candidates is meaningless and I should pick randomly.
Recall that when Blais asked respondents whether they could influence the election outcome, they generally understood that they had an exceptionally low probability of being decisive. But Opp asked the question differently. He didn't ask whether you think you'll be decisive; he asked whether he thinks you'll have an influence--and most think they do have a "strong influence" on the election. To what extent does this difference in questioning have an effect? It appears that Opp's respondents are not deluded; they interpret the question in terms of contributory causation. Like Schlozman's respondents, Opp's respondents want to be a part of bringing about a collective outcome.
For the record, here's Opp's specific question: "Please, tell me to what extent 'you personally' could exert influence in politics when you carry out the actions on these cards" (p 364). In response to the "voting" card, many voters thought they could strongly influence politics.
Research on similar subjects