Quattrone and Tversky: Contrasting rational and psychological analyses of political choice
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.
Quattrone and Tversky. 1988. Contrasting rational and psychological analyses of political choice. APSR 82:719-36.
The authors illustrate how prospect theory (admittedly a descriptive enterprise) predicts certain outcomes better than rational choice theory. They do this by framing a choice between two options (with identical expected value) in terms of either gains or loss. Rational choice would predict an identical outcome, since the expected outcomes of each choice are identical now matter how they are phrased; yet they find consistently different results by changing the reference point (i.e. by framing it as gain or loss), suggesting a weakness in rational choice.
See also Quattrone and Tversky (1984).
The authors ask Stanford and Berkeley undergrads hypothetical questions. Different groups get slightly different questions; for example, a policy question is framed in terms of the unemployment rate to one group, but in terms of the employment rate to another group. Both questions have the same real outcomes, but they are presented differently. The fact that students respond differently to the two questions reveals interesting findings that rational choice would not predict. The method raises questions about external validity, however, as it based entirely on hypothetical questions.
Key Concepts and Findings
- CONSISTENCY (invariance): preference order amongst equal prospects should not depend on how they are described or presented (according to rational choice). Rational choice depends on the real outcomes, not their presentation.
- RISK AVERSION: We aren't willing to take risks in order to get a gain. Given the choice between $100 (for sure) or $200 (with 50% chance of getting nothing), people prefer the $100, even though the expected utility is the same for both options.
- RISK SEEKING: We will gladly take a risk to avoid a loss. Given the choice between losing $100 (for sure) or $200 (with 50% chance of losing nothing), we gamble on the $200 loss, even though the expected utility is the same either way.
- LOSS AVERSION: We would rather keep a less-than-satisfying status quo than risk an even worse outcome. For example: Suppose the US economy is a little stagnant, but it's still doing better than comparable countries. The incumbent proposes to continue with status quo policies, a low-risk strategy. The challenger proposes risky policies; some economists predict economic growth, others predict decline. We do not want to risk a loss. We will stick with the incumbent (and keep stagnation) rather than vote for the challenger (who might bring growth or not), even though the expected utility is the same either way (i.e. the status quo).
- REFERENCE POINTS: It matters what the status quo or reference point is. In the "loss aversion" example I just gave, the conclusion would be completely different if the US were currently doing worse than comparable countries. Compare Problems 3 and 4 on p 723.
- RATIO-DIFFERENCE: We don't perceive raw numbers, we perceive ratios. Thus, if you tell me that one city's unemployment rate is 5% and 10%, I will register that as "twice as much," a big deal. But if you give me employment rates instead (90% and 95%), I will register that as "essentially the same" and not worry about it.
- DECISION WEIGHTING: See Fig 2 (p 730). We underweight most probabilities. However, we overweight small probabilities.
- CAUSAL VS DIAGNOSTIC CONTINGENCIES. Right or wrong, people perceive correlations as causation. In other words, we perceive "diagnostic contingencies," not "causal contingencies." Thus, we expect a correlation of attitudes and actions; this leads to the VOTER'S ILLUSION (below).
Implications for Politics
- INCUMBENTS are less risky than challengers. "The less risky incumbent should fare better when conditions are good than when they are bad" (p 724). It's worth gambling on a risky challenger only if a continuation of the status quo is unsatisfactory.
- In BARGAINING, loss aversion can be a problem. Each concession I make (a loss) looms larger than the concessions gained from the other guy. Thus, we will each think we are offering more than we are getting.
- VOTER'S ILLUSION: If attitudes and actions appear correlated, then we presume some causal linkage. Thus, we conclude that "if people like me don't vote, then my side will lose;" because I see my attitudes as correlated with my participation, I assume a causal linkage between having attitudes like mine and participating. As a result, we dramatically overestimate the probability that our individual vote will matter, since we assume that "if I don't vote, then people like me won't vote either."
- The illusion is manipulable: If you are a partisan and you believe that the election will be driven by independents, then you won't worry about voting. But if you are a partisan and believe that the election will be driven by which party gets higher turnout, then I believe that I must turn out for my party to win. If I don't, then neither will members of my party.
Implications for Political Science
If our formal models are based on an imperfect rational choice, then what do we need to change to eliminate this systematic flaw?