Notes on Degrees of Rationality in Politics ( Dennis Chong )

Assumptions of Rational Choice Theory

• Rational choice theory is both a normative standard and an empirical model of behavior.
• As a theory of behavior, it predicts (or prescribes) how an individual will (or should) choose from alternative courses of action given objectives and beliefs about the instrumental relationship between alternatives and goals.
• The economic version is often called omniscient rationality: it assumes perfect, unbounded, or substantive rationality.
• A perfectly rational individual has:
  • a complete and coherent set of preferences,
  • gathers information as needed by the significance of the choice,
  • forms beliefs about alternatives reflecting relevant information,
  • and chooses the action optimally related to beliefs and goals.
• Central empirical questions: Do people choose actions optimally related to their beliefs/goals? Are preferences coherent? Do beliefs reflect the evidence gathered? Do individuals seek the proper amount of evidence?
• Observed departures in politics: people often support policies or engage in behavior (e.g., voting) that do not maximize self-interest; preferences can be unstable, inconsistent, and framed by how options are presented; beliefs may not update with new evidence; information gathering may be insufficient.
• Paradox: it is irrational for the average citizen to invest substantial time and effort in becoming informed and making political decisions (Downs, 1957; Hardin, 2006).
• The economics of information constrains rational choice because initial conditions (economizing on effort) can compromise preferences/decisions.
• Substantive rationality depends on optimal use of information to achieve goals; questions remain: can citizens who rationally pay slight attention to politics still make rational decisions?
• Expected quality of decisions varies with decision context and decision-maker.
• Standard vs bounded rationality:
  • Omniscient/Perfect rationality: predict choices given goals and circumstances, assuming capacity/knowledge to infer properly.
  • Bounded rationality: instrumental in actions but with limited time/resources; decisions may involve mistakes in gathering/assessing evidence and reasoning from means to ends (Hirschman, 1982; Riker, 1995).
• To explain/predict behavior we study:
  • subjective motivations/goals,
  • information possessed,
  • inferences about consequences of actions.
• All deliberative procedures can be viewed as boundedly rational, with varying degrees of substantively rational outcomes (Simon, 1985, 1995).
• If not universally rational, people may be more rational in high-stakes or highly motivated contexts; they may be rational within their bounded knowledge, capacity, and motivation.
• The chapter aims to explore both economic and psychological rationality, outlining assumptions and variations to accommodate a more realistic psychology, and evaluating whether citizens make optimal self-interested policy choices, whether low/information rationality is substantively rational, whether beliefs are updated rationally, and whether preferences are stable across issue framings.
• Key methodological implication: since decision-making incurs costs, rational choice must account for incentives, learning, and bounded processing.
• Normative standards of decision-making vary across contexts: information responsiveness is valued in some contexts, stability of preferences in others; there is inconsistency in normative evaluations (e.g., party cues and motivated reasoning).

Economic and Psychological Rationality

• Economic rationality vs bounded rationality differ in information level and cognitive ability.
• Bounded rationality allows variation across individuals and contexts; decision procedures depend on issue importance, motivations, abilities, and predispositions.
• Humans are not always able to make optimal choices; mistakes are common as they balance effort and optimization.
• Famous quotation: Hirschman (1982) on mistake making as a characteristic of human action; Riker (1995) on instrumental accuracy vs. attempted instrument choice.
• Psychological/experimental evidence reveals irrationality in many contexts: people err in decision-making, fail to maximize utility, or lack consistent preferences (Tversky & Thaler, 1990).
• Debates exist on external validity of experimental findings: incentives, learning opportunities, and debate can affect performance; many experiments involve novel problems with little learning opportunity; debates guide evaluation of alternatives (Camerer & Hogarth, 1999; Chase et al., 1998; Gigerenzer, 1991; Riker, 1995; Wittman, 1995).
• Inconsistencies in choice and miscalculations are more likely when stakes are low and problems unfamiliar.
• Elster (1990) asks whether people deal irrationally with important problems.
• Acknowledgement that incentives, education, and debate can improve rationality, and decision-making can be placed on a continuum from intuitive System 1 to reasoned System 2 processing (Kahneman, 2003).
• Dual-process models distinguish central/systematic (effortful) vs peripheral processing; central processing requires more effort but yields more accurate judgments (Petty & Cacioppo; Fazio & Olson).
• Satisficing: individuals settle on an adequate outcome and stop searching when a satisfactory standard is met (Simon, 1957).
• The role of incentives: monetary incentives can increase effort and performance when effort matters and cognitive capital exists; when capital is insufficient, extra incentives may be futile (Camerer & Hogarth, 1999).
• In mass politics, the impact of incentives is ambiguous due to small expected value of individual political decisions (e.g., voting) and availability of shortcuts (partisan cues, ideology).
• The broad takeaway: bounded rationality extends economic reasoning rather than contradicts it; decision-making is shaped by costs, information, and cognitive limits.

Do People Maximize Self-Interest?

• A direct test: do people maximize utility when taking public policy positions?
• In survey data, self-interest often weakly predicts policy preferences compared with general orientations like ideology, partisanship, and values.
• Observations: voters often focus on national economic conditions and broad welfare goals rather than their personal economic status when evaluating candidates/policies.
• Implication: self-interest has only modest explanatory power for political choices; this challenges a simple, ubiquitous self-interest model of political behavior.
• The chapter explores explanations for this weak correlation, conditions that strengthen self-interest effects, and theoretical implications for rationality.
• Two broad sources of explanation:
  • conceptual/measuring issues: how we define and detect self-interest (narrow vs broad concepts);
  • cognitive/political constraints: information, framing, and perception influence self-interest’s impact.

2.1. Defining and Measuring Self-Interest

• Self-interest is often defined narrowly as tangible, immediate, personal or family benefits of a policy; broad conceptions include expressive actions or long-term effects.
• Narrow self-interest helps separate effects of interests from values, aiding assessment of how much interests matter.
• Survey analyses commonly infer self-interest by evaluating whether groups most affected by a policy support it more than others, assuming similar beliefs about policy impact.
• Risk: respondents may oppose or support policies based on mistaken beliefs about impacts; preferences may be consistent with their analysis of interests, or they may knowingly choose non-self-interested positions.
• Information economics: citizens have little policy knowledge to base self-interest on; immediate personal gains may be uncertain or perceived through partisan/ideological frames.
• Thus, self-interest can be misestimated, and its apparent influence can be biased by information and beliefs.

2.2. When Costs and Benefits Are Magnified

• Strong self-interest effects emerge when policy impacts are clear and direct for some groups (e.g., tangible costs or benefits).
• Examples:
  • Homeowners closer to Proposition 13 benefits vs non-homeowners; tangible, immediate tax relief increased support among homeowners.
  • Smokers more opposed to smoking restrictions and taxes than nonsmokers;
  • Regular drinkers less supportive of enhanced drunk-driving controls;
  • Gun owners less likely to support handgun bans or waiting periods.
• Common thread: when a policy clearly benefits or harms a specific group, instrumental reasoning and self-interested decision-making are more likely to manifest.
• Agreement across traditions (social psychology and rational choice) that visible, targeted effects amplify self-interest.

2.3. Priming Self-Interest

• Awareness and attention modulate the relation between self-interest and policy choice.
• People with higher political information or who are attentive are more likely to know how a policy affects them; less-informed individuals rely more on cues.
• Critical factor: whether material benefits are visible/accessible to a decision-maker.
• Empirical findings:
  • Sears & Lau (1983): self-interest effects strengthen when respondents are asked about personal economic situations before candidate preferences.
  • Priming self-interest increases egoistic justifications and reduces acceptance of broad, symbolic reasons (Young et al., 1991; Taber & Young).
  • Chong, Citrin, and Conley (2001): in three policy issues (Social Security, home mortgage interest deductions, domestic partner health benefits) those with clear beneficiaries act on self-interest without prompting; but priming magnifies this effect; those with less stake rely more on values when primed.

2.4. Uninformed Self-Interest

• Bartels (2008) on Bush tax cuts: public support correlated with perception of paying too much tax, not nuanced considerations of inequality or program effects.
• Myopic voters may ignore downstream consequences (e.g., tax cuts funded by cuts in government programs or increased state/local taxes).
• Other findings: Californians who supported Proposition 13 may have been driven by short-term personal relief rather than broader tax/allocations considerations.
• The public often treats themselves as consumers; short-sighted decisions resemble consumer behavior (Conlisk, 1996).
• Surveys show many people overestimate or misattribute effects of tax policy; information framing changes perceptions of trade-offs (Hacker & Pierson, 2005).
• In competitive frames, supporters may favor short-term gains while opponents favor long-term social benefits; both views can be consistent with self-interest, depending on framing.

2.5. When Self-Interest Matters

• Core question: not whether self-interest matters, but when it does.
• Self-interest matters when people have a stake in a policy and can recognize it; benefits are visible and policy is transparent.
• When stakes are debatable, consequences are hard to discern, or individuals are not directly affected, people rely on general political orientations (ideology, partisanship) to guide decisions.
• Future research suggestions:
  • study a wider range of issues with varying size/clarity of benefits/costs;
  • improve measurement of both self-interest and values by refining beneficiary class definitions and relevant values.

Low-Information Rationality?

• Politically, citizens often economize on deliberation, hoping to make reasonable decisions with limited information.
• Is bounded rationality compatible with good outcomes? Do limited-information strategies yield similar results to fully informed decisions?
• Consequences of being uninformed may not be as dire as once thought: people can learn just enough through everyday routines and by using heuristics available in the political environment (Downs, 1957; Fiorina, 1981; Popkin, 1991; Lupia & Sears, 2000).
• Voters may use simpler data such as recent economic trends, candidate partisanship, ideology, and cues from opinion leaders/endorsements to evaluate candidates.
• This suggests people can form reasonable judgments without deep issue knowledge, by relying on accessible indicators.

3. Measuring Performance

• Normative standard: a perfectly informed voter would know all relevant information, analyze it properly, and choose the option that maximizes expected utility.
• A neutral method: compare preferences of the better-informed individuals to those of less-informed individuals, controlling for demographics. Differences imply information changes beliefs/preferences.
• Experimental method: randomly assign information to a treatment group and compare to a control group; or use before/after designs to gauge information effects.
• Another approach: compare the preferences of the heuristic-informed group with those of the more fully informed group.
• Findings generally show information increases alignment between preferences and goals, but the effect varies by issue, stakes, and individual characteristics.

3.1. Measuring Performance

• Normative standard is idealized (omniscient voters); in practice, researchers test how better-informed individuals differ from less-informed ones.
• Approaches include:
  • comparing preferences of more-informed vs less-informed groups while controlling for demographics,
  • experiments giving new information to assess shifts in preferences,
  • comparing heuristic-informed voters to fully informed voters.

3.2. The Impact of Information on Preferences

• Across studies, more information tends to shift preferences away from uninformed positions, indicating information matters for optimizing choices.
• Representative findings:
  • Delli Carpini & Keeter (1996): knowledge gaps sharpen with information within groups (men vs women, blacks vs whites, younger vs older) among the best-informed members.
  • Bartels (1996): average deviation from fully informed preferences across elections (1972–1992) was about 10%; with random choice, deviation could be ~20%; thus information reduces error by roughly half.
  • Althaus (1998): fully informed opinion tends to be more liberal on government services/taxes but more cautious on expanding government power; on social policy, generally more liberal except on affirmative action.
  • Gilens (2001): specific information alters preferences beyond general knowledge (e.g., crime rates and prison construction; foreign aid as a small federal share).
  • Lau & Redlawsk (1997; 2006): most voters vote consistently with stated beliefs/interests; two-thirds to three-quarters may be voting correctly in light of complete information.
• Open question: even if information changes beliefs, is the cost of obtaining that information worth the benefit of new preferences? Some decisions (like voting) have small personal impact; the practical value of right decisions can be limited.

3.3. Knowing Just Enough versus Knowing a Lot

• Heuristics are not a complete substitute for knowledge; some issue salience can prompt cue acquisition, e.g., Lupia (1994) on auto insurance initiatives: knowing the position of the industry is enough for voting behavior; additional details added little predictive value.
• Ambiguity: using better-informed as a standard for optimality may be problematic; information value depends on relevance to the decision and how it changes preferences.
• Mixed evidence: information improves decision quality, but top-informed individuals are often the most partisan/ideological, which can bias interpretation of evidence (leading to motivated reasoning).
• Price of cutting corners exists, but deviations from fully informed preferences may be tolerable depending on standards for decision-making. In some cases, highly informed individuals may still engage in ideological distortions.

Are Beliefs Formed and Updated Rationally?

• Normative model: gather information, weigh applicability, adjust beliefs if information is relevant to how alternatives relate to goals.
• Evidence shows that reasoning with limited information and low motivation can produce motivated reasoning, which can undermine the informational value of new evidence.
• Although motivated reasoning can be logically coherent given preferences, it reduces the value of information in politics relative to simpler cues.

4.1. Biased Information Processing

• A core bias: people shape beliefs to fit preferences rather than forming preferences from objective evaluations of the world.
• The foundational principle of rationality: desires should not guide beliefs; this is violated in motivated reasoning and cognitive dissonance reduction (Elster, 1990).
• Classic demonstrations:
  • Lord, Ross, and Lepper (1979): people interpret evidence in a biased manner consistent with pre-existing beliefs.
• In politics, where disputes over interpretation are common, individuals often interpret the same facts differently depending on partisan or ideological perspectives.
• Implication: even rational actors may end up with biased beliefs due to cognitive constraints and motivated reasoning.

4.2. Motivated Reasoning

• Motivated reasoning refers to the tendency to interpret information in a way that supports already-held beliefs or desired policy outcomes.

• While it can be logically coherent, it undermines normative expectations of objective information processing.

• More detailed discussion points (in practice): interaction of information, framing, identity, and incentives; role of party cues; asymmetries in processing of evidence depending on affective or identity relevance.

• The chapter highlights that normative standards of decision-making vary across topics; information responsiveness is valued in some contexts, but stability of preferences or resistance to information might also be valued in others.

• Overall takeaway: beliefs are formed/updated in bounded rationality contexts, where information, framing, and motivated reasoning interact to shape political judgments.

• Notes for exam preparation:

  • Distinguish between omniscient rationality and bounded rationality; know definitions of complete, transitive preferences and the role of a utility function.
  • Be able to express the rational choice rule under uncertainty: $EU(a) = \sum<em>i p</em>i(a) \; u(o<em>i)$ and the optimal action: $a^* = \arg\max</em>{a} EU(a).$
  • Understand System 1 vs System 2 (dual-process) and the concept of satisficing by Simon (1957).
  • Recognize the paradox: rationality vs information costs in politics; bounded rationality as an extension rather than a contradiction.
  • Remember key empirical findings:
  • self-interest often weaker predictors than ideology/values (2.1–2.5)
  • when costs/benefits are magnified, self-interest effects strengthen (2.2)
  • priming/self-interest can amplify self-interest explanations (2.3)
  • uninformed self-interest can mislead; information framing matters (2.4, 2.5)
  • low-information rationality: info can shift preferences, though effects vary (3.1–3.3)
  • beliefs updated rationally vs biased processing; motivated reasoning (4.1–4.2)