Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, POLITICS ( (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy and Legal Notice).

date: 05 December 2019

The Representativeness Heuristic in Political Decision Making

Summary and Keywords

The representativeness heuristic was defined by Kahneman and Tversky as a decision-making shortcut in which people judge probabilities “by the degree to which A is representative of B, that is, by the degree to which A resembles B.” People who use this cognitive shortcut bypass more detailed processing of the likelihood of the event in question but instead focus on what (stereotypic) category it appears to fit and the associations they have about that category. Simply put: If it looks like a duck, it probably is a duck. The representativeness heuristic usually works well and provides valid inferences about likelihood. This is why political scientists saw it as an important part of a solution to an enduring problem in their field: How can people make political decisions when so many studies show they lack even basic knowledge about politics? According to these scholars, voters do not need to be aware of all actions and opinions of a political candidate running for office. To make up their mind on who to vote for, they can rely on cues that represent the performance and issue position of candidates, such as the party they are affiliated with, their ranking in the polls, and whether (for instance) they act/appear presidential. In other words, they need to answer the question: Does this candidate fit my image of a successful president? The resulting low-information rationality provides voters with much confidence in their voting decision, even though they do not know all the details about the history of each candidate. Using heuristics allows relatively uninformed citizens to act as if they were fully informed.

Despite this optimistic view of heuristics at their introduction to the discipline, they originated from research showing how heuristic use is accompanied by systematic error. Tversky and Kahneman argue that using the representativeness heuristic leads to an overreliance on similarity to a category and a neglect of prior probability, sample size, and the reliability and validity of the available cue. Kuklinsky and Quirk first warned about the potential effect of these biases in the context of political decision-making. Current research often examines the effects of specific cues/stereotypes, like party, gender, race, class, or more context-specific heuristics like the deservingness heuristic. Another strand of research has started exploring the effect of the representativeness heuristic on decision-making by political elites, rather than voters. Future studies can integrate these findings to work toward a fuller understanding of the effects of the representativeness heuristic in political decision-making, more closely consider individual differences and the effects of different contexts, and map the consequences that related systematic biases might have.

Keywords: heuristics, biases, decision making, stereotypes, representativeness, low-information rationality, cognitive shortcut, similarity, probability, political decision making

Roughly stated: People who use the representativeness heuristic judge something based on the mental category it resembles most rather than take account of all individual aspects and information available (Tversky & Kahneman, 1974). Doing so makes decision-making much easier, requires much less prior knowledge/information, and usually produces adequate inferences (Kahneman & Frederick, 2002). The representativeness heuristic is a general purpose heuristic that applies to many (political) situations and is used frequently (Gilovich, Griffin, & Kahneman, 2002; Mondak, 1993).

For political science the discovery of the representativeness heuristic was important, since it provided a way out of an age-old dilemma: How can citizens participate meaningful in political life if (most of them) know little or nothing, lack coherent belief structures, and hold unstable and inconsistent opinions, and how are they, at the same time, able to provide answers with great confidence to the most obscure political questions asked them in opinion surveys (Sniderman, Brody, & Tetlock, 1991)? Heuristics provide a way out, since they enable citizens to form political opinions from a “low-information” environment (Popkin, 1994). People have many scenarios and ideas to help them navigate everyday life, which they might draw upon by linking them to the heuristic cues available in the political context. For example, their ideas about who is trustworthy, competent, or deserving of help might help voters decide whether a political candidate is trustworthy and competent and whether he or she talks/acts kindly to those who are deserving help, without requiring this voter to know and understand all this candidate’s policy ideas and the details of his or her political track record.

However, there is one important problem with this view: The representativeness heuristic does not necessarily yield correct inferences. A candidate who looks trustworthy and acts trustworthy does not always turn out to be trustworthy. In fact, this was the origin of the psychological research into the representativeness heuristic: that it is accompanied by predictable and systematic biases (Tversky & Kahneman, 1974). One famous example is the conjunction error, illustrated by the “Linda problem.” Participants in an experiment are provided with the following description: “Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy in college. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.” They are then asked what it is likely she is doing now: Is she a bank teller (A), or is she a bank teller and active in the feminist movement (A&B)? Tversky and Kahneman (1983) find that a large majority of people indicate that she is more likely to be a feminist bank teller (A&B) than a bank teller (A). They show that this is because people believe that Linda resembles a feminist bank teller more than she resembles a bank teller. A different way to approach this question is to use extensional reasoning, that is, to compare the extensions (set of elements, here individuals, belonging to) of each of the categories (here bank tellers and feminist bank tellers). Using such logic, it cannot be the case that Linda is more likely to be a feminist bank teller than a bank teller, “because every feminist bank teller is a bank teller, but some women bank tellers are not feminists, and Linda could be one of them.” (Tversky & Kahneman, 1983, p. 299). This is called the conjunction error as the conjunction of two sets (i.e., all bank tellers and all feminists) is necessarily smaller or equal to each individual set. This example shows that inferences based on representativeness (Linda resembles a feminist but not a bank teller) may lead to incorrect estimates when reality does not match the stereotype.

As Kuklinski and Quirk (2000) note, such systematic biases are very relevant for political decision-making. It matters which cues voters use to form their representations,1 since some of them are more valid than others. According to Kuklinsky and Quirk, people would do better to use party labels, which are based on longstanding features, rather than campaign appeals about issues, which are made to trigger a less reliable representation. They fear that the use of heuristic cues by voters opens up the door for mass manipulation by political entrepreneurs, since they will be able to sway the public by providing a certain image, rather than a reality.

This article starts by outlining the origin and definition of the representativeness heuristic. It then describes the state of the present literature by focusing on the role of this heuristic for decision-making by voters and how recently scholars have started to investigate its influence on elite decision-making as well. The second part of the article presents a research agenda organized around three related themes: individual and contextual differences, integrating findings on cues, and the consequences of systematic biases. Together these research proposals contribute to a more encompassing, coherent, and systematic description of the role of the representativeness heuristic on political decision-making, as well as provide insights that may aid the study into the effects of individual cues.

What Is Representativeness?

Kahneman and Tversky (1974) explain that when using the representativeness heuristic, “probabilities are evaluated by the degree to which A is representative of B, that is, by the degree to which A resembles B” (p. 1124). The heuristic essentially replaces the difficult assessment of probability, with the easier assessment of resemblance (Kahneman & Frederick, 2002). Using the representativeness heuristic thus involves two kinds of representations:

(1) a prototype (a representative exemplar) is used to represent categories (e.g. bank tellers) in the prediction task; (2) the probability that the individual belongs to a category is judged by the degree to which the individual resembles (is representative of) the category stereotype.

(Kahneman & Frederick, 2002, p. 93)

For example, in judging a presidential candidate, a voter might think of her category of competence and might judge the resemblance of the candidate to the category of competency by using available cues, such as the shape of the candidate’s chin or the forcefulness of her speeches (Kahneman, 2011).

This definition highlights that representativeness is about probability. In this context, probability refers to the assessments of uncertain events. Voters do not know whether a candidate will make a good president, but to decide who to vote for they need to make a prediction about the likelihood that he or she will. Similarly, when judging what Linda currently does for a living, we have to judge which of the given options (bank teller vs. feminist bank teller) is more likely to develop from the provided college description. The focus on probability thus relates to many political questions, as uncertainty is a central element of politics: Voters and politicians alike do not know for sure what all the consequences of a policy will be or whether a certain politician will keep his or her promises.

Representativeness is also about similarity. Linda appears like a feminist because she shares certain features with our mental representation of feminists. However, the workings of representativeness are not restricted to similarity. Representativeness can also refer to a causal or correlational belief (Tversky & Kahneman, 1983). Certain causes or consequences fit better together than others. When the pattern between cause and consequence appears to match one’s representation of what that pattern should look like, the cause and consequence can be thought of as being more representative of each other. Representativeness thus extends beyond the direct attributes of an instance (like that Linda is 31 years old) to its likely causes and consequences (like a socially engaged philosophy student is likely to become a feminist, so feminism is also representative for Linda). Tversky and Kahneman, for example, found that people believed it to be more likely that the Russia and the United States would suspend diplomatic relations due to a Russian invasion of Poland in the coming year (1983) than that Russia and the United States would suspend diplomatic relations that year without a specific cause being provided. If it is easy to think of a cause for something, it is easier to imagine that it will happen. So when a cause is provided for an unlikely event, which fits that event, the occurrence of that event is more representative for things that take place and hence is judged to be more probable. This again might produce biases, since, similar to the Linda example, the (causal) conjunction of a suspension of diplomatic relations and an invasion of Poland can never be more likely than the larger set of possibilities defined by a suspension of diplomatic relations in general, at least according to extensional reasoning.

The use of the representativeness heuristic involves attribute substitution: the target attribute (e.g., the probability of Linda being a feminist bank teller) is assessed through the value of an heuristic attribute (e.g., the degree to which Linda resembles a feminist bank teller) (Kahneman & Frederick, 2002). Since the heuristic attribute is different from the target attribute, systematic biases are likely to occur. In addition to the conjunction fallacy, another prominent bias is called base-rate neglect.2 Kahneman and Tversky (1973) told participants in one of their experiments that they had drawn up personality descriptions of 70 lawyers and 30 engineers. They then presented participants with a personality description of a stereotypical engineer and asked them to rate how likely they believed it was that this person was an engineer or a lawyer. They gave the same personality description to a different group, whom they told the descriptions were drawn from a set of 30 lawyers and 70 engineers instead. When analyzing the answers, Kahneman and Tversky found that the likelihood assessments of both these groups were similar; in other words neither group had taken into account that a random personality description from a pool consisting of 30%—the base rate—engineers should be less likely to be an engineer than a random personality description from a pool consisting of 70% engineers (Kahneman & Tversky, 1973). They repeated this with various other personality descriptions and found the same result. The explanation they give for this neglect of base rates is that the stereotypical representation resulting from the personality description overshadows other considerations in determining likelihood: Participants rely on the representativeness heuristic rather than considering all information presented to them in detail.

In terms of its reliance on attribute substitution, the representativeness heuristic works similar to the closely related availability heuristic. People apply the availability heuristic if they “assess the frequency of a class or the probability of an event by the ease with which in-stances or occurrences can be brought to mind” (Tversky & Kahneman, 1974, p. 1127). Both heuristics work through assessing the probability of an event by the probability of something less complex. They differ in that, in the case of the representativeness heuristic, prototypical associations are used to predict probability, while in the case of the availability heuristic instances that come quickly to mind are used. Such instances might come quickly to mind because they are highlighted in some way, for example, by being encountered very recently. There might be overlap between the two heuristics as well, since prototypical associations likewise are easier to come to mind than nonprototypical associations.

Through the years the representativeness heuristic and the heuristic approach in general have faced various criticisms (for an overview, see Kelman, 2011). One important one can be labeled the “vagueness” critique (see Gigerenzer, 1996). Critics of the representativeness heuristic argue that it is difficult to judge a priori which heuristic will be used by whom, which mental representation will be applied to which situation, and with what effect. Moreover, the replications of the original experiments by, for example, Gigerenzer, Hell, and Blank (1988) show that the extent to which people rely on the representativeness heuristic is sensitive to the problem formulation. If there are no clear predictions about which representations will be used by which people in which context, how can we study their effect outside the controlled environment of the academic lab (i.e., in political life)? In retrospect any event can be explained by heuristics, because it is always possible to infer that someone judged something with respect to the category representative for the outcome once that outcome is known. Representativeness appears to be so general that it is able to explain everything and thereby nothing at all.

Part of an answer to this critique might be that the empirical study of the representativeness heuristic is a work in progress, and further research may better delineate which representations will be used by which people in which context. It is important to consider that Tversky and Kahneman (1974, 1983) did not intend to argue that the representativeness heuristic (or any other heuristic) fully explains all decision making. In their earlier work they rather speak about its role in invoking a “natural assessment.” Such assessments occur spontaneously and consequently may influence a decision but do not determine it, as they may be overridden by conscious deliberation. In fact, many participants upon given the explanation of the Linda experiment indicate that in retrospect they would have decided differently, which illustrates that they are fully capable of using nonrepresentativeness heuristic guided reasoning (Tversky & Kahneman, 1983). Kahneman and Frederick (2002) explain that there are two families of cognitive operations. System 1 is quick, intuitive, and guided by heuristic thinking while the other, System 2, is more deliberative and monitors the quality of System 1 intuitions and might endorse, correct, and override them. Consequently, the more System 2 is activated, the less likely it will be that the resulting decisions will conform to any prediction made based upon representativeness. Researchers should, thus, not be mainly concerned with evaluating whether representativeness drives decision-making but rather (empirically) investigate the conditions that enhance or reduce its influence, as well as when biases are more or less likely to occur (Gilovich & Griffin, 2002).

Voters and the Representativeness Heuristic

Initially, the greatest interest in the representativeness heuristic came from scholars seeking to explain meaningful voter participation in politics. At the time, the “minimalist perspective” argued that most voters know little or nothing, lack coherent belief structures, and hold unstable and inconsistent opinions (Luskin, 1987; Sniderman et al., 1991). By cleverly using heuristics, voters could still attain “low-information rationality,” in which they judged candidates by, for example, personal characteristics, campaign performance, and associated party leaders they had learned about in the past (Brady & Sniderman, 1985; Popkin, 1994). From this information voters can roughly deduce their policy preferences, governmental competence, a candidate’s place in the party, and the credibility of her platform. Since the true character of a candidate, his or her future performance, or the outcome of a political decision tends to be uncertain, voters need to make probabilistic inferences about many different elements to reach political decisions. Therefore, the representativeness heuristic may account for the influence of cues on many different political judgments. This section reviews such research on voter decisions, as well as discusses findings on context and conditioning effects.

In trying to define the role the representativeness heuristic played in voter reasoning, scholars pointed to the various cues from which voters could infer relevant information. For example, Popkin (1994) proposed that voters used their personal assessment of a candidate to assess what kind of leader he or she was in previous offices or what kind of president he or she will be in the future. Popkin also acknowledges that demographic cues and racism are related to the representativeness heuristic. Sniderman et al. (1991) pointed to likability and endorsements. Party and partisan stereotypes were found to be especially powerful (Rahn, 1993). With the shift to studying the effective leveraging of cues, the term representativeness ran out of fashion in political science.

Much research continues to be done on cues as information shortcuts that operate in the spirit of the representativeness heuristic: through the associations they invoke and the inferences based upon those associations. Lau and Redlawsk (2001) grouped them into five major categories: party affiliation (e.g., Arceneaux, 2008; Rahn, 1993); ideology (e.g., Conover & Feldman, 1984, 1989); endorsements (Brady & Sniderman, 1985); opinion polls (Mutz, 1998). and candidate appearance, which includes many social stereotypes like age and race (e.g., Piston, Krupnikov, Milita, & Ryan, 2018); and gender (e.g., Teele, Kalla, & Rosenbluth, 2018). More recently, scholars studied the effect of class as a heuristic cue (e.g., Carnes & Sadin, 2013). Other scholars did continue to use the heuristic concept explicitly in their work. Petersen and colleagues introduced the deservingness heuristic (Jensen & Petersen, 2017; Petersen, 2015; Petersen, Slothuus, Stubager, & Togeby, 2011). This heuristic explains attitudes toward social benefits based on the perception of whether or not the recipient is at fault. However, the perception of that recipient appears to be based on his or her representative exemplar: the stereotypical sick person or unemployed person. Jensen and Petersen examine the case of modern lifestyle diseases. Such diseases are related to personal choices and thus differ from the stereotypical representation of a sick person as being helpless. However, they do not find that such diseases alter the general perception of sick people. This is in line with the expectations based on the representativeness heuristic: Unrepresentative aspects of a phenomenon (here sick people) are neglected in favor of more representative aspects. In this sense, the deservingness heuristic can thus be seen as a special case of the representativeness heuristic.

Context and Conditioning Effects on the Use of the Representativeness Heuristic

The dual process (System 1 vs. System 2) perspective suggests that there are certain conditions that make heuristic processing more or less likely. Such studies can help explain who will be more likely to rely on such cues in what context. Several studies have explored the consequences of such variables in the context of voter decision-making (e.g., Coan, Merolla, Stephenson, & Zechmeister, 2008; Lau & Redlawsk, 2001). These studies explore factors that facilitate/inhibit heuristic processing in System 1 but also study the factors that that facilitate/inhibit more elaborate processing in System 2, as each might contribute to which system will be dominant in which decision.

In line with the original emphasis on biases, early work examined factors that might contribute to System 2 overriding System 1. For example, Slovic and Tversky (1974) propose that a deeper understanding of a problem can contribute to System 2 thinking. When the logic behind each option in the Linda problem is made more transparent, the number of people believing that it is more likely that Linda is a feminist bank teller rather than a bank teller in general decreases markedly (Tversky & Kahneman, 1983). This might suggest that those with a higher cognitive ability might be more likely to recognize this logic in the first place and thus be less likely to commit the conjunction error even without explicit explanation. To test this, scholars compared the IQ (measured in SAT scores) of people selecting the representativeness-based and nonrepresentativeness-based answers in replications of the original tests. The results show IQ only plays a modest role in the sensitivity to using the representativeness heuristic (Kelman, 2011; Stanovich & West, 1998).

Apparently, cognitive ability is not enough. The dual process (System 1 vs. System 2) approach suggests various other factors that might be important. Lau and Redlawsk (2006) build on the literature and their own research to propose four factors that promote heuristic processing and four that promote more detailed processing. They propose that people are more likely to rely on heuristics when they have competing obligations/interest, when they are familiar with the topic and complacent about the consequences, when a task is difficult, and when they have to decide under time pressure. Alternatively, when the decision is perceived as important, the stakes are high, when people feel anxious about potential consequences, or if the phenomenon is novel, people are more likely to process information more elaborately.

Both the degree of competing interest and task difficulty are factors making a decision more or less complex. Much work has examined the effect of decision complexity on heuristic use (e.g., Bodenhausen & Lichtenstein, 1987; Coan et al., 2008; Ottati, 1990). Sniderman (2000) proposes that, in politics, the political institutional structure plays an important role in determining complexity. A two-party system facilitates the long-term formation of two rival blocks, with which voters can become familiar over time and which yield adequate associations about the likely policy preferences of its candidates. In multiparty systems the choice set tends to be much larger. There are more parties to become familiar with, and new party entry tends to occur more frequently, both increasing the complexity facing a voter (Sniderman, 2000). Consequently, voters in multiparty systems would rely less on heuristic reasoning. An experiment by Coan et al. seems to support this hypothesis. They tested the effect of issue complexity on the influence of party labels. They found that party labels had more effect on opinions about complex issues, especially when the voter was familiar with and had trust in that party. Minor party labels, such as the Green Party in the United States, were less effective.

Note here that for most political scientist the main question has not been how to avoid the biases associated with the representativeness heuristic but rather whether voters are able to use the representativeness heuristic effectively to compensate for their lack of detailed knowledge. Their concern is usually not with activating System 2 to override System 1 but with how to facilitate System 1 as best as possible. Sniderman’s (2000) argument about the importance of political institutions described earlier, for example, deals mainly with how institutional features enable voters to draw adequate heuristic inferences. The political system must supply useful categories to order the political world. Other research looked at whether voters were able to process the cues presented to them. Such studies have found that a certain level of political sophistication is important to be able to interpret heuristic cues (Lau & Redlawsk, 2001). Mondak (1993) explains that voters with low political sophistication rely on cues when the decision is of low salience, but, in contrast to those with more political sophistication, they deduce more diverse information from a cue. In other words, the representations triggered by those cues might be less predictive about the evaluation in question. The more sophisticated are better able to interpret cues adequately (see also Lau & Redlawsk, 2001). Druckman, Kuklinski, and Sigelman (2009) explain that heuristics are a processing shortcut, not an informational shortcut: To work well people still need to have adequate information available to them.

Tversky and Kahneman (1983) already noted the importance of whether or not certain information is presented on the likelihood of heuristic processing. When probabilities are stated as observed frequencies, rather than percentages, for example, people find it easier to process this information and appear to rely less on the representativeness heuristic (see also Gigerenzer et al., 1988). If such frequencies are not available, people apparently have difficulty constructing them from probabilities. Mondak (1993) similarly argues that the supply of information is important in the political context and argues that when substantive information presented to voters is minimal, they rely more on heuristic inferences. This makes sense: How can System 2 override System 1 when it lacks the information to do so, but also how can System 1 yield adequate inferences if it is based on unpredictive associations?

People with a higher political sophistication have more background knowledge about politics but are also likely to pay more attention to campaigns. Lau and Redlawsk (2001) find that they are more likely to search for cues, such as party labels, ideology, and polls, when engaging with a campaign. More attention yields more knowledge and thereby more relevant and informative mental categories (see Hafner-Burton, Hughes, & Victor, 2013). In this way sophistication facilitates heuristic processing, since the mental categories from which the people draw their associations (i.e., their mental representations to which they associate the event in question) and to which they judge the similarities to the information provided are likely to be more relevant and informative. However, more attention is also likely be related to less reliance on heuristic processing, in favor of more elaborate System 2 processing. So are these cues really processed heuristically?3 Kam (2005) finds that the more sophisticated rely less on party cues and instead rely more on issue-relevant information, while those less politically sophisticated rely more on the party cue. It appears that sophistication is needed to interpret political cues like party but also leads to more systematic processing. As the emphasis in political science has been on finding out how voters make political inferences based on little information, rather than studying the potential biases resulting from heuristic processing, this question has not been very central. Maybe it is also the wrong question to ask, since the growing consensus is that there is no strict boundary between System 1 and System 2 but rather they rather work together while emphasizing one or the other to different extents in different situations (Kahneman & Frederick, 2002).4

Political Elites and the Representativeness Heuristic

The focus on specific cues and more specific heuristics among political scientists is understandable in the context of explaining voter decision-making based on little information. However, the likely influence of the representativeness heuristic on political decision-making more generally is not restricted to such applications. Recently, the representativeness heuristic has been rediscovered in political decision-making research and applied to other decision domains, like decision-making by political elites.

Whether or not the political elite’s decisions are sensitive to the representativeness heuristic is open for question. Elites have more resources and more support staff; they are also held accountable for their decisions by their colleagues and their voters. Elites, thus, do not suffer from the shortage of information like voters do but rather struggle to deal with the overload of information they receive (Vis, 2018). The high stakes, their high motivation, and the fact that their decisions tend to be important all indicate that they should be more likely to rely on more elaborate System 2 processing (cf. Lau & Redlawsk, 2006). However, in their decisions they often face complex issues, competing interests, and time pressure, which all could point to System 1 heuristic processing instead (e.g., Lau & Redlawsk, 2006). Their decisions are not necessarily structured by a neat two-party institutional system; instead they face a myriad of potential solutions to both enduring and novel problems. They might be more politically sophisticated than the average voter but often are not policy experts on all areas they have to decide upon. Hafner-Burton et al. (2013) reviewed the literature on the effect of heuristic use by elites and found that expertise helps elites make more effective use of heuristics, similar to the examples of voters described earlier. However, the advantage of expertise is lost if decisions refer to other domains, even if they are very close to the domain of expertise. There is thus good reason to believe that the representativeness heuristic might also influence elite political decision-making (see Vis, 2018).

Like voters, political elites also rely on cues. Studies found that political elites infer race from names of constituents and use this as a shortcut for the likelihood that this person will vote for him or her, and consequently in determining whether or not to respond to a request (e.g., Butler & Broockman, 2011). Politicians may also inaccurately infer the opinion distribution in their constituency from the constituents who contact them the most (assuming these constituents are representative for all their constituents) (Broockman & Skovron, 2018).

McDermott (2001) argues that the representativeness heuristic applies to many other politically relevant phenomena. The causal conjunction error makes certain potential scenarios for the future appear much more likely than others, only because it is easier to think of reasons about why they should occur. Such inferences might contribute to an overemphasis in policy formation on preventing the negative aspects of such scenarios and a neglect of addressing the risks of less easily imaginable, but more likely, scenarios (cf. Tversky & Kahneman, 1983). Remember the example, discussed in the section “What Is Representativeness?”, of people believing it to be more likely that Russia and the United States would suspend diplomatic relations due to a Russian invasion of Poland in 1983 than that Russia and the United States would suspend diplomatic relations without a specific cause being provided.

Similarly, historical analogies, like those of Vietnam or Munich in foreign policy, provide easy accessible representations of likely consequences of certain policy choices but might not always be very representative of the situation at hand (McDermott, 2001). People overestimate the likelihood of things that tell a good story, and policymakers are people too. Weyland (2007) shows how successful pension reform in Chile was believed to be representative of the likely consequences of similar reforms in other Latin American countries. Politicians argued: “we can do this too!” The use of the representativeness heuristic led these politicians to downplay the differences in the political situation. This contributed to the failure of the subsequent reforms in these countries.

A second effect mentioned by McDermott (2001) is related to policy evaluation. Kahneman and Tversky (1973) find that people fail to recognize the tendency for stochastic processes to regress to the mean: An extreme instance is usually followed by one closer to the mean. Israeli flight instructors found that when they praised a pilot after successfully executing a complex maneuver, this typically resulted in decreased performance in the next try. Therefore they believed that punishment must work better than praise, as punishment after bad performance typically resulted in better performance in the next try. This dynamic of very good performance followed by less good performance and very bad performance followed by less bad performance can be explained by regression to the mean alone, rather than by the effect of punishment. Kahneman and Tversky explain that people tend to view the predicted outcome as maximally representative of the input: They fail to acknowledge that they deal with outliers but rather see past performance as representative of future performance.

The biased assessment of the predictive value of past performance may impact political decision-making in two ways. First, when defining success of a policy, a reference point is needed; usually this reference point is provided by the situation in the year before or under the previous government. If a policy of the previous government yielded particularly bad results, which prompted the new government to alter this policy, then politicians might infer that the new policy, which produces better results, is much more effective than it actually is. Better performance might just reflect regression to the mean. This appearance of success might contribute to reinforcing ineffective policy measures and potentially wasting public funds in the process. Second, regression to the mean might lead to many short-time changes in performance indicators, like gross domestic product and inflation, but also the number of robberies in a year, and so on. A failure to acknowledge the regressive process might prompt people to come up with all kinds of ad hoc explanations for decreases or increases in these indicators, which have no basis in reality but might influence future policymaking.

A Research Agenda

So far, this article has discussed the nature of the representativeness heuristic and reviewed the literature on its implications for voter decision-making, as well as decision-making by political elites. There might be many domains other than voter and political elite decision-making in which the representativeness heuristic might play a role, for example, agenda-setting (some issues might appear more legitimate/important/urgent than others due to the mental representations they are associated with), persuasion, opinion formation, or decision-making by other kinds of political actors, like social movements. Each of these provide an interesting opportunity for further research. However, for the purpose of this article, the recommendations for further research outlined in this section focuses on reintroducing the concept of the representativeness heuristic as a key concept in political decision-making, how to proceed in working out its consequences, on using this heuristic to integrate findings on the effects of separate cues and on the democratic consequences of the systematic biases that accompany the use of the representativeness heuristic in political decision-making.

Because of the complexity inherent to political decisions, it is usually impossible to fully comprehend all aspects relating to all possible alternatives available. Every political actor (e.g., voter, activist, or political elite) is forced to simplify in one way or another. As explained in the section “What Is Representativeness?,” using the representativeness heuristic helps to make decisions more manageable by extrapolating on the information presented (in cues or otherwise) through matching it to one’s pre-existing mental categories based on similarity or association. In doing so, a lot of information is inferred/added to the event in question, without much effort, which can be used to make probability judgments. Which information is inferred is likely to depend on one’s available mental categories with which cues can be associated, so individual differences probably play a role. On the other hand, the representativeness heuristic applies to widespread and fundamental political questions, like assessments of what political issues need addressing (i.e., likely to be of great consequence) and how to address them (i.e., which policy is likely to be successful). Since this heuristic presents a common mechanism for systematic biases to occur across contexts and specific cues, and since it applies to the process of decision-making by many actors, it might pay off to study its effect coherently and map its boundary conditions more precisely. For example, considering the effects of cues as instances of the representativeness heuristic, rather than as separate entities, highlights the potential biases involved in the use of all cues, not just those relating to social stereotypes.

This section proposes an agenda for further research emphasizing three specific areas: mapping individual and contextual differences, integrating findings on cues, and evaluating the consequences of systematic biases. By further investigating individual and contextual differences, researchers can address the “vague-ness critique” described in the section “What Is Representativeness?” and yield more concrete predictions. Integrating the findings of the various studies on different cues can help in this regard as well. A stronger focus on the systematic biases that accompany the use of the representativeness heuristic might open up new areas of interest. Together these research proposals aim to contribute toward a more encompassing, coherent, and systematic description of the role of the representativeness heuristic on political decision-making.

Individual and Contextual Differences

Several studies discussed in this article deal with individual and contextual differences but mainly relate them to the effect of specific cues (e.g., Lau & Redlawsk, 2001; Sniderman, 2000). Over time, the literature on voter decision-making has put more emphasis on studying particular cues and their effects, rather than examining the effect of the representativeness heuristic in general. An advantage of this approach is that it bypasses part of the vague-ness critique of the representativeness heuristic. These studies define a priori which representation (i.e., related to gender, class, party) they are studying, so they do not have to answer the question of which domain of representations (e.g., class or party or physical appearance of competence) is more applicable in different situations. Instead, they focus on the whether a cue elicits a specific representation and the effect of that cue for different people and in different contexts. In this way, they address the important question of how voters use information in their political decision-making. However, the study of cues might divert attention from questions related to the representativeness heuristic more generally, such as how representations are formed, how they differ across individuals, and how they might be applied differently in different contexts. The processes generating such individual differences might be related to the representative heuristic in general, rather than be tied to a specific cue. Also starting from the perspective of the representativeness heuristic yields hypotheses about how these differences come about. This section reviews findings of studies on specific cues that contribute to establishing individual and contextual differences in the use of the representativeness heuristic, as well as outlines the contribution studies on the representativeness heuristic more generally can make to account for the boundary conditions of the effects of specific cues in turn.

As described earlier, Sniderman (2000) already wondered about the role of political institutions in the formation of the mental categories that form the basis for the representativeness heuristic to work. Likewise, research on the importance of political sophistication for the efficient use of the representativeness heuristic shows that it matters which mental categories are available for a person to form associations with (Bellur & Sundar, 2014). The more sophisticated have more accurate representations of the issue positions related to different ideologies, and, compared to those less sophisticated, they are better able to match candidates with their own preferences based on limited information about the candidate in question (Lau & Redlawsk, 2001). In other words, the mental categories the more sophisticated learned over time better explain reality, so when a cue invokes such a mental category it adds more accurate information to that cue, making the cue more informative, and using the representativeness heuristic with this cue as input becomes more efficient.

However, focusing on individual differences as restricted to the accuracy of one’s available representations might lead the “heuristic” aspect of using cues to be overlooked. The bias side of the representativeness heuristic still applies, since associating a cue with a mental representation is a judgment based on similarity/association rather than deductive inference. Having a more accurate representation of feminists in the Linda problem (discussed in the section “What Is Representativeness?”) is not likely to help much in avoiding the conjunction fallacy. Individual differences in the effect of using the representativeness heuristic are thus not necessarily equal to individual differences in avoiding biases. Systematic biases are a consequence of using the representativeness heuristic, but whether using the representativeness heuristic leads to biased decision-making also depends on whether the invoked representation suggests a different decision from using deductive inference. In some situations, such as the Linda problem, deciding based on Linda’s feminist association is at odds with the deductive logic that there are more bank tellers than feminist bank tellers, but in many other situations these two different systems of decision-making (see “What Is Representativeness” section) do not lead to different conclusions. As Kahneman and Frederick (2002) note, using the representativeness heuristic in most cases works quite well. What Lau and Redlawsk’s (2001) findings rather suggest is that there might also be many situations in which using the representativeness heuristic leads to (more) biased conclusions if one’s mental categories do not (or less so) match reality. This could be either due to less well developed mental categories (as in the case of less sophisticates), which makes the use of the representativeness heuristic less efficient, or because the cue invoked is not representative for the event in question (as in the case of a candidate taking a counter-stereotypical position).

So what explains these individual differences? Gualtieri and Denison (2018) studied the sensitivity of young children to the representativeness heuristic. They found that 4-year-olds used the heuristic, but its influence increases over time as they acquire more knowledge about social categories and stereotypes. This suggests that social learning plays an important part in the formation of mental categories. This makes it likely that the exact representations relevant to political life differ between individuals and across decision domains depending on the knowledge and experience acquired over the life course.

Little research has been done to map those individual differences in mental categories. Nielsen and Larsen (2014) provide an example of how political scientists might proceed in this regard. They first measured the associations with party labels on an individual basis and then tested the relation between these representations and voting behavior; they found that their concept of “party brands” had additional explanatory value over party identification as such. Future studies can extend this perspective to more detailed analysis of the representations surrounding many other political cues. As a next step, researchers might analyze how these different representations are formed over time. It might be worthwhile, in this respect, to transcend the focus on individual cues. It is likely, for example, that some individuals have a larger set of more detailed representations than others, which might depend on the domain in question.

A small number of studies have proposed a number of variables that might help explain how mental categories differ across people. For example, Hafner-Burton et al. (2013) suggest that experience is important. They found that the representations influencing the policy of the Bush administration toward North Korea shifted from “how parents deal with children throwing tantrums” in 2002 to “a game of strategic containment” in 2006 (Hafner-Burton et al., 2013, p. 378). This study illustrates how similar events, and similar cues, might invoke different representations among the same people over time. Similarly, Baldassarri and Schadee (2006) showed that different voters, who hold different conceptions of politics, consider different sets of representations as more/less relevant in processing campaign information. Druckman et al. (2009) highlighted the role of availability in the media. The media might make certain issue domains and their related representation more salient and consequently increase the probability that voters will draw on representations related to that domain in evaluating political events and actors (Iyengar & Kinder, 1987).5

Other studies have taken the reverse perspective and investigated how people adapt their representations to the situation. For example, Arceneaux (2008) finds that whether voters punish a candidate for taking counter-stereotypical issue positions is contingent on the salience of the issue and the political awareness of the voter. Fortunato and Stevenson (2019) build on this finding and examine the extent to which constituents of loyal partisan senators differ in their reliance on party cues, compared to constituents of less loyal partisan senators. Their results confirm that these voters indeed differ in their reliance on party cues, which could point to a learning effect in whether or not a certain representation is applicable. Such a learning effect might apply to other representations/cues as well.

These findings on the differences in politically relevant mental categories held by different people, considered in different situations, and how these representations are shaped over time and across contexts present ample opportunities to further extend knowledge about the effect of the representativeness heuristic in political decision-making. Individual differences in mental categories might be cue related or they might be general reflections of domain interest (i.e., related to political sophistication), but they might also be subject to more intricate variations, like the study on party brands by Nielsen and Larsen (2014) illustrates. Systematically mapping those representations can be a first step toward uncovering the influence of individual differences in mental categories.

Integrating Findings on Cues

In addition to differences in the effect of the representativeness heuristic due to individual and contextual variations, there is also much to gain in further delving into commonalities across different applications of this heuristic. To assess what studies on specific cues help explain the effects and boundary conditions of the representativeness heuristic, they need to be somehow integrated into a framework that shows how they fit together. The neglect of the common heuristic origin of the effects of these cues may lead scholars to assume this is a rather straightforward task: The separate effects of different cues can be simply added to each other as independent effects to determine how they work together to form the informational basis of political decision-making: One cue accounts for 3%, the other for 7%, a third for 4% of an attitude, and so on. This section outlines how acknowledging the heuristic origin of the effects of these cues leads to a different expectation: When different cues are presented in combination, certain cues are likely to dominate other cues in forming political judgments depending on the context. This is because of the fundamental premise behind heuristic processing that elements are not included additively. Reasoning is based on association, not on deduction. In the case of the representativeness heuristic, the representative associations dominate other cues. This means that the result of one study cannot simply be added to that of another, since the effect of a cue in the context of one study can be very different from its effect in another context. This makes it difficult to add results of studies of different cues together. For example, when two cues are present, how do researchers know whether each cue will be considered side by side and one will not trump the other? It is argued in this section that the heuristic approach can aid the study of the effect of cues by providing a research agenda for how to proceed in integrating existing findings.

An example of one cue overshadowing another is the lawyer/engineer problem discussed in the section “What Is Representativeness” (Kahneman & Tversky, 1973). The results of that problem showed that base rates tend to be neglected in the presence of a stereotypical personality description. When no personality description was given, participants did rely on the base rates. Participants thus realized the importance of base rate information but still neglected it in the face of the stereotypic personality description. A study by Riggle, Ottaki, Wyer, Kuklinski, and Schwarz (1992) yields a similar result more tailored to the political context. They find that candidates’ physical attractiveness had substantial influence on the evaluations of those candidates. However, when information about the voting record or the party membership of the candidate was made available, physical attractiveness lost its influence. To make matters more complex, they found that when evaluating one candidate, the voting record was the dominant cue for forming evaluations, but when participants had to compare two candidates, they relied more on party membership.6 This illustrates how subtle differences in context or the exact combination of cues present might greatly affect the influence of each specific cue.

Tversky and Kahneman (1983) provide an example of how a cue that may be dominant in a certain study situated in certain context may not be dominant in another study in another context. They found in one of their studies that participants believed it was more representative for a Hollywood actress to be divorced more than four times than to vote Democratic (Tversky & Kahneman, 1983). So when the participants in this experiment would rely on the representativeness heuristic to asses Hollywood actresses, they would be more likely to draw on their associations about people who divorce many times than associations about the Democratic party. So in this context the party cue is not the dominant one. In addition, Tversky and Kahneman also found that a parallel group of participants believed that here are more Hollywood actresses that vote Democratic than who are divorced more than four times. The stereotype of the Democratic Party thus seems to be relevant for a larger number of actresses than the stereotype related to multiple divorces. Tversky and Kahneman explain that the dominant stereotype about a group tends to be determined by the aspects that set this group apart from other groups, rather than by the frequency of attributes within the group: participants appear to believe that Hollywood actresses differ from other people more in their number of divorces than in their support for the Democratic Party. The results also show that a representation that might be important in the domain of politics, like party identification, need not be the most representative in all circumstances.

Alternatively, there are also some studies that indicate that different cues might have effects alongside each other. Burns, Eberhardt, and Merolla (2013) found that a description that mixes feminine and masculine attributes leads to a higher overall evaluation for U.S. vice presidential candidate Sarah Palin than descriptions of either feminine or masculine attributes. Apparently, pure gender stereotypes are not the dominant consideration in evaluating Palin. These results could mean that people have more fine-grained gender representations or that different representations might combine to produce an overall judgment in certain cases. This would be in line with a finding by Kahneman, Schkade, and Sunstein (1998) about punitive damages that should be awarded to victims who suffered bodily harm as a result of using a product. The damages indicated by the participants appeared to be correlated not just by the degree of outrage felt by participants in view of the incidents (which they presumed would be the heuristic cue used) but rather by the product of the outrage felt and the amount of harm suffered by the victim. Apparently, the representations of improper conduct of the company behind the product (outrage) and that of the harm suffered by the victim worked in tandem to produce the dollar estimate of the punitive damages to be paid (see also Kahneman & Frederick, 2002). These studies suggest that in some conditions different representations might work in parallel; further research can investigate how this works precisely and under which conditions this occurs.

The Consequences of Systematic Biases

In addition to helping explain why adding the results of studies on specific cues together might be problematic, a stronger focus on the heuristic origin of cues also highlights the systematic biases involved in the use of all cues. The trend toward studying the effects of cues has helped to make the biases related to the representativeness heuristic a little more central to political science than in the early days. As argued earlier, the studies that introduced the representativeness heuristic in political science mainly dealt with its ability to help voters make judgments based on little information. Consequently, they paid little attention to the biases involved (Kuklinski & Quirk, 2000). Studies on the effects of social stereotypes do put biases at center stage (e.g., Teele et al., 2018). Still, Kahneman and Tversky (1973; see also Tversky & Kahneman, 1974, 1983) described several systematic biases related to the representativeness heuristic that are likely to be relevant regardless of the specific cue involved. Several such biases have been recognized by political scientists, but they have yet to be systematically investigated. For example, Lau and Redlawsk (2001) found that more politically sophisticated voters made better use of heuristic cues, like ideology, than less sophisticated voters. However, they were also led astray by this cue more often than less sophisticated voters, when a candidate took counter-stereotypical issue positions. So far the conjunction fallacy and base rate neglect have been mentioned as systematic biases related to the representativeness heuristic. Some other biases are likely to be relevant for political judgment as well. Kahneman and Tversky mention insensitivity to predictability and the illusion of validity, and Kahneman and Frederick (2002) point to scope neglect and duration neglect. Each is discussed in turn, followed by an examination of how scholars can help efforts to address the adverse effects of these biases.

Insensitivity to predictability refers to the neglect of the degree to which a cue is actually predictive for the assessment in question, in deciding whether to base one’s assessment on this cue and to what extent. For example, in one experiment participants were given assessments of a practice class of a student teacher. One group of participants was asked to rank the quality of the class delivered based on this description; a different group was asked to predict the performance of that student teacher five years after the practice class. Although it appears unlikely that such a practice class has a strong predictive power over performance over such a long time span, the answers of the two different groups of participants correlated strongly (Kahneman & Tversky, 1973). This means that the participants making the prediction neglected to weigh in the low predictability of the cue (performance in the practice class) in their prediction of performance five years later. They neglected the likely regression to the mean described earlier in the section on political elites (see also Jarvstad & Hahn, 2011). The effect of this neglect on policy/candidate evaluation by voters and political elites alike may be further investigated. One example of such an application is Kuklinski and Quirk (2001), in which the authors asked participants to assess the percentage of all families that are on welfare. They found that the percentage differed markedly between those favoring and those opposing welfare spending. Both groups were significantly off the mark though and gave answers that varied considerably. Still, most participants indicated they were very or fairly confident in their estimates (regardless of whether they opposed or favored welfare spending). Since the estimates of most people were rather off the mark, it would have been prudent for many individuals to be very cautious about their estimates. That they were not illustrates that people are often unaware of the lack of reliability of their estimates and of the biases that influence their judgment. This might lead to overconfidence in many political decisions (cf. Sheffer & Loewen, 2019). Further studies can investigate the effect that insensitivity to predictability has on all kinds of political decisions, for elites as well as voters.

The second bias discussed here is the illusion of validity. This bias is related to the insensitivity to predictability bias in that it deals with how people weigh the evidence for their estimates. The illusion of validity entails that people have more confidence in their prediction when they receive multiple cues that all fit the representation with which they assess it. The more consistent the evidence, the more representative it will appear, and the more confident people feel in their assessment (Tversky & Kahneman, 1974). Although this appears to be a valid inference, Tversky and Kahneman illustrate why illusion of validity might be an illusion. They find that when people have to judge whether a personality description belongs to a librarian, they feel more confident in their assessment when the description contains several cues that fit the stereotype of a librarian, no matter how vague, unreliable, or outdated these cues are. In addition, they argue that when people encounter such uniform cues in real life, this is likely the result of a correlation between these cues. For example, the person writing the personality descriptions might have been looking for certain cues. Such cues then provide more information about the bias of the writer than about the personality of the person in question: Correlated cues do not provide independent sources of information, so their additional informational value is reduced. Cues that are independent from each other provide more information than cues that are correlated. The tricky part is that independent cues are also more likely to differ from each other compared to correlated cues. The result is the counterintuitive situation in which cues that differ more provide a better basis for prediction, but cues that are more similar feel like stronger evidence. The failure to take the correlation of cues into account produces the illusion of validity. Scholars might explore whether this illusion applies to political decision-making by examining to what degree different cues used by voters or political elites are correlated, whether voters and political elites indeed neglect this correlation in their assessments, and what effect this has on political decision-making (e.g., see Ortoleva & Snowberg, 2015).

The third and fourth biases are scope neglect and duration neglect. These biases are related to the base rate neglect bias discussed before. Both refer to the neglect of an extensional aspect of an attribute in the face of a stereotypical exemplar. For example, when politicians debate about how to deal with terrorism due to a single small-scale incident in their country in over a decade, it is easy to neglect the long period without attacks and instead focus on their attitudes toward the current vivid representative instance. According to scope neglect, politicians (and voters) will base their evaluation of terrorism on their feelings toward the stereotypical exemplar of a terrorist attack and neglect to adjust for the scope of the current threat. To support their argument, Kahneman and Frederick (2002) point to a study that shows that Toronto residents were willing to pay as much to clean up a few lakes in Ontario as they were to clean up all lakes in Ontario (Desvousges, Mathews, & Train, 2012; Kahneman & Knetsch, 1992): They neglected the difference in scope and instead based their willingness to pay on their feelings evoked by their mental representation of polluted lakes in general. Duration neglect is rather similar but refers to duration rather than scope. Such biases are likely to be relevant for political decision-making, since politicians need to divide the budget across many areas, some of which have a larger scope than others. Based on this research surrounding the representativeness heuristic, it can be expected that areas with vivid stereotypical examplars, such as terrorism, receive more budget than problems that are arguably larger in scope or more enduring but lack such examplars. Note that differences in spending could be the result of representativeness-based reasoning on the part of voters, as well as policymakers.

The research agenda outlined here thus proposes to map differences in the effect of using the representativeness heuristic for different contexts and individuals, as well as explore the commonalities between those situations in the variables that determine those variations and pay increased attention to the biases that accompany the use of the representativeness heuristic. If it is indeed the case, as argued in this section, that systematic biases negatively affect political decision-making, then a relevant question is what to do about this. Several options are available. As discussed before, Kahneman and Frederick (2002) proposed that to reduce the influence of the representativeness heuristic and its accompanying biases, System 2 should be activated to override System 1. In addition, other research suggests two further ways to deal with biases: socialization and training. The influence of socialization is illustrated, for example, by the study discussed earlier that showed that young children increase their reliance on stereotypes as they learn more about them (Gualtieri & Denison, 2018). Other studies looked at the effect of retraining. For example, one study showed beneficial effects of giving people insight in biases by letting them experience their own mistakes and subtely confronting them with the consequences of their logic (Cox & Mouw, 1992). Another approach might be to retrain the associations people have with stereotypes (see Forbes & Schmader, 2010; Leicht, Randsley de Moura, & Crisp, 2014). It must be noted that altering associations is no easy task. For example, Herrmann and Tepe (2017) investigated the effect of salient counter-stereotypical examples in a recent German election. Such examples might contribute to altering the associations with that stereotype. They found, however, that these counterexamples had little effect. Further studies could investigate how System 2 can be stimulated in the context of political decision-making and whether, when, and how socialization and retraining might be beneficial in improving political judgment.


Arceneaux, K. (2008). Can partisan cues diminish democratic accountability? Political Behavior, 30(2), 139–160.Find this resource:

Baldassarri, D., & Schadee, H. (2006). Voter heuristics and political cognition in Italy: An empirical typology. Electoral Studies, 25(3), 448–466.Find this resource:

Bellur, S., & Sundar, S. S. (2014). How can we tell when a heuristic has been used? Design and analysis strategies for capturing the operation of heuristics. Communication Methods and Measures, 8(2), 116–137.Find this resource:

Bodenhausen, G. V., & Lichtenstein, M. (1987). Social stereotypes and information-processing strategies: The impact of task complexity. Journal of Personality and Social Psychology, 52(5), 871–880.Find this resource:

Brady, H. E., & Sniderman, P. M. (1985). Attitude attribution: A group basis for political reasoning. American Political Science Review, 79(4), 1061–1078.Find this resource:

Broockman, D. E., & Skovron, C. (2018). Bias in perceptions of public opinion among political elites. American Political Science Review, 112(3), 542–563.Find this resource:

Burns, S., Eberhardt, L., & Merolla, J. L. (2013). What is the difference between a hockey mom and a pit bull? Presentations of Palin and gender stereotypes in the 2008 presidential election. Political Research Quarterly, 66(3), 687–701.Find this resource:

Butler, D. M., & Broockman, D. E. (2011). Do politicians racially discriminate against constituents? A field experiment on state legislators. American Journal of Political Science, 55(3), 463–477.Find this resource:

Carnes, N., & Sadin, M. L. (2013). The “mill worker’s son” heuristic: How voters perceive politicians from working-class families—and how they really behave in office. The Journal of Politics, 77(1), 285–298.Find this resource:

Coan, T. G., Merolla, J. L., Stephenson, L. B., & Zechmeister, E. J. (2008). It’s not easy being green: Minor party labels as heuristic aids. Political Psychology, 29(3), 389–405.Find this resource:

Conover, P. J., & Feldman, S. (1984). How people organize the political world: A schematic model. American Journal of Political Science, 28(1), 95–126.Find this resource:

Conover, P. J., & Feldman, S. (1989). Candidate perception in an ambiguous world: Campaigns, cues, and inference processes. American Journal of Political Science, 33(4), 912–940.Find this resource:

Cox, C., & Mouw, J. T. (1992). Disruption of the representativeness heuristic: Can we be perturbed into using correct probabilistic reasoning? Educational Studies in Mathematics, 23, 163–178.Find this resource:

Desvousges, W., Mathews, K., & Train, K. (2012). Adequate responsiveness to scope in contingent valuation. Ecological Economics, 84, 121–128.Find this resource:

Druckman, J. N., Kuklinski, J. H., & Sigelman, L. (2009). The unmet potential of interdisciplinary research: Political psychological approaches to voting and public opinion. Political Behavior, 31(4), 485–510.Find this resource:

Forbes, C. E., & Schmader, T. (2010). Retraining attitudes and stereotypes to affect motivation and cognitive capacity under stereotype threat. Journal of Personality and Social Psychology, 99(5), 740–754.Find this resource:

Fortunato, D., & Stevenson, R. T. (2019). Heuristics in context. Political Science Research and Methods, 7(2), 311–330.Find this resource:

Gigerenzer, G. (1996). On narrow norms and vague heuristics: A reply to Kahneman and Tversky. Psychological Review, 103(3), 592–596.Find this resource:

Gigerenzer, G., Hell, W., & Blank, H. (1988). Presentation and content: The use of base rates as a continuous variable. Journal of Experimental Psychology, 14(3), 513–525.Find this resource:

Gigerenzer, G., & Selten, R. (Eds.). (2001). Bounded rationality: The adaptive toolbox. Cambridge, MA: MIT Press.Find this resource:

Gilovich, T., & Griffin, D. (2002). Introduction—heuristics and biases: Then and now. In T. Gilovich, D. Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment (pp. 1–18). New York, NY: Cambridge University Press.Find this resource:

Gilovich, T., Griffin, D., & Kahneman, D. (Eds.). (2002). Heuristics and biases: The psychology of intuitive judgment. New York, NY: Cambridge University Press.Find this resource:

Gualtieri, S., & Denison, S. (2018). The development of the representativeness heuristic in young children. Journal of Experimental Child Psychology, 174, 60–76.Find this resource:

Hafner-Burton, E. M., Hughes, D. A., & Victor, D. G. (2013). The cognitive revolution and the political psychology of elite decision making. Perspectives on Politics, 11(2), 368–386.Find this resource:

Herrmann, M., & Tepe, M. (2017). Does exposure to stereotype-disconfirming politicians reduce the effect of stereotypes on voting? Evidence from seven plagiarism scandals in Germany. Political Psychology, 39(2), 303–324.Find this resource:

Iyengar, S., & Kinder, D. R. (1987). News that matters: Television and American opinion. Chicago, IL: University of Chicago Press.Find this resource:

Jarvstad, A., & Hahn, U. (2011). Source reliability and the conjunction fallacy. Cognitive Science, 35(4), 682–711.Find this resource:

Jensen, C., & Petersen, M. B. (2017). The deservingness heuristic and the politics of health care. American Journal of Political Science, 61(1), 68–83.Find this resource:

Kahneman, D. (2011). Thinking, fast and slow. London, U.K.: Penguin.Find this resource:

Kahneman, D., & Frederick, S. (2002). Representativeness revisited: Attribute substitution in intuitive judgment. In T. Gilovich, D. Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment (pp. 49–81). New York, NY: Cambridge University Press.Find this resource:

Kahneman, D., & Knetsch, J. L. (1992). Valuing public goods: The purchase of moral satisfaction. Journal of Environmental Economics and Management, 22(1), 57–70.Find this resource:

Kahneman, D., Schkade, D., & Sunstein, C. R. (1998). Shared outrage and erratic awards: The psychology of punitive damages. Journal of Risk and Uncertainty, 16(1), 49–86.Find this resource:

Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological Review, 80(4), 237–251.Find this resource:

Kam, C. D. (2005). Who toes the party line? Cues, values, and individual differences. Political Behavior, 27(2), 163–182.Find this resource:

Kelman, M. (2011). The heuristics debate. Oxford, U.K.: Oxford University Press.Find this resource:

Kirkland, P. A., & Coppock, A. (2018). Candidate choice without party labels: New insights from conjoint survey experiments. Political Behavior, 40(3), 1–21.Find this resource:

Kuklinski, J. H., & Quirk, P. J. (2000). Reconsidering the rational public: Cognition, heuristics, and mass opinion. In A. Lupia, M. D. McCubbins, & S. L. Popkin (Eds.), Elements of reason: Cognition, choice, and the bounds of rationality (pp. 153–182). New York, NY: Cambridge University Press.Find this resource:

Kuklinski, J. H., & Quirk, P. J. (2001). Conceptual foundations of citizen competence. Political Behavior, 23(3), 285–311.Find this resource:

Lau, R. R., & Redlawsk, D. P. (2001). Advantages and disadvantages of cognitive heuristics in political decision making. American Journal of Political Science, 45(4), 951–971.Find this resource:

Lau, R. R., & Redlawsk, D. P. (2006). How voters decide: Information processing in election campaigns. Cambridge, U.K.: Cambridge University Press.Find this resource:

Leicht, C., Randsley de Moura, G., & Crisp, R. J. (2014). Contesting gender stereotypes stimulates generalized fairness in the selection of leaders. Leadership Quarterly, 25(5), 1025–1039.Find this resource:

Lupia, A. (1994). Shortcuts versus encyclopedias: Information and voting behavior in California insurance reform elections. American Political Science Review, 88(1), 63–76.Find this resource:

Luskin, R. C. (1987). Measuring political sophistication. American Journal of Political Science, 31(4), 856–899.Find this resource:

McDermott, R. (2001). The psychological ideas of Amos Tversky and their relevance for political science. Journal of Theoretical Politics, 13(1), 5–33.Find this resource:

Mondak, J. J. (1993). Public opinion and heuristic processing of source cues. Political Behavior, 15(2), 167–192.Find this resource:

Mutz, D. C. (1998). Impersonal influence: How perceptions of mass collectives affect political attitudes. Cambridge, U.K.: Cambridge University Press.Find this resource:

Nicholson, S. P., & Hansford, T. G. (2014). Partisans in robes: Party cues and public acceptance of supreme court decisions. American Journal of Political Science, 58(3), 620–636.Find this resource:

Nielsen, S. W., & Larsen, M. V. (2014). Party brands and voting. Electoral Studies, 33, 153–165.Find this resource:

Ortoleva, P., & Snowberg, E. (2015). Overconfidence in political behavior. American Economic Review, 105(2), 504–535.Find this resource:

Ottati, V. C. (1990). Determinants of political judgments: The joint influence of normative and heuristic rules of inference. Political Behavior, 12(2), 159–179.Find this resource:

Petersen, M. B. (2015). Evolutionary political psychology: On the origin and structure of heuristics and biases in politics. Political Psychology, 36(Suppl. 1), 45–78.Find this resource:

Petersen, M. B., Slothuus, R., Stubager, R., & Togeby, L. (2011). Deservingness versus values in public opinion on welfare: The automaticity of the deservingness heuristic. European Journal of Political Research, 50(1), 24–52.Find this resource:

Piston, S., Krupnikov, Y., Milita, K., & Ryan, J. B. (2018). Clear as Black and White: The effects of ambiguous rhetoric depend on candidate race. The Journal of Politics, 80(2), 662–674.Find this resource:

Popkin, S. L. (1994). The reasoning voter: Communication and persuasion in presidential campaigns. Chicago, IL: University of Chicago Press.Find this resource:

Rahn, W. M. (1993). The role of partisan stereotypes in information processing about political candidates. American Journal of Political Science, 37(2), 472–496.Find this resource:

Riggle, E. D., Ottaki, V. C., Wyer, R. S., Kuklinski, J., & Schwarz, N. (1992). Bases of political judgements: The role of stereotypic and non-stereotypic information. Political Behavior, 14(17), 67–87.Find this resource:

Sheffer, L., & Loewen, P. (2019). Electoral confidence, overconfidence, and risky behavior: Evidence from a study with elected politicians. Political Behavior, 41(1), 31–51.Find this resource:

Slovic, P., & Tversky, A. (1974). Who accepts Savage’s axiom. Behavioral Science, 19(6), 368–373.Find this resource:

Sniderman, P. M. (2000). Taking sides: A fixed choice theory of political reasoning. In A. Lupia, M. D. McCubbins, & S. L. Popkin (Eds.), Elements of reason: Cognition, choice, and the bounds of rationality (pp. 67–84). Cambridge, U.K.: Cambridge University Press.Find this resource:

Sniderman, P. M., Brody, R. A., & Tetlock, P. E. (1991). Reasoning and choice: Explorations in political psychology. New York, NY: Cambridge University Press.Find this resource:

Stanovich, K. E., & West, R. F. (1998). Individual differences in framing and conjunction effects. Thinking and Reasoning, 4(4), 289–317.Find this resource:

Teele, D. L., Kalla, J., & Rosenbluth, F. (2018). The ties that double bind: Social roles and women’s underrepresentation in politics. American Political Science Review, 112(3), 525–541.Find this resource:

Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124–1131.Find this resource:

Tversky, A., & Kahneman, D. (1983). Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment. Psychological Review, 90(4), 293–315.Find this resource:

Vis, B. (2018). Heuristics and political elites’ judgment and decision-making. Political Studies Review. [Advance online publication]Find this resource:

Weyland, K. (2007). Bounded rationality and policy diffusion: Social sector reform in Latin America. Princeton, NJ; Oxford, U.K.: Princeton University Press.Find this resource:


(1.) The terms “representations” and “mental categories” are used interchangeably throughout this article. In addition “stereotypes” are considered to form a subclass of “representations,” which refer to social categories.

(2.) For a discussion of other biases see Kahneman and Tversky (1973), Tversky and Kahneman (1974), and Gilovich, Griffin, and Kahneman (2002).

(3.) Consider also Druckman et al. (2009) who argue that if the use of an heuristic is only defined by the criterion that a decision is not based on complete information, than anything can be a heuristic, since it is always possible to find more information.

(4.) Some would argue that the applications of heuristics in political science fit better with the fast and frugal heuristic approach associated mainly with Gerd Gigerenzer (e.g., Gigerenzer & Selten, 2001), rather than the heuristics and biases approach of Tversky and Kahneman (for an explanation of how these approaches differ and overlap see Kelman, 2011).

(5.) This is an example of the close link that may exist between the availability heuristic and the representativeness heuristic; see the section “What Is Representativeness?”

(6.) For a recent study showing how a party cue might dominate a Supreme Court cue, see Nicholson and Hansford (2014). For a study on how other cues carry more weight in the absence of party labels, see Kirkland and Coppock (2018).