Sage Journals: Discover world-class research

Abstract

Direct estimates based on election returns show that corruption is mildly punished at the polls. A large majority of survey respondents, however, often tend to state that they do not like corruption and will not support corrupt politicians. This has been interpreted as a product of social desirability bias: interviewees prefer to report socially accepted attitudes (rejection of corruption) instead of truthful responses (intention to vote for their preferred candidates regardless of malfeasance). We test to what extent this is the case by using a list experiment that allows interviewees to be questioned in an unobtrusive way, removing the possible effects of social desirability. Our results show that the great majority of respondents report intentions to electorally punish allegedly corrupt candidates even when asked in an unobtrusive way. We discuss the implications of this finding for the limited electoral accountability of corruption.

Keywords

Social desirability corruption list experiment accountability

Introduction

According to a large number of empirical analyses, voters do not electorally punish corrupt politicians as much as expected by democratic theory (Chang et al., 2010; Costas-Pérez et al., 2012; Dimock and Jacobson, 1995; Eggers and Fisher, 2011; Peters and Welch, 1980; Reed, 1999; Vivyan et al., 2012). This situation could reflect citizens' tolerance towards corruption, but surveys show that a large majority of citizens hold a negative view of corruption and, when asked, overwhelmingly report an intention to punish malfeasant incumbents in elections. A common interpretation of this paradox is that surveys suffer from social desirability bias, interviewees prefer to report socially accepted attitudes (rejection of corruption) instead of truthful responses (intention to vote for corrupt politicians).

This paper assesses whether vote intention towards allegedly corrupt candidates suffers from social desirability bias. For this purpose, we use a list experiment that allows interviewees to be questioned in an unobtrusive way, removing the possible effects of social desirability. Our results show that the great majority of respondents report intentions to electorally punish allegedly corrupt candidates even when asked in an unobtrusive way. When people express negative views towards corrupt politicians, they are not merely giving a socially desirable answer. Our reading of these results combined with the results of other studies (e.g. Breitenstein, 2019; Klašnja et al., 2021), is that voters prefer, ceteris paribus, the non-corrupt candidate. When choosing a candidate in a specific election, however, other considerations may play an important role in voter’s deliberations and the ceteris paribus condition is seldom realistically fulfilled. Therefore, the answer to the gap between attitudes towards corruption and electoral behavior should be looked for in the possible trade-offs voters may face in elections, but not on social desirability issues.

Social desirability in attitudes towards corrupt politicians

Evoking truthful answers can be challenging when addressing sensitive issues, such as racism, homophobia, or corruption, as citizens may adapt their answers to social expectations or simply avoid answering these questions (Corstange, 2009). The effect of social desirability bias has been widely demonstrated in issues such as vote buying (Gonzalez-Ocantos et al., 2012), racism (Kuklinski et al., 1997), turnout (Holbrook and Krosnick, 2010), and support for female candidates (Streb et al., 2008). Research on these issues has shown that when questioned directly, respondents tend to underreport socially undesirable attitudes.

This could also be the case in the electoral punishment of corrupt candidates. Confronted with a direct question, respondents usually report a very critical view of corruption (A clear majority of respondents across all countries consider it wrong and punishable to engage in different types of corrupt activities Afrobarometer Round 3; World Values Survey Wave, 6, V202). At the same time, people report very low levels of vote intention for corrupt candidates (Muñoz et al., 2016; Weitz-Shapiro and Winters, 2016). However, in real elections, voters only mildly punish corrupt politicians. When considering election results, corruption is punished in some cases (e.g. Ferraz and Finan, 2008; Wood and Grose, 2022), but many empirical analyses report that the actual electoral punishment of corruption cases is limited (Chang et al., 2010; Dimock and Jacobson, 1995; Eggers and Fisher, 2011; Peters and Welch, 1980; Reed, 1999).

This paradox could be explained by social desirability bias. This argument would suggest that people may not openly admit that they would vote for a corrupt candidate in a survey, but they may do so when the behavior is not visible, in the real election, because privately, corruption is not such a big concern.

This is a common interpretation in recent literature on corruption voting. For instance, when arguing for the need of direct measures of electoral outcomes instead of survey statements, Chong et al. (2015) remark that self-reported voting behavior suffers from social desirability bias. Other authors have also expressed similar concerns (Konstantinidis and Xezonakis, 2013; McCann and Dominguez, 1998; Winters and Weitz-Shapiro, 2013). The assumption that citizens are hiding their truthful attitudes regarding corrupt politicians is so extended, that the performance of survey data, including survey experiments, in measuring corruption accountability is being questioned (Boas et al., 2019; Incerti, 2020). For example, Incerti (2020) argues that social desirability may likely bias the ability of survey experiments to mimic real-world voting behavior.

Empirical strategy

To assess the degree of social desirability bias in vote intention for corrupt politicians, this paper uses a list experiment embedded in an online survey that took place in Spain in January 2019. The Internet-based survey was completed by 1200 Spanish citizens between 18 and 89 years of age, recruited through the online panel of the firm Netquest. The sample included quotas for sex, education, and age to resemble the Spanish population.

Spain is a country with moderate levels of corruption, scoring 58 on a scale of 0 (highly corrupt) to 100 (very clean) according to the 2018 Corruption Perception Index of Transparency International. Corruption scandals are often a salient issue in Spanish politics, and they were at the top of the agenda at the time the survey was conducted. This makes the topic relevant and realistic, as Spanish citizens are faced with these issues when casting their vote.

Furthermore, Spain represents a typical case where citizens report a highly negative view of corruption in surveys but often do not hold malfeasant politician accountable in elections, hence making it a plausible case for the argument of social desirability. According to the official survey institute of the Spanish government (CIS 2905), 87% of respondents considered corruption a problem of paramount importance in Spanish democracy in 2011. According to the same survey, only 10% of respondents would vote for a corrupt but efficient candidate rather than for an honest but inefficient candidate.

In actual election results, however, electoral punishment is very limited if at all present. According to Costas-Pérez et al. (2012), the mean electoral punishment of corrupt mayors in Spain is around 4%. Analyzing mayors under criminal investigation in two regions of Spain, Rivero & Fernández-Vázquez (2011) did not find a significant electoral punishment.

List experiments provide a way to reduce the incentives of the interviewee to hide socially undesirable answers by asking the question in an unobtrusive way and explicitly assuring anonymity (Corstange, 2009). List experiments have been successfully used to remove measurement biases from sensitive issues such as racism (Kuklinski et al., 1997), self-reported voter turnout (Holbrook and Krosnick, 2010), and vote buying (Gonzalez-Ocantos et al., 2012; Kramon, 2016) among others. If the mechanism behind the discrepancy between real data and survey responses in corruption voting is due to desirability bias, a list experiment should reveal it.

Respondents were randomly assigned to three different groups. One group was assigned to a direct question and two groups were assigned to the list experiment, either the control or treatment group. In both the control and the experimental groups, respondents were presented with a list of plausible reasons for not voting for a mayoral candidate, and then were asked how many of these were good arguments for not supporting a candidate of their preferred political party, without having to specify which ones. The list was the same for both groups except for the sensitive item (being under investigation for corruption) that was only added to the list of items for the treatment group. This technique allowed respondents to freely answer without fearing that the interviewer or the researcher might disapprove of the answers, as one cannot know what items the respondent is referencing (Kuklinski et al., 1997). A posteriori, the difference in the means between the control and treatment group allow the estimation of the prevalence of the sensitive item in the treatment group.

Given that condoning corruption is only realistic in cases where the candidate possesses other valued characteristics, the candidate’s partisan label was matched to the respondent’s preferred party. Partisanship has been found to decrease the severity of the assessment of corruption cases (Anduiza et al., 2013; Dimock and Jacobson, 1995) and to moderate the electoral punishment of wrongdoings (Solaz et al., 2019). Furthermore, the candidate was described as a well preforming politician positively valued by the municipality’s residents.

The direct question was worded as follows:

Imagine a candidate for mayor of your municipality who has a long management experience that residents value positively and who belongs to [respondent’s preferred party]. The candidate is being investigated for the urban redevelopment of land owned by a relative. What would you do in the next election?

• I would vote for him/her

• I would vote for a candidate of another party

• I would not vote

The experimental question was as follows:

Imagine a candidate for mayor of your municipality who has a long management experience, whom residents value positively and who belongs to [respondent’s preferred party]. HOW MANY of the following items do you consider to be valid reasons not to vote for this candidate of [respondent’s preferred party]?

The candidate…

The list of items for the control group (without the sensitive item) includes:

• was convicted in the past for sexually assaulting his secretary

• is very religious and against abortion and gay marriage

• slightly stutters

• is an atheist and supporter of removing religion from primary education

The list of items for the treatment group was exactly the same list but including:

• is being investigated for the urban redevelopment of land owned by a relative

Corruption is usually defined as “the abuse of public office for private gain” (World Bank), this broad definition entails very different corrupt activities such as nepotism, embezzlement, accepting and/or offering bribes. Moreover, all these activities can be minor or vast infractions, for example, giving a family member preference in seeing a doctor or accepting millions of dollars in exchange of a public contract. Following an experimental logic, we aim to control what exactly respondents think about when reading about a corrupt politician. This restricts us to using a description of a certain corrupt activity instead of the term corruption, as different type of individuals might differ in what they interpret when they read the word corruption.

The exact wording of the corruption item (is being investigated for the urban redevelopment of land owned by a relative) was designed to closely match the actual cases of corruption in Spain. This way we increase the external validity of our experiment, as it closely reflects corruption cases in Spain, and we ensure that respondents are familiar with the type of corruption and clearly identify it as such. In the last two decades Spain has experienced an expansion of corruption scandals related to the urban development (Jiménez, 2009; Jiménez and Villoria, 2019; Quesada et al., 2013; Solé-Ollé and Sorribas-Navarro, 2018). The fall in Spain’s Corruption Perception Index (Transparency International) from 7.1 in 2004 to 6.5 in 2008 has been attributed to the rise of urban planning corruption (Jiménez, 2009). Corruption related to the redevelopment of land is thus one of the most common types of corruption in Spain and citizens are well aware of this problem because famous scandals have continuously made news headlines (Jiménez, 2009; Palau and Davesa, 2013). According to an analysis of urban corruption cases published in local and national newspapers between 1999 and 2007 the most common scandals is that of a “local politician taking a bribe in exchange of introducing changes in municipal land use regulation” (Costas-Pérez, 2014: 8). As such, our treatment describes a very common pattern of behavior in cases of political corruption in Spain.

Furthermore, the item states that the candidate is under investigation, and not already convicted because this is the most common situation voters face at the polls. Politicians are rarely condemned for corruption when elections are held because corruption scandals are usually uncovered much later and judicial procedures take a long time to be solved. Furthermore, politicians are more likely to retire once there is a firm condemnation of corruption (Wood and Grose, 2022).

The other items of the list were chosen to comprise a large range of circumstances, so that these could be either accepted or rejected according to different arguments. Furthermore, following the logic put forwards by Glynn (2013), the list includes a negative correlation between two items so that respondents will only choose one of both. Respondents that like a religious politician that is against abortion and gay marriages are also more likely to be in favor of obligatory religious education in school, and vice versa. This facilitates that each respondent at least selects one item and leaves without selecting another item.

To avoid further ceiling or floor effects, we include a low-prevalence non-sensitive item (slightly stutters) and a high-prevalence very sensitive item (was convicted in the past for sexually assaulting his secretary). While being convicted for sexual harassment is likely to be selected by the majority of respondents as a reason no to vote for a candidate, slightly stuttering is likely to be selected by very few respondents. The sexual harassment item was also included so that our sensitive item of interest (being under investigation for corruption) does not stick out as the most sensitive issue in the list, and so avoid forcing respondents to react on it.

By limiting the possibility of ceiling and floor effects, we ensure the unobtrusiveness of our list experiment (Janus, 2010). According to our results, the potential of ceiling and floor effects was minimal. In the control list, only 5% answered “none” and 8% answered “all of them.”

To match respondents’ and candidate’s partisanship, party identification was assessed in a previous question, “For which of the following parties do you feel more sympathy, or which one do you feel is closer to your own ideas?” About 31% of the respondents declared that they did not feel close to any of the parties with representation in the Spanish Congress of Deputies. In that case, they were asked a follow-up question of what party they would select if they had to choose one. Even so, 19% of the respondents declared that they did not feel close to any of the political parties. These respondents were randomly assigned to one of the four parties with more representation in the Spanish parliament (PP, PSOE, Podemos, and Ciudadanos).

Our dependent variable was integrated in the treatment condition, as it was the number of features in which the respondent wouldn’t vote for the candidate. We performed a randomization check (reported in the appendix in Table A1) that does not show any special problem in how the random assignment of respondents to treatment groups was performed.

Results

The experiment does not reveal any hidden corruption voting due to potential social desirability. Using an unobtrusive way of inquiring does not elicit support for candidates under corruption allegations. As shown in Table 1, introducing the treatment item to the list results in a strong and significant increase of the dependent variable (0.825). (See Table A2 in Appendix for the piecewise proportions). This means that 82.5% of the respondents wouldn’t vote for a candidate being investigated for corruption, while 17.5% would do so. When we only take into consideration those that responded feeling close to a party and were therefore shown a candidate of their preferred party, the support for the allegedly corrupt politician increases substantially, as up to 26% of these respondents would vote for the candidate investigated for corruption.

Table 1.

Results across different treatment conditions.

	Mean in control list	Mean in treatment list	Difference in means	% vote for corrupt in list	% voter for corrupt in direct question
All respondents	2.02 (0.046)	2.845 (0.060)	0.825	17.5%	22.75%
N	400	400			400
Only partisans*	1.96 (0.052)	2.70 (0.069)	0.74	26%	28.27%
N	271	280			283

Note: First two columns show the mean of Y across control list and treatment list, standard errors in parentheses. Third column shows difference in means. Last two columns show the percent of respondents that would vote for the corrupt candidate in the list experiment and in the direct question. *Only respondents that answered feeling close to a party and were therefore assigned their favorite party in the list experiment (i.e. excluding those that answered not feeling close to any party).

These results conform to the results we obtain when the question was asked directly, as up to 23% of respondents accept that they would vote for the allegedly corrupt candidate even when asked undeviatingly. When only taking into consideration partisans, up to 28% would vote for the allegedly corrupt candidate of their preferred party. In fact, the support for allegedly corrupt politicians is even higher when respondents were directly asked what they would do. In Table A3 in the appendix, we provide evidence that our experiment does not suffer from design effects that could affect the validity of the experiment and our results (see Blair and Imai, 2012 for more information regarding this test).

Our results do not provide support for the idea that direct survey questions on voting for corrupt politicians suffer from social desirability bias. If that were the case, we would have observed a higher acceptance or “forgiveness” of corruption in responses to an unobtrusive measure like the one included in our list experiment. On the contrary, results show that a very similar proportion of respondents accept voting for an allegedly corrupt politician both in the unobtrusive question as in the direct question.

These outcomes suggest the need to identify alternative explanations when trying to make sense of the discrepancies between survey responses and actual behavior regarding corruption voting. Results from our experiment do not support the idea that voters have a true structure of preferences in which corruption is unimportant in their vote choice that gets masked by social desirability concerns in survey research. Voters seem to genuinely dislike corruption and give weight to it in their choice function.

Conclusions

According to the list experiment results, social desirability does not explain the discrepancy between the reported rejection of corruption in surveys and the actual punishment of corruption in elections. When people express a negative evaluation of corruption cases, we must assume that they report these attitudes sincerely, as they do not reveal different evaluations in unobtrusive settings.

The fact that election results tend to show limited punishment for corruption allegations and thus contradict survey responses, should therefore not be attributed to social desirability issues without further research. Such a result is compatible with alternative explanations. Voters might truly dislike corruption and consider it a liability in their evaluations of candidates. Yet, this is compatible with lower actual punishment in the election if other considerations influence the vote choice decision (Muñoz et al., 2016; Rundquist et al., 1977), or if voters do not receive or trust the information on corruption (Ferraz and Finan, 2008; Weitz-Shapiro and Winters, 2017).

The discrepancy between what is measured in surveys and what is observed in real elections is probably a consequence of the inability of standard survey questions to estimate the relative importance of corruption in elections. Direct questions in surveys may not be failing due to their lack of uncovering truthful answers but due to their struggles in estimating the relative importance of certain issues. Voters might dislike corruption, yet corruption may not be their most important consideration when deciding their vote. When it comes to real elections, voters have to consider additional aspects beyond the politician’s integrity, and they may trade integrity for other valued characteristics. The literature has highlighted the importance of partisanship (Anduiza et al., 2013; Chang and Kerr, 2017; Solaz et al., 2019), issue preferences (Rundquist et al., 1977), ideology (Charron and Bågenholm, 2016), or the politicians’ perceived competence (Breitenstein, 2019; Muñoz et al., 2016; Rosas and Manzetti, 2015) in moderating the punishment of corruption in elections.

Future work should take this into account when conceiving survey instruments to estimate the electoral effects of corruption, and when interpreting results. Survey instruments should present situations where the choices mirror the complexity of decision making in real elections. Excessively simplistic scenarios, where respondents are not confronted with the trade-offs that exist in reality, will be unlikely to provide valid information on how people punish (or not) electoral corruption at the polls. Not because respondents hide their true attitudes in surveys, but because these are unable to mimic decision making in real elections, where voters choose among a limited set of options that differ from each other in many potentially relevant dimensions.

Supplemental Material

Supplemental Material - Do they really care? Social desirability bias in attitudes towards corruption

Supplemental Material for Do they really care? Social desirability bias in attitudes towards corruption by Sofia Breitenstein, Eva Anduiza and Jordi Muñoz in Research & Politics

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the “la Caixa” Foundation (2016 ACUPO177), Ministerio de Ciencia e Innovación (CSO2017-83086-R), and Ministerio de Economía y Competitividad (BES-2015-072756).

Correction (June 2025):

The article has been updated with correct dataverse link in the supplementary material section. For more details, please see the correction notice: .

ORCID iDs

Sofia Breitenstein

Eva Anduiza

Jordi Muñoz

Supplemental Material

Supplemental material for this article is available online.

The files can be found at

References

Anduiza

Gallego

Muñoz

(2013) Turning a blind eye: experimental evidence of partisan bias in attitudes toward corruption. Comparative Political Studies 46(12): 1664–1692.

Blair

Imai

(2012) Statistical analysis of list experiments. Political Analysis 20(1): 47–77.

Boas

Hidalgo

Melo

(2019) Norms versus action: why voters fail to sanction malfeasance in Brazil. American Journal of Political Science 63(2): 385–400.

Breitenstein

(2019) Choosing the crook: A conjoint experiment on voting for corrupt politicians. Research and Politics 6(1).

Chang

ECC

Golden

Hill

(2010) Legislative malfeasance and political accountability. World Politics 62(2): 177–220.

Chang

ECC

Kerr

(2017) An Insider–outsider theory of popular tolerance for corrupt politicians. Governance 30(1): 67–84.

Charron

Bågenholm

(2016) Ideology, party systems and corruption voting in European Democracies. Electoral Studies 41: 35–49.

Chong

Ana

De La

, et al. (2015) Does corruption information inspire the fight or quash the hope? A field experiment in Mexico on voter turnout, choice, and party identification. The Journal of Politics 77(1): 55–71.

Corstange

(2009) Sensitive questions, truthful answers? Modeling the list experiment with LISTIT. Political Analysis 17(1): 45–63.

10.

Costas-Pérez

(2014) Essays on the Political Economy of Local Corruption. Available at: https://www.ub.edu/school-economics/recent_thesis/essays-on-the-political-economy-of-local-corruption-2/

11.

Costas-Pérez

Solé-Ollé

Sorribas-Navarro

(2012) Corruption scandals, voter information, and accountability. European Journal of Political Economy 28(4): 469–484.

12.

Dimock

Jacobson

(1995) Checks and choices: the house bank scandal’s impact on voters in 1992. The Journal of Politics 57(4): 1143–1159.

13.

Eggers

Fisher

(2011) “Electoral Accountability and the UK Parliamentary Expenses Scandal: Did Voters Punish Corrupt MPs?” http://papers.ssrn.com/abstract=1931868

14.

Ferraz

Finan

(2008) Exposing corrupt politicians: the effects of Brazil’s publicly released audits on electoral outcomes. The Quarterly Journal of Economics 123(2): 703–745.

15.

Glynn

(2013) What can we learn with statistical truth serum?: Design and analysis of the list experiment. Public Opinion Quarterly 77(S1): 159–172.

16.

Gonzalez-Ocantos

de Jonge

Melendez

, et al. (2012) Vote buying and social desirability bias: experimental evidence from Nicaragua. American Journal of Political Science 56(1): 202–217.

17.

Holbrook

Krosnick

(2010) Social desirability bias in voter turnout reports tests using the item count technique. Public Opinion Quarterly 74(1): 37–67.

18.

Incerti

(2020) Corruption information and vote share: a meta-analysis and lessons for experimental design. American Political Science Review 114(3): 761–774.

19.

Janus

(2010) The influence of social desirability pressures on expressed immigration attitudes. Social Science Quarterly 91(4): 928–946.

20.

Jiménez

(2009) Building boom and political corruption in Spain. South European Society and Politics 14(3): 255–272.

21.

Jiménez

Villoria

(2019) Political finance, urban development and political corruption in Spain. In Money, Corruption, and Political Competition in Established and Emerging Democracies. Lexington Books.

22.

Klašnja

Lupu

Tucker

(2021) When do voters sanction corrupt politicians? Journal of Experimental Political Science 8(2): 161–171.

23.

Konstantinidis

Xezonakis

(2013) Sources of tolerance towards corrupted politicians in Greece: the role of trade offs and individual benefits. Crime, Law and Social Change 60(5): 549–563.

24.

Kramon

(2016) Where is vote buying effective? Evidence from a list experiment in Kenya. Electoral Studies 44: 397–408.

25.

Kuklinski

Cobb

Gilens

(1997) Racial attitudes and the New South. The Journal of Politics 59(02): 323–349.

26.

McCann

Domıinguez

(1998) Mexicans react to electoral fraud and political corruption: an assessment of public opinion and voting behavior. Electoral Studies 17(4): 483–503.

27.

Muñoz

Anduiza

Gallego

(2016) Why do voters forgive corrupt mayors? Implicit exchange, credibility of information and clean alternatives. Local Government Studies 42(4): 598–615.

28.

Palau

Davesa

(2013) El Impacto de La Cobertura Mediática de La Corrupción En La Opinión Pública Española/The Impact of Media Coverage of Corruption on Spanish Public Opinion. Revista Española de Investigaciones Sociológicas, 144: 97–126.

29.

Peters

Welch

(1980) The effects of charges of corruption on voting behavior in congressional elections. The American Political Science Review 74(3): 697–708.

30.

Quesada

Jiménez-Sánchez

Villoria

(2013) Building local integrity systems in southern europe: the case of urban local corruption in Spain. International Review of Administrative Sciences 79(4): 618–637.

31.

Reed

(1999) Punishing corruption: the response of the Japanese electorate to scandals. In Political Psychology in Japan. Nova Science Publishers.

32.

Rivero

Fernández-Vázquez

(2011) Las Consecuencias Electorales de Los Escándalos de Corrupción Municipal 2003-2007. Estudios de Progreso 59(201): 1.

33.

Rosas

Manzetti

(2015) Reassessing the trade-off hypothesis: how misery drives the corruption effect on presidential approval. Electoral Studies 39: 26–38.

34.

Rundquist

Strom

Peters

(1977) Corrupt politicians and their electoral support: some experimental observations. American Political Science Review 71(03): 954–963.

35.

Solaz

De Vries

Roosmarijn

, et al. (2019) In-group loyalty and the punishment of corruption. Comparative Political Studies 52(6): 896–926.

36.

Solé-Ollé

Sorribas-Navarro

(2018) Trust no more? On the lasting effects of corruption scandals. European Journal of Political Economy 55: 185–203.

37.

Streb

Burrell

Frederick

, et al. (2008) Social desirability effects and support for a female American president. Public Opinion Quarterly 72(1): 76–89.

38.

Vivyan

Wagner

Tarlov

(2012) Representative misconduct, voter perceptions and accountability: evidence from the 2009 house of commons expenses scandal. Electoral Studies 31(4): 750–763.

39.

Weitz-Shapiro

Winters

(2017) Can citizens discern? Information credibility, political sophistication, and the punishment of corruption in Brazil. The Journal of Politics 79(1): 60–74.

40.

Winters

Weitz-Shapiro

(2013) Lacking information or condoning corruption: when do voters support corrupt politicians? Comparative Politics 45(4): 418–436.

41.

Wood

Grose

(2022) Campaign finance transparency affects legislators’ election outcomes and behavior. American Journal of Political Science 66(2): 516–534.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.38 MB