Public Opinion and Foreign Electoral Interventionto know whether Russian intervention caused public sentiment about democracy to slide, especially since the ... denounced Russian interference

American Political Science Review, Page 1 of 18

doi:10.1017/S0003055420000064 © American Political Science Association 2020

Public Opinion and Foreign Electoral InterventionMICHAEL TOMZ Stanford UniversityJESSICA L. P. WEEKS University of Wisconsin–Madison

Foreign electoral intervention is an increasingly important tool for influencing politics in othercountries, yet we know little about when citizens would tolerate or condemn foreign efforts to swayelections. In this article, we use experiments to study American public reactions to revelations offoreign electoral intervention. We find that even modest forms of intervention polarize the public alongpartisan lines. Americans are more likely to condemn foreign involvement, lose faith in democracy, andseek retaliation when a foreign power sides with the opposition, than when a foreign power aids their ownparty. At the same time, Americans reject military responses to electoral attacks on the United States, evenwhen their own political party is targeted. Our findings suggest that electoral interference can divide andweaken an adversary without provoking the level of public demand for retaliation typically triggered byconventional military attacks.

INTRODUCTION

The discovery that Russia executed awide-rangingplan to influence the 2016 U.S. Presidential racehas sparked a global debate about foreign in-volvement in democratic politics.Although countries havelong interfered in each other’s elections (Bubeck andMarinov 2017, 2019; Levin 2016, 2019b),1 the scope andsophistication of Russian activities signaled the arrival ofanewera.Changes in information technologynowmake itpossible for states to undertake ambitious influence cam-paigns in faraway countries, even when outmatched froma conventional military standpoint. Moreover, observershave struggled to identify effective strategies for stoppingthis potentially powerful form of foreign influence.2 One

can, therefore, expect more foreign efforts to shape elec-tions in the future.

In this article, weuse survey experiments to investigatethree fundamental questions about how Americanswould respond to revelations of foreign electoral in-tervention. First, when would U.S. citizens tolerate for-eign involvement in American elections, instead ofcondemning external efforts to tip the scales? Polls con-ducted after the 2016 election show that Democrats andRepublicans expressed different opinions about Russianinterference. Democrats were more likely to believe thatRussia interfered,more likely to think thatRussia alteredthe outcome of the election, and more concerned aboutthe potential for foreign meddling in the future.3

It is difficult to draw general conclusions from thishistorical episode, however. We do not know, for in-stance, how Americans would respond if the shoe wereon the other foot. If citizens learned that a foreigncountry had intervened on behalf of a Democraticcandidate, would Democrats denounce the foreign in-tervention as an unacceptable attack on Americandemocracy, or would they condone the foreign assis-tance? Would Republicans change their tune, as well,disapproving more strongly of foreign aid for a Demo-cratic candidate than for a Republican one? Wouldpublic reactions depend not only on the intendedbeneficiary of the intervention but also on the form ofmeddling?Data from 2016 cannot provide the answers,but we can investigate these issues systematicallythrough experiments.

This article also uses experiments to address a secondfundamental question: when would news of foreign elec-toral intervention undermine confidence in democraticinstitutions? One ostensible goal of the 2016 Russian in-tervention was to make Americans doubt their own po-litical system. Although Americans espouse less approvalof domestic institutions now than before 2016, it is difficult

Michael Tomz , William Bennett Munro Professor in PoliticalScience, Stanford University, Senior Fellow, Stanford Institute forEconomic Policy Research, [email protected].

Jessica L. P. Weeks , Associate Professor of Political Science,University of Wisconsin–Madison, [email protected].

This material is based in part on work supported by the NationalScience Foundation under award numbers SES-1226855 and SES-1226824. For extremely helpful comments on earlier versions of thisarticle, we thank participants at the American Political Science As-sociation conference (2018, 2019), the International Studies Associ-ation conference (2018), the Barcelona-Gothenburg-BergenWorkshop in Experimental Political Science (2018), the MidwestPolitical Science Association conference (2019), the Upper MidwestInternational Relations Conference (2018), and the Pacific In-ternational Politics Conference (2018), and seminar participants atAcademia Sinica, Berkeley, Columbia, Cornell, Facebook, HebrewUniversity, Princeton, Stanford, the University of Wiscon-sin–Madison, and Washington University in St. Louis. We are alsograteful for comments from Kirk Bansak, Mark Bell, Andrew Blin-kinsop,SarahBush,StephenChaudoin,BenFordham,HerbLin, JoshKertzer, Jon Krosnick, Carrie Lee, Dov Levin, Jordi Muñoz, UrtePeteris, Jon Pevehouse, Katy Powers, Lauren Prather, and JonathanRenshon. Replication files are available at the American PoliticalScience Review Dataverse: https://doi.org/10.7910/DVN/E3BAO5.

Received: January 16, 2019; revised: December 4, 2019; accepted:February 1, 2020.

1 Levin (2016) calculates that theUnited States and theUSSR/Russiaintervened to help specific candidates in one-ninth of all competitivenational-level executive elections from 1946 to 2000.2 Tenove et al. 2018.

3 For examples of polls, see the June 2017 Suffolk University/USAToday Poll at https://www.suffolk.edu/-/media/suffolk/documents/academics/research-at-suffolk/suprc/polls/national/marginal-and-table/final_june_national_crosstabs_6-28_pdftxt.pdf?la5en&hash5F8CCE8E75ECA9AC39155ADCF0AC9B4CC-E9250823, pp.21–2,or theAugust2017PRRIPollathttps://www.prri.org/research/poll-trump-russia-investigation-impeachment-republican-party/.

1

https://doi.org/10.1017/S0003055420000064https://orcid.org/0000-0002-0453-024Xmailto:[email protected]://orcid.org/0000-0002-5019-2261mailto:[email protected]://doi.org/10.7910/DVN/E3BAO5https://www.suffolk.edu/-/media/suffolk/documents/academics/research-at-suffolk/suprc/polls/national/marginal-and-table/final_june_national_crosstabs_6-28_pdftxt.pdf?la=en&hash=F8CCE8E75ECA9AC39155ADCF0AC9B4CCE9250823https://www.suffolk.edu/-/media/suffolk/documents/academics/research-at-suffolk/suprc/polls/national/marginal-and-table/final_june_national_crosstabs_6-28_pdftxt.pdf?la=en&hash=F8CCE8E75ECA9AC39155ADCF0AC9B4CCE9250823https://www.suffolk.edu/-/media/suffolk/documents/academics/research-at-suffolk/suprc/polls/national/marginal-and-table/final_june_national_crosstabs_6-28_pdftxt.pdf?la=en&hash=F8CCE8E75ECA9AC39155ADCF0AC9B4CCE9250823https://www.suffolk.edu/-/media/suffolk/documents/academics/research-at-suffolk/suprc/polls/national/marginal-and-table/final_june_national_crosstabs_6-28_pdftxt.pdf?la=en&hash=F8CCE8E75ECA9AC39155ADCF0AC9B4CCE9250823https://www.suffolk.edu/-/media/suffolk/documents/academics/research-at-suffolk/suprc/polls/national/marginal-and-table/final_june_national_crosstabs_6-28_pdftxt.pdf?la=en&hash=F8CCE8E75ECA9AC39155ADCF0AC9B4CCE9250823https://www.suffolk.edu/-/media/suffolk/documents/academics/research-at-suffolk/suprc/polls/national/marginal-and-table/final_june_national_crosstabs_6-28_pdftxt.pdf?la=en&hash=F8CCE8E75ECA9AC39155ADCF0AC9B4CCE9250823https://www.suffolk.edu/-/media/suffolk/documents/academics/research-at-suffolk/suprc/polls/national/marginal-and-table/final_june_national_crosstabs_6-28_pdftxt.pdf?la=en&hash=F8CCE8E75ECA9AC39155ADCF0AC9B4CCE9250823https://www.prri.org/research/poll-trump-russia-investigation-impeachment-republican-party/https://www.prri.org/research/poll-trump-russia-investigation-impeachment-republican-party/

to know whether Russian intervention caused publicsentiment about democracy to slide, especially since thedownward trendbegan longbefore the2016election.Howmuch stronger would faith in American democracy be ifforeign powers refrained from interfering in U.S. elec-tions? This question is difficult to answer with historicaldata, but it becomes tractable with survey experiments.

Finally, we use experiments to shed light on a thirdfundamental question: when would Americans allowforeign intervention to pass instead of demanding re-taliation? In theaftermathof 2016, someU.S. politiciansdenounced Russian interference as an act of war andlikened it to the attacks of September 11, 2001, whichprecipitated the U.S. war on terror and military in-tervention in Afghanistan. Senator Ben Cardin (D-MD) explained, “when you use cyber… to compromiseour democratic, free election system, that’s an attackagainst America. It’s an act of war.”4 Others counteredthat Russian behavior was neither “an initiation ofarmed conflict” nor “a violation of the U.N Charter”andwouldnot justify amilitary response.5Howdoes therevelation of election interference affect public supportfor diplomatic, economic, and military retaliationagainst the aggressor, and to what extent might re-taliatory sentiments split along partisan lines?

To answer these questions, we embedded experimentsin a large-scale survey of the American public. AllrespondentsreadavignetteaboutafutureU.S.presidentialelection. In some vignettes, a foreign government verballyendorsedoneofthecandidates, threatenedconsequencesifits preferred candidate did not win, or supported a candi-date by providing funding, manipulating information, orhacking into voting machines. In other vignettes, the for-eign country stayed out of the election entirely.

In addition to randomizing the existence and natureof the electoral intervention, we randomized whichcandidate—Democratic or Republican—the foreigncountry favored. Finally, we randomized informationabout the identity of the foreign country and confidencein that assessment. Having presented the vignettes, wemeasured three sets of dependent variables: condem-nation of the intervention, faith in American de-mocracy, and support for retaliation.

Our experiments revealed that news of foreign in-tervention polarizes the public along partisan lines. In-stead of rejecting foreign involvement tout court,Americans exhibited a partisan double standard. BothDemocratsandRepublicansweremorelikely tocondemnforeign involvement, lose faith in democracy, and call forretaliation when a foreign power sided with the opposi-tion, than when a foreign power aided their own party.

Our experiments also revealed that even modestforms of electoral intervention can divide and

demoralize the country. Although operations such asfunding, defamation, and hacking were most corrosive,mere endorsements by foreign countries also provokedsubstantial public ire, undermined faith in democraticinstitutions, and split the nation along partisan lines. Atthe same time, Americans—including the victims ofelectoral intervention—were unwilling to retaliateharshly. These findings suggest that electoral in-terference can be a destructive offensive tool, sowingpublic discord and eroding faith in democracy withoutprovoking the level of public demand for retaliationtypically triggered by conventional military attacks.

HOW DOES FOREIGN ELECTORALINTERVENTION AFFECTPUBLIC ATTITUDES?

In recent decades, it hasbecome increasingly common forcountries to involvethemselves inforeignelections.Often,this involvement aims to enhance democracy withoutfavoring a particular candidate or party. Before elections,foreign governments and NGOs assist with electoralreforms, and during elections, they monitor activities todetect and deter irregularities.6 Given the growth offoreign election assistance, an expanding literatureinvestigates how external observers affect domestic per-ceptions of the quality of elections (e.g., Brancati 2014;Bush and Prather 2017, 2018, 2019; Robertson 2015).

In some cases, however, countries seek to tip the scales;they use rhetoric and/or resources to give specific partiesor candidates an electoral advantage (Bubeck and Mar-inov 2017, 2019; Bush and Prather 2020; Corstange andMarinov 2012; Levin 2016, 2019b; Martin and Shapiro2019). We refer to these types of activities as foreignelectoral interventions.7 The Russian interference of2016, which aimed to help Donald Trump defeatHillary Clinton, exemplifies this form of intervention.

Past Research about ForeignElectoral Interventions

Researchers have recently begun to investigate theeffects of foreign electoral intervention. In a pioneeringstudy, Levin (2016) examined how interventions by

4 Nelson, Louis. “Cardin: Russia’s Election Meddling is ‘An Act ofWar’.” Politico, November 1, 2017. https://www.politico.com/story/2017/11/01/russia-meddling-us-elections-ndi-event-244414.5 Nakashima, Elle. “Russia’s Apparent Meddling in U.S. Election isNot anAct ofWar,CyberExpert Says.”WashingtonPost. February 7,2017. https://www.washingtonpost.com/news/checkpoint/wp/2017/02/07/russias-apparent-meddling-in-u-s-election-is-not-an-act-of-war-cyber-expert-says/?utm_term5.55e860dba0b9.

6 E.g., Hyde 2011; Hyde and Marinov 2014; Kelley 2008.7 Our use of the term “electoral intervention” is closest to what Levincalls a “partisan electoral intervention” (Levin 2016, 2019). Levin(2019b, 90) defines a “partisan electoral intervention” as “a situationin which one or more sovereign countries intentionally undertakesspecific actions to influenceanupcomingelection in another sovereigncountry in an overt or covert manner which they believe will favor orhurt one of the sides contesting that election andwhich incurs, or mayincur, significant costs to the intervener(s) or the intervened country.”As discussed below, however, we include one type of electoral in-tervention that Levin excludes, namely, “positive/negative things saidabout a candidate/party by the intervener before an election with noconcrete threats/promises” (Levin 2019b, 91). Our use of the termelectoral intervention is narrower than Bubeck and Marinov (2017,2019) and Corstange and Marinov (2012), who define electoralinterventions to include not only partisan interventions but also de-mocracy promotion.

Michael Tomz and Jessica L. P. Weeks

2

https://www.politico.com/story/2017/11/01/russia-meddling-us-elections-ndi-event-244414https://www.politico.com/story/2017/11/01/russia-meddling-us-elections-ndi-event-244414https://www.washingtonpost.com/news/checkpoint/wp/2017/02/07/russias-apparent-meddling-in-u-s-election-is-not-an-act-of-war-cyber-expert-says/?utm_term=.55e860dba0b9https://www.washingtonpost.com/news/checkpoint/wp/2017/02/07/russias-apparent-meddling-in-u-s-election-is-not-an-act-of-war-cyber-expert-says/?utm_term=.55e860dba0b9https://www.washingtonpost.com/news/checkpoint/wp/2017/02/07/russias-apparent-meddling-in-u-s-election-is-not-an-act-of-war-cyber-expert-says/?utm_term=.55e860dba0b9https://www.washingtonpost.com/news/checkpoint/wp/2017/02/07/russias-apparent-meddling-in-u-s-election-is-not-an-act-of-war-cyber-expert-says/?utm_term=.55e860dba0b9

great powers affectedelectionoutcomes in target states.He found that partisan electoral interventions by theUnited States and the USSR/Russia during the years1946–2000 increased the vote share of the favoredcandidateby threepercentagepoints, onaverage.Levinalso showed that partisan electoral interventions con-tributed to political instability by encouraging the for-mation of domestic terrorist groups, increasing the riskof terrorist incidents, and raising the probability ofa democratic breakdown (Levin 2018, 2019a).

Weknowlessabouthowcitizenswould judgetheactofforeign intervention itself. If citizens became aware offoreign involvement,whenwould they tolerateandwhenwould they condemn efforts to influence their ownelections? Only two studies, to our knowledge, haveexamined this important question. Both studies madesignificant strides, but as we explain below, they reachedconflicting conclusions. Moreover, by focusing on spe-cific episodes in Ukraine and Lebanon, the studies leftopen how citizens in mature democracies would react toforeign electoral intervention, how different modes ofintervention would influence public reactions, andwhether foreign interference would erode confidence indemocracy and provoke retaliatory foreign policies.

In one groundbreaking study, Shulman and Bloom(2012) analyzed public approval of Russian andWestern involvement in the 2004 Ukrainian presiden-tial elections. During that election, Russia offered“nakedly partisan” support for incumbent ViktorYanukovych (455). The Russian government contrib-uted money to Yanukovych’s campaign, and RussianPresident Vladimir Putin backed him publiclythroughout the election. Western efforts were not asopenly one-sided. The United States, the EU, andWestern organizations took care not to campaign forYanukovych’s rival, Viktor Yuschenko, and they didnot contribute money to his campaign. Western coun-tries did, however, disproportionately fund oppositionparties, and Western election monitoring and exitpolling contributed toYanukovych’sdefeatbyexposingelectoral fraud.

To study how Ukrainians reacted to Russian andWestern involvement, Shulman and Bloom analyzedpublic opinion surveysfieldedapproximately a year afterthe election. They found that the public disapproved ofbothWestern and Russian activities, and concluded that“foreign influence over any aspect of a state’s politicaldevelopment, especially one that so closely symbolisesself-rule such as elections, risks unleashing a backlashfueled by citizens jealously guarding their national au-tonomy and national identity” (470).8

A second study, which focused on the 2009 parlia-mentary elections in Lebanon, reached a differentconclusion (Corstange and Marinov 2012). Corstangeand Marinov innovated methodologically by running

a survey experiment that randomly conveyed in-formation about American or Iranian support for oneside in the Lebanese election. They then measured thepublic’s desire to protect foreign relations with theUnited States and Iran, as well as satisfaction with therole the foreign country played.

Corstange and Marinov found that foreign in-tervention “did not provoke a nationalistic backlashagainst anymeddling in domestic affairs” (667). Indeed,unlike Ukrainian voters who generally rejected foreigninterference, Lebanese voters sometimes appeared toappreciate foreign intervention on behalf of their pre-ferred candidate. When Lebanese citizens heard thatthe United States had favored one side, those whoagreed with the American position increased theirdesire for cooperation with the United States, whereasthose who disagreed with the American positiondowngraded the importance of U.S. relations. News ofIranian interference did not provoke a similar splitreaction, however, raising questions about why someinterventions would divide the public, whereas otherswould not.9

Those two studies not only reached different con-clusions but also left several fascinating questionsunanswered. For instance, how would voters in a long-standing democracy such as the United States judgeforeign electoral intervention? As Corstange andMarinov (2012, 659) argue, voters in “fragile and un-consolidated” democracies may tolerate interventionsthat help their side. When the future of democracy is indoubt, it could be rational to prioritize short-term po-litical gains over the potential negative effects of foreignmeddling. Consequently, voters in unstable de-mocracies may react more positively on average, and ina more polarized way, than voters in longstanding de-mocracies such as the United States.

Previous research also left unclear how voters wouldrespond to different types of electoral interventions.Corstange and Marinov (2012) reminded respondentsthat theUnitedStatesorIran“madeitclear that it stronglypreferred one side over the other,” without specifyingwhether the intervention went beyond verbal endorse-ments. Shulman and Bloom (2012), in contrast, studiedhistorical interference that included both endorsementsandothermeasures,making itdifficult todeterminewhichactions provoked the most public ire. By experimentallyrandomizing what the foreign country did, one couldcompare reactions to various forms of interference.

Finally, previous studies did not reveal how foreigninterference affected faith in democracy or support forreprisals. We develop hypotheses about these themes,thereby setting the stage for our survey experiments. Indeveloping our predictions, we adapt previous work tothe American context and extend it to cover a broaderrange of potential foreign intrusions.

8 Shulman and Bloom found that Ukrainians disapproved more ofWestern involvement than of Russian involvement, even thoughRussian interference was more openly partisan. They hypothesizedthat many Ukrainians identified more strongly with Russia than theWest, and therefore viewedRussian involvement as less of a challengeto Ukrainian sovereignty.

9 As Corstange and Marinov (2012, 667) point out, reactions to theIranian treatment might have been weaker because, prior to theexperiment, Lebanese were more familiar with the objectives andbehaviorof Iran thanof theU.S. Ina follow-uparticle,Marinov (2013)used the same experimental data to examine how education andpolitical sophistication moderated individuals’ reactions.

Public Opinion and Foreign Electoral Intervention

3

Hypotheses about Toleranceversus Condemnation

We hypothesize that American tolerance of foreign in-tervention shoulddependon the typeof interventionandthe intended beneficiary. We distinguish three modes ofinterference: verbal endorsements, threats, and oper-ations. Endorsements occur when foreign countries ex-press their opinions about candidates. Threats combineanendorsementwithapromiseof futurerewardor threatof future punishment, such as threatening to downgradefuture relations if the preferred candidate loses. Finally,we use the term operations when foreign powers un-dertake efforts such as spreading embarrassing in-formation about a candidate, hacking into votingsystems, or donating money to an election campaign.10

Onemight suppose thatAmericans would disapproveequallywhethera foreigncountrystates itsopinionabouta candidate, issues a threat, or engages in operations.After all, each of these activities has the potential toinfluence the election and could be seen as inappropriateforeign involvement in U.S. domestic affairs.11 Wesuggest, however, that operations should provoke moreAmerican disapproval than threats, and that threatsshould provoke more ire than endorsements.

First, operations such as information campaigns andhacking could be perceived as greater challenges todemocracy. Polls show that Americans rate democracyas the best form of government and express over-whelming support for fair elections.12 We anticipatethat citizens will judge foreign intervention based on itspractical consequences for the election, as well as itsconsistency with democratic norms. In terms of con-sequences, we expect citizens to object more strongly tointerventions that they believe affected the outcome.Normatively speaking, citizens should recoil at behaviorthat seems inconsistentwithdemocratic values,whetheror not they think it shaped the outcome.13 By thisreasoning, we anticipate that many Americans wouldregard foreignendorsements as harmless and legitimateforms of free speech, while viewing operations asconsequential and inherently antidemocratic. We ex-pect threats to provoke an intermediate reaction.

Second, operations and threats may be regarded asgreater violations of sovereignty. The norm of sover-eignty—that countries should not interfere in the in-ternal affairs of other countries—is well established ininternational law and a foundation of the U.N. Charter.

Shulman and Bloom (2012, 470) found that the com-mitment to sovereignty was “alive and well” in Ukraineand helped explain why Ukrainians rejected foreigninterference during the presidential elections of 2004.We anticipate that perceptions of consequences andnorms will influence public judgments about whetherthe intervention violated U.S. sovereignty. We predictthat citizens will be most concerned about operationssuch as hacking into voting systems or donating money,as these directly advantage the favored candidate andinvolve behavior the U.N. has classified as impermis-sible interference in the internal affairs of another na-tion.14 Americans should be more tolerant of threatsandmost tolerant of endorsements, which could be seenas legitimate and harmless expressions of opinion thatdo not intrude on American sovereignty.

We also hypothesize that revelations of foreign in-terventionwillgeneratepolarizedpartisanresponses.Onemight expectAmericans to disapprove any time a foreigncountry becomes involved in a U.S. election, just as theyrecoil against conventionalmilitaryattacksonU.S. troops,territory,orequipment.However,unliketraditional formsof foreign intervention, partisan electoral interventionscreate domestic winners and losers: they help one can-didate or party at the expense of others.15 Given thepossibility of asymmetric partisan gain, we anticipate thatAmerican voters will disapprovemore strongly of foreignmeddling on behalf of political opponents, than of foreignmeddling to assist their own party.16

At least three mechanisms could contribute to thisdouble standard.The first is consequentialist. In additionto valuing democracy and sovereignty, voters care aboutpolicy outcomes. Many voters believe that victory bytheir own party would produce better policies than vic-tory by the opposition. Voters should therefore disap-provemorestronglyof interferenceonbehalfofdomesticpolitical adversaries, since such interference could con-tribute to bad policy outcomes. Conversely, they shouldbemoretolerantofassistancefordomesticpolitical allies,since foreign assistance could help domestic allies winand contribute to more desirable policies.

The secondmechanism is perceptual.We argued thatwhen voters judge whether an intervention underminesdemocracy and sovereignty, they consider how conse-quential the intervention was for the outcome of theelection. Such perceptions are, however, prone topartisan bias. Psychological studies have shown thatpeople overestimate the extent to which others sharetheir opinions (Ross, Greene, and House 1977). This“false consensus effect” is evident in many spheres,including politics, where members of a political partyoverestimate public support for their own side (Dela-vande andManski 2012). Studies also show that citizens

10 Operations are typically designed to be hidden from the public eyeandmay remain secret for years afterward.Our experiments evaluatehow citizens react once information about electoral interventionbecomes public.11 For example, endorsements could send signals about the quality ofcandidates or lead the public to draw conclusions about future rela-tions with the foreign country.12 Drutman,Diamond, andGoldman2018;PewResearchCenter2018.13 Public reactions could differ in countries where democratic normsare more tenuous. Indeed, Corstange andMarinov (2012, 659) foundthat Lebanese citizens tolerated foreign meddling in their fledglingdemocracy,while stressing that acquiescencewouldbe“implausible inconsolidated democracies.”Weexpect a stronger negative reaction inthe United States than Corstange and Marinov found in Lebanon.

14 See, for example, U.N. Resolution 60/164 (16 December 2005).15 Although foreign military attacks may provide opportunities fornarrow domestic groups such as weaponsmanufacturers, such attackstypically do not benefit large groups of U.S. citizens.16 Our prediction dovetails with research showing that citizens applya partisan double standard when thinking about domestic issues, in-cluding political corruption (Anduiza, Gallego, and Muñoz 2013),election fraud(Beaulieu2014), andelection rules (Ahlquist etal. 2018).


4

interpret data selectively, accepting news that portraystheir party in a positive light while dismissing news thatportrays their party in a negative light (Bush andPrather 2017).17 These types of cognitive biases couldcause citizens to perceive foreign intervention onbehalfof their own party as less consequential—and thereforeless objectionable—than foreign intervention on behalfof the opposition.18

The final mechanism is symbolic. In sports, peopledisapprove when fans cheer for the opposition, evenwhen cheerleading does not affect the outcome ofa match.19 A similar logic applies to politics: expressingenthusiasm forone’sownparty seems lessobjectionablethan expressing support for the opposition, even whenexpressions of support would not undermine de-mocracy or sovereignty or alter the outcome of anelection. As such, even when foreign meddling has noreal impact, we predict that citizens will disapprovemore stronglyofmeddlerswho took thewrong side thanof meddlers who supported the right political team.

In summary, we suggest three reasons why citizensmight exhibit a partisan double standard. Foreign in-tervention on behalf of the opposition could be seen ascontributing to bad policy outcomes, be perceived asmore likely to influence the outcome of the election, and/or be castigated as symbolic support for the wrong team.Ofcourse,notall votershavefirmpartisanaffiliations.Weanticipate that independent voters will not discriminatebased on which party the intervention aimed to help.

Hypotheses about Faith in Democracy

In addition to provoking disapproval, the discovery offoreign involvement in elections could change attitudesabout democracy. Previous studies have found that med-dling bydomestic actors raises doubts about the integrity ofelections, triggering a chain reaction that delegitimizes thepolitical system, depresses voter turnout, and encouragesmass protest (Norris 2014; Tucker 2007; Wellman, Hyde,and Hall 2018). Research has documented the prevalenceandconsequencesof domestic threats to electoral integrity,including efforts to block opposition parties, censor themedia, launder campaign funds, gerrymander districts,suppress turnout, buy votes, stuff ballots, and manipulatethe rules that translate votes into seats (e.g., Ahlquist et al.2018; Birch 2011; Simpser 2013).

We extend this line of inquiry to the internationalrealm by assessing how foreign interference affectsattitudes about democracy.20 One might think thatAmericans would view foreign interference as a minor

annoyance that would not shake their confidence inAmerican democracy.We expect, however, that news offoreign interference will harm faith inU.S. elections andinstitutions. Interventions could raise suspicions aboutwhether electoral outcomes reflect the will of theAmericanpeople.Learningof foreign interventioncouldalso sap faith in democratic institutions and depress fu-ture political participation (Norris 2014), although weanticipate a bigger impact on proximate outcomes (suchas distrusting the results of the immediate election) thanon distant and diffuse ones (such as losing faith in de-mocracy or abstaining from future elections).

We further predict that some types of interventionwillinflict more damage than others. Although democracy isa contested concept, most would regard free and fairelections as the sine qua non of a democratic system(Dahl 1971).Foreignbehavior thatbiases theoutcomeofan election should, therefore, heighten public concernsabout the health of American democracy. Operationssuch as hacking, funding, and disinformation seemmostlikely to raise suspicions of bias, whereas endorsementsseem least likely to cause bias, with threats somewhere inbetween. We therefore predict that operations will sowdistrust and undermine public confidence in politicalinstitutions to a greater degree than threats, and thatthreats will inflict more damage than endorsements.

Finally, we hypothesize that foreign interference willhave especially corrosive effects on the democraticconfidence of citizens whose party was attacked. Asexplained earlier, motivated reasoning should lead citi-zens to perceive attacks on their own party as moreconsequential—and therefore more threatening todemocracy—than attacks on the opposition. Moreover,research has shown that citizens on the losing side of anelection exhibit less faith in democratic institutions thancitizens on the winning side (e.g., Anderson et al. 2005;Anderson and Guillory 1997).21 Following this logic,foreign intervention should be especially demoralizingwhen it appears to help the opposition win.

To what extent would such reactions weaken Amer-ican democracy? Luckily, researchers expect long-standing democracies to be resilient in the face ofsporadic election irregularities (Norris 2014).One or twoinstances of foreign meddling would, therefore, be un-likely to trigger a collapse of America’s seasoned dem-ocratic institutions. In recent years, however, Americanshave expressed high levels of dissatisfaction with U.S.elections, citing problems with gerrymandering, votingrights, and unrestricted campaign contributions.22 In thatcontext,Americansmight see foreign interventionaspartof a systemic problem, rather than an isolated setback.Whether or not American faith in democracy reachesa crisis point, any downward trend would be worrisomebecause when citizens distrust political institutions,leaders find it more difficult to govern effectively (Levi1998) and to rally public support for government activity(Chanley,Rudolph, andRahn2000;Hetherington2004).

17 According to the classic Michigan model, partisanship “raisesa perceptual screen through which the individual tends to see what isfavorable to his partisan orientation” (Campbell et al. 1960, 133).18 Both processes are examples of motivated reasoning, see Bolsen,Druckman, and Cook 2014; Leeper and Slothuus 2014; Lodge andTaber 2013.19 On in-groupbias and intergroupdiscriminationmore generally, seeTajfel and Turner 1979.20 Whereas others have studied responses to foreign electionobservers (e.g., Brancati 2014; Bush and Prather 2017), we examineforeignefforts to giveone sideapolitical advantage. SeealsoBushandPrather (2019).

21 The gap could arise because losers react negatively to loss, winnersreact positively to victory, or both (Esaiasson 2011; Wellman, Hyde,and Hall 2018).22 Pew Research Center (2018).


5

In addition, declining turnout is concerning from thestandpoint of democratic representation, particularlywhen turnout falls unevenlyacross thepolitical spectrum.Thus, dwindling faith in U.S. democracy could haveimportant political and normative implications.

Foreign Electoral Intervention and ForeignPolicy Preferences

Finally, foreignelectoral interventioncouldspurdemandsfor retaliation. We consider two broad categories ofoptions. The United States could take nonmilitarymeasures such as severing diplomatic relations with theoffending country or imposing economic sanctions, ormilitarymeasures such as threatening to use force or evenlaunching a military strike against the meddling nation.

Which measures would citizens support, and underwhat conditions? We expect lower public support formilitary options than for nonmilitary ones. This pre-diction may reflect not only prudential concerns aboutthe human and economic costs of military engagementbut also normative skepticism about whether violencewould constitute an appropriate response to nonkineticattacks (Kreps and Das 2017; Kreps and Schneider2019). Although some American politicians charac-terized the Russian interference of 2016 as an “act ofwar,” election interference presumably does not qualifyas an “armed attack” that would justify military re-taliationunder justwar theoryor international law.23Tothe extent that citizens share this view, they shouldprefer nonmilitary options over military ones.

However, support for retaliation of any kind shoulddepend on the mode and partisanship of the foreignintervention. Previously, we explained why citizenswould be most upset about operations and least upsetabout endorsements. We further argued that Ameri-cans would object more strongly to interventionsdesigned to harm their own party. Carrying thesearguments to their logical conclusion, citizens should bemost inclined to retaliate against operations, followedby threats and verbal endorsements. They should alsobemorewilling topunish attacks against their ownpartythan attacks on the opposition.24

Moreover, the desire for retaliation should increasewith certainty about the identity of the perpetrator(Kreps and Das 2017). In some cases, intelligenceagencies may have a hard time inferring who wasmeddling, especially if the foreign power used coverttactics to launder campaign contributions, spread dis-information,orhack intovoting systems. Inothercases, itmight be obvious which country carried out the electoralintervention.Weexpect that citizenswill bemorewillingto retaliate if they are certain they are punishing the realoffender, rather than a likely but unproven suspect.

RESEARCH DESIGN

To test these hypotheses, we administered a survey ex-periment to a diverse sample of 3,510 U.S. adults inMarch–April 2018. The sample was recruited by Lucid,which selected participants to resemble the gender, age,geographic, and racial distribution of the U.S. adultpopulation.

We began by telling participants: “On the next fewpages, we will describe a situation that could take placein the future. Please read thedescriptioncarefully.Afteryou have read about the situation, we will ask for youropinions.” All respondents then received a vignetteabout the U.S. Presidential election of 2024.

We randomly assigned each participant to one of fourgroups, which varied in the degree to which a namedforeign country interferedwith the election.Members ofthe endorsement group received a scenario in which thecountry publicly announced its preference for one of thecandidates. Members of the threat group received a vi-gnette in which the country not only announced itspreference but also intimated that a disappointing out-comewould prompt it to rethink its relationship with theU.S. Members of the operation group read a story inwhich agents from the foreign country used money, in-formation, or hacking in an attempt to give their favoredcandidate an electoral advantage. Finally, members ofthe stayoutgroupreceivedavignette inwhich the foreigncountry did not meddle in the U.S. election.25

We now describe each treatment in more detail.Members of the endorsement group read the followingvignette, with randomized components in italics:

In 2024, the government of [country] made several publicstatements during the U.S. Presidential election campaign.[Country] said that it stronglypreferred[candidate]andhoped[candidate] would win the U.S. Presidential election. In theend, [candidate]wontheU.S.Presidential election.Observersbegan debating whether [country]’s statements during thecampaign might have affected the results of the election.

Country was assigned to be China, Pakistan, or Tur-key26 and candidate was either “the Democratic can-didate” or “theRepublican candidate.”27 In our survey,country and candidatewere randomized independently,resulting in 3 3 2 5 6 variations.

Members of the threat group received the followingvignette, which not only included an endorsement butalso implied that victory by the opponent might have

23 E.g., “McCain:RussianCyberintrusions an ‘Act ofWar’”December30, 2016. CNN. https://www.cnn.com/2016/12/30/politics/mccain-cyber-hearing/index.html; Goodman 2017; Hathaway et al. 2012.24 The latter prediction relates toBush andPrather (2020), who foundthat citizens in Tunisia and the United States were less willing toengage economically with foreign countries that sided against them indomestic politics.

25 Our scenarios implied no ambiguity about whether a foreigncountry intervened. Future experiments could study how citizensrespond when unsure whether an intervention took place.26 We selected countries that were technologically advanced enoughto interveneandhadplausiblemotives forgetting involved.Theonlineappendix shows that the specific countrymentionedhad little effect onpublic perceptions of intervention. Future research could examinehow the American public would respond to interference by othercountries, both friendly and unfriendly.27 In this scenario (and in the threat and operation scenarios, de-scribed below), the candidate that received support from the foreigncountryultimatelywon theelection.Futureexperiments could includescenarios in which the foreign-supported candidate lost the election.


6

https://www.cnn.com/2016/12/30/politics/mccain-cyber-hearing/index.htmlhttps://www.cnn.com/2016/12/30/politics/mccain-cyber-hearing/index.htmlalyssHighlightplease remove the link from the parentheses

consequences for economic and military relations withthe United States:

In 2024, the government of [country] made several publicstatements during the U.S. Presidential election campaign.[Country] said that it strongly preferred [candidate] andhoped [candidate] would win the U.S. Presidential election.[Country] said that, if [opponent] won, it would rethink itseconomic andmilitary relationshipswith theU.S. In the end,[candidate] won the U.S. Presidential election. Observersbegan debating whether [country]’s statements during thecampaign might have affected the results of the election.

We independently randomized country and candidate(leaving the other politician as the opponent), yielding 33 2 5 6 variations.

The operation scenario involved either giving moneyto support a campaign, spreading true or false in-formation, or hacking into voting machines. Subjectsread the following text:

In 2024, a foreign country developed a plan to influence theU.S. Presidential election. There was a [percent] chance thatthe foreign country was [country]. The plan was designed tohelp [candidate] and hurt [opponent]. According to the plan,agents fromtheforeigncountrywould [typeofoperation].Theforeign country carried out its plan to help [candidate] andhurt [opponent]. In the end, [candidate] won the U.S. Presi-dential election. Authorities began investigating whether theforeigncountrymighthaveaffected the resultsof theelection.

Country and candidate (and, by implication, opponent)were randomized as described earlier. Recognizing thatcitizensmightnotbesurewhichforeigncountrycarriedoutthe intervention, we randomized percent to be 50%, 75%,95%, or 100%. Finally, we randomized the type of oper-ation. The foreign country’s agents would give money(“give$50million to support thecampaignof [candidate]”),spread truth (“use socialmedia to spread embarrassing buttrue information about [opponent]—accurately revealingthat [opponent] had broken laws and acted immorally”),spread lies (“use social media to spread embarrassing liesabout [opponent]—falsely claiming that [opponent] hadbroken laws and acted immorally”), or hack machines(“hack into voting machines and change the official votecount to give [candidate] extra votes”). Overall, the oper-ation scenario included 33 25 6 combinations of country

and candidate, crossed with 4 3 4 5 16 combinations ofpercent and type of operation. Below, we simplify the ex-positionand increase statisticalpowerbyanalyzing someofthese variations while averaging over the others.

Finally, in the stay out story, the named country nevercarried out an intervention. The text appears below,with randomized components in italics:

In 2024, there was a false rumor that [country] had de-velopedaplan to influence theU.S. Presidential election. Infact, [country] never had such a plan. The election pro-ceeded without any involvement by [country], and [can-didate] won the U.S. Presidential election.

As in the other scenarios, country was China, Pakistan,or Turkey, and candidate was either the Democraticcandidate or the Republican candidate. Thus, the stayout group involved 3 3 2 5 6 variations. Table 1summarizes all experimental treatments.28

After assigning each participant to one group andpresenting them with one scenario, we measuredopinions about three topics.29 First, would they approveor disapprove of how the foreign country behaved?Second, howwould suchevents affect their confidence inU.S. elections and American democracy more gener-ally? Finally, what foreign policies would they supportwith respect to the country in the scenario?Weorganizetheremainderof thepaperaroundthese threequestions.

FINDINGS

Public Disapproval of ForeignElectoral Intervention

We first investigate when the public would condemn orcondone foreign electoral intervention. We predictedthat Americans would object most to operations and

TABLE 1. Experimental Treatments

Endorsement group Threat group Operation group Stay out group

Country China, Pakistan,Turkey

China, Pakistan,Turkey

China, Pakistan, Turkey China, Pakistan,Turkey

Candidate Democratic,Republican

Democratic,Republican

Democratic, Republican Democratic,Republican

Percent N/A N/A 50, 75, 95, 100 N/A

Type ofoperation

N/A N/A Give money, spread truth, spread lies,hack machines

N/A

28 When presenting the scenarios, we asked a series of post-treatmentattention checks. The analyses in this article include all 3,510 peoplewhotookoursurvey, regardlessofattentiveness.Thefindings reportedhere should therefore be interpreted as “intention-to-treat” analyses.The online appendix shows that conclusions were similar when weexcluded subjects whoanswered less than 80%of the checks correctly.29 Weallocated fewer respondents to the stayout andoperationconditionsthan to the endorsement and threat conditions, given our ex ante expect-ations about the number of respondents needed to detect causal effects.


7

least to endorsements, and would express greater dis-approval of foreign efforts to help political opponentsthan of foreign assistance to their own party.

To evaluate these predictions, we asked voterswhether they approved or disapproved of how theforeign country behaved. There were five responseoptions: approve strongly, approve somewhat, neitherapprove nor disapprove, disapprove somewhat, anddisapprove strongly. For simplicity, we focus on a nat-ural and easily interpretable quantity of interest, thepercentage of respondents who disapproved, but theonline appendix documents that our conclusions holdwhen we analyze the full five-point scale, as well.

Figure 1 shows the average rate of disapproval in eachof our intervention treatment groups: endorsement,threat, and operations. The estimates in Figure 1 in-tegrate over the other experimental conditions in Table1, and therefore reflect average levels of disapprovalregardless of the foreign countrywementioned, the levelof certainty about the perpetrator, and the party of thecandidate who was favored.30 In this figure, the dotsrepresent point estimates, and the horizontal lines are95% confidence intervals.

When the foreign country endorsed a candidate, 37%of our subjects disapproved, even though the foreigncountry did no more than express its preferences. Aspredicted, disapproval was higher (55%) when theforeign countrynot only expressedapreferencebut alsothreatened to downgrade relations with the UnitedStates if its favorite candidate lost. These findings werenot preordained. One might think that foreign coun-tries, like domestic political actors, would be entitled tovoice their opinions and engage with some partnersmore than others. However, many who read the en-dorsement scenario and most who read the threatscenario reacted with disdain.

As expected, disapproval was highest (77%) whenthe country implemented operations to bolster itsfavored candidate. The bottom portion of Figure 1disaggregates the four types of operations in our ex-periment. Approximately 72% of respondents dis-approved when the foreign country spreadembarrassing but true information about a candidate.Reactions were even more negative when the foreigncountry spread lies, gave money for campaigning, orhacked into voting machines. In those situations,disapproval hovered between 78% and 79%.31

The bottom portion of Figure 1 also suggests somesurprising conclusions. Citizens apparently did not drawa sharp distinction between spreading truth andspreading lies. In the “spreading truth” scenario, agentsfrom the foreign country used social media to spreadembarrassing but true information about one of the

candidates, accurately revealing that the candidate hadbroken laws and acted immorally. One might think thatsome Americans would welcome, or at least tolerate,information about actual improprieties by a U.S. presi-dential candidate. Instead, 72% of respondents dis-approved when the foreign country disseminated trueinformation. A higher proportion, 79%, disapproved ofspreading lies, but thedifference in reactions to these twotreatments was only 7 percentage points.

Moreover, respondents reacted just as negatively toforeign campaign contributions as to spreading lies orhacking electionmachines.Onemight expect citizens toview campaign contributions as more legitimate thanfalsely scandalizing a candidate or rigging the electoraltally. On the contrary, Americans in our study viewedforeign money as no less objectionable than disinfor-mation and cheating.

We also investigated how partisanship moderatedreactions to foreign interference.In our sample, 36% ofrespondents were Democrats, 29% were Republicans,and the remainder were Independents who did notidentify with either party. For each of these sets ofrespondents, we measured the percentage who dis-approved of each mode of intervention, while holdingthe winner of the election constant.

Figure 2 summarizes how Republicans, Independents,andDemocratsreactedtoeachtypeofforeignintervention.Consider how Republicans responded to foreignendorsements, shown in the top left corner of the figure. Inour experiment, 50% of Republicans disapproved whenthe foreign country endorsed the Democrat (CountryFavored D), whereas only 22% of Republicans dis-approved when the foreign country endorsed the Re-publican (Country Favored R). This example fits ourhypothesis thatAmericans reactmorenegatively to foreignintervention on behalf of the opposition, than to otherwiseequivalent intervention in support of their own party.

Reactions among Democratic voters were similarlypartisan. As the lower left corner of Figure 2 shows,disapproval among Democrats was 53 percentagepointswhen the foreigncountryendorsedaRepublican,

FIGURE 1. Disapproval of Foreign ElectoralIntervention

Note: The figure shows the percentage of respondents whodisapproved. Sample sizes were 1,202 for endorsement, 996 forthreat, and 751 for operations. Sample sizes for the four types ofoperations (spread truth, spread lies, give money, and hackmachines) were 181, 209, 180, and 181, respectively.

30 The online appendix shows that the public reacted similarlywhether the country was China, Pakistan, or Turkey.31 The online appendix probes why respondents objected moststrongly to operations, followed by threats and then endorsements.Respondents perceived that operations were most effective in givingthe favored candidate an advantage, followed by threats, and thenendorsements. Furthermore, all else equal, respondents disapprovedmuchmore stronglyof interventions theydeemed tobeconsequential.


8

versus 28 percentage points when the foreign countryendorsed a Democrat.32 These examples illustratea remarkable double standard that arose throughoutour experiment. As expected, the reactions of In-dependent voters were far less sensitive to which partythe foreign country endorsed.33

We found similar patterns when the foreign countrythreatened that the outcome could affect future eco-nomic and military relations with the United States(middle columnof Figure 2).AmongRepublicans, 71%disapproved when the country sided with the Demo-cratic candidate, versus only 51% when the countrysided with the Republican candidate. Democrats alsoresponded in a partisan fashion; their disapproval was71% when the foreign country backed the Republican,compared with 39% when the foreign country backedthe Democrat. As expected, the effect of threats onIndependent voters did not depend on which side theforeign country took.

The right side of Figure 2 shows the results when theforeign country undertook operations. Here, we foundstronger bipartisan pushback: strong majorities ofrespondents disapproved of operations, includingoperations designed to help their own party. Nonethe-less, partisan double standards persisted. AmongRepublicans, disapproval was 87% when foreign oper-ations aimed to help the Democrat, compared with 67%when foreign operations aimed to help the Republican.

Likewise, Democrats reacted more negatively (87%)when the foreign country sidedwith theRepublican thanwhen the foreign country sided with the Democrat(72%). Once again, the reactions of Independents wereless sensitive to which side the foreign country took.34

Figure 2 also shows how foreign electoral in-terference could split the American electorate. Forexample, when the foreign country issued threats onbehalf of a Democratic candidate, most Republicans(71%) disapproved, but only a minority of Democrats(39%) objected. Likewise, when a foreign countryprivileged the Republican in this way, the vast majorityof Democrats (71%) protested, but only half (51%) ofRepublicans balked. These results illustrate how for-eign interference sows domestic divisions.

It isworth considering howour experimental designmight have affected the magnitude of the partisandifferencesweuncovered.Our scenarios conveyednoambiguity about whether the foreign country tam-pered with the U.S. election. On the one hand, theendorsements and threats we studied are by naturepublic, affording little opportunity for observers todisagree about what the foreign country said (Levin2016, 193). On the other hand, countries typically tryto keep operations secret, creating uncertainty—atleast initially—about what the foreign country did.35

For example, after the 2016 U.S. election, some Re-publican elites, including President Donald Trump,

FIGURE 2. Disapproval of Foreign Electoral Intervention, by Partisanship

Note: The figure gives a partisan breakdown of the percentage of respondents who disapproved of the foreign country’s behavior. For eachtype of intervention, the figure shows how Republicans, Independents, and Democrats reacted when the foreign country favored theRepublican candidate (Country FavoredR), andwhen the foreign country favored theDemocratic candidate (Country FavoredD). For eachof the six combinations of partisanship on the vertical axes, there were between 155 and 233 observations in the endorsement group,144–193 observations in the threat group, and 99–148 observations in the operations group.

32 Corstange andMarinov (2012) found a partisan reaction in a muchmore fragile democracy, Lebanon.33 The online appendix sheds light on one reason why these patternsmight have arisen. As predicted, respondents thought that inter-ventionsonbehalfof theopposingpartyweremoreconsequential thaninterventions to help their own side. Furthermore, they respondedmore negatively to interventions they perceived as consequential,particularly when the consequences favored the opposing party.

34 We found no consistent differences in partisan bias amongrespondents with more vs. less education. See Online Appendix.35 Ambiguity dissipates over time, however. As Levin (2019b, 13 ofOnline Appendix A) explains, acting secretly “usually enabled theintervener to hide such covert components from the voters in thetarget prior to the elections,” but interveners were “usually unable/unwilling to hide it from later investigations/scholars etc. in the yearsfollowing….”


9

dismissed Russian operations as a hoax that Demo-crats had fabricated to rationalizewhy their candidatelost. If we conducted experiments in which elitesdebated or questioned whether operations occurred,would we find even sharper partisan differences? Weleave this as a topic for future research.

One might also ask whether the events of 2016 mighthave affected average disapproval in our scenarios. Ontheonehand,mainstreamnewsmedia tended toportrayRussia’s behavior in a negative light, potentially am-plifying the disapproval we observed in our 2018experiments. On the other hand, Russia’s interferencein 2016 spurred Americans to develop reasoned opin-ions about foreign involvement in U.S. elections, po-tentially increasing the external validity of ourexperiments. Future research could evaluate theseconjectures.

Finally, did Russian interference in 2016 affect ourfindings about partisanship? After 2016, DemocratsandRepublicanswere exposed to different portrayalsof Russian behavior. Democratic leaders andmuch ofthe newsmedia tended to depict Russian interferenceas an unacceptable and consequential attack on theUnited States. By contrast, Republican opinionleaders often portrayed Russian meddling as fabri-cated or inconsequential. Our study suggests that,despite the markedly different messages Democratsand Republicans may have internalized about 2016,voters from both parties apply similarly biasedstandards when judging hypothetical future inter-ventions. Future research could evaluate whetherthese partisan portrayals influenced reactions to ourscenarios in subtle ways and,more generally, whetherpast experiences of intervention shape futurereactions.

Foreign Electoral Intervention and Attitudesabout Democracy

We next consider how foreign electoral interventionaffected attitudes about democracy. We hypothesizedthat intervention would undermine trust in the electionresults, erode faith in democratic institutions, and de-press future political participation.We further expectedthat operations would inflict more damage than threats,threats would cause more damage than endorsements,and voters would react more negatively if the foreigncountry sided with the opposition.

To gauge attitudes about American democracy, weasked respondents in each of the four treatmentgroups (stay out, endorsement, threat, and oper-ations): “If the 2024 election happened just as wedescribed, would you agree or disagree with the fol-lowing statements?” The three statements were, “Iwould trust the results of the election,” “I would beunlikely to vote in future elections,” and “I would losefaith in American democracy.” We calculated thepercentage of respondents who agreed or disagreedwith each statement.

We found that foreign intervention greatly increaseddistrust in the results of the election (first graph in Figure3). When the foreign country stayed out, 25% voiced

distrust, reflectingpreexistingcynicismabout the integrityof American elections. But distrust increased to 38%when the foreign country offered an endorsement, 42%when the country coupled the endorsementwith a threat,and 71% when the country carried out operations.36

Foreign interventionnotonly soweddoubts about thecurrent election but also eroded faith inU.S. democracy(second graph in Figure 3). Although 23% lost faitheven when the foreign country refrained from

FIGURE 3. Attitudes about Democracy, byMode of Foreign Electoral Intervention

Note:The figure shows, by treatment condition, the percentage ofrespondents who said they would distrust the results of theelection (top panel), lose faith in American democracy (middlepanel), or avoid voting in the future (bottom panel). Sample sizeswere 561 for stay out, 1,202 for endorsement, 996 for threat, and751 foroperations, comprisedof181,209,180,and181 forspreadtruth, spread lies, give money, and hack machines, respectively.

36 Figure 3 presents levels of distrust and other attitudes about de-mocracy. The online appendix contains complementary figures thatre-express the values in Figure 3 as intention-to-treat effects, calcu-lated as attitudes when the foreign government intervened, minusattitudes when the foreign government stayed out.


10

intervening, that figure increased to 29% when theforeign country endorsed one of the candidates, 36%when the endorsement came with a threat, and 49%when the country undertook concrete operations.

Finally, foreign intervention modestly depressedfuture intentions to vote (third graph in Figure 3).When the foreign power stayed out of the election,20% of subjects said they would abstain from voting infuture elections. Avoidance of voting rose to 21%when the foreign countrymerely expressed its opinion,24%when the country put future relations on the line,and 29% when the foreign country carried outoperations.

Overall, these findings suggest that foreign in-volvement could have profoundly negative effects onAmerican democracy. Interference in a presidentialelection would increase distrust in the immediateresults, while also affecting—to a lesser degree—moredistant outcomes such as faith in American democracyand participation in future elections. Moreover, ourexperiments indicate that foreign countries can ad-versely affect the American psyche through wordsalone. By verbally endorsing a Presidential candidate,with or without threats, foreign powers have the po-tential to undermine confidence in the American po-litical system.

We hypothesized that intervention would not onlybe corrosive on average but also prompt differentreactions depending on partisanship. Figure 4 showsthe effect of intervention on attitudes about de-mocracy among Republicans, Independents, andDemocrats. We calculated intention-to-treat effectsby taking attitudes about democracy in each treatmentcondition and subtracting attitudes in the baselinecondition, when the country stayed out but the samecandidate won.37 The black dots are average in-tention-to-treat effects, and the thin lines are 95%confidence intervals.

Figure 4 shows evidence of a partisan doublestandard. Consider, for example, Democrats’ reac-tions to foreign endorsements, shown on the left-handside of each of the three panels. When the foreigncountry endorsed the Republican (Country FavoredR), Democrats became substantially more likely todistrust the results (23-point effect), lose faith indemocracy (19-point effect), and avoid voting in thefuture (10-point effect). But when the foreign countryendorsed the Democrat (Country Favored D),Democrats were typically indifferent, exhibiting nostatistically significant changes in attitudes aboutdemocracy.

Republicans displayed a double standard, as well.Republican distrust of the outcome and skepticismabout democracy swelled to a much greater degreewhen the foreign country endorsed a Democrat thanwhen the foreign country endorsed a Republican. (Bycontrast, endorsements did not elicit a double stan-dard in Republican willingness to vote.) Finally,

endorsements caused Independents to sour on de-mocracy, but not to the same degree as citizens whowitnessed interference against their own party. For-eign threats produced similar patterns, as shown in themiddle column of Figure 4.

The right side of Figure 4 presents intention-to-treateffects when the foreign country carried out operations.On the one hand, all estimates were positive, and mostwere significantly distinguishable from zero. Thus,operations undermined the democratic ethos evenamong citizens the intervention was designed to help.On the other hand, voters tended to be more negativeabout American democracy when the foreign countryfavored the opposing side.

In sum, revelations of evenmodest forms of electoralintervention undermined confidence in American de-mocracy while also pitting Republicans and Democratsagainst each other. Thus, foreign interference providesforeign countries with a potent weapon for weakeningthe United States.

Foreign Electoral Intervention and ForeignPolicy Preferences

Finally, we investigated public support for retaliationagainst foreign interference in U.S. elections. Weexpected higher public support for nonmilitary options,such as diplomatic or economic sanctions, than formilitary responses such as threatening or initiatingarmed conflict. We further predicted that support forboth kinds of retaliation would be strongest for oper-ations and weakest for endorsements, and would behigher when the foreign country aided the opposingparty. Finally, we expected that the desire to retaliateagainst operations would rise with certainty aboutwhich country was responsible.

To measure support for retaliation, we askedrespondents whether they would support or opposeeach of the following policies if the 2024 electionhappened just as we described: cutting off diplomaticrelations with [country], imposing economic sanctionson [country], threatening to use military force against[country], and launching a military strike against[country]. Figure 5 shows the percentage ofrespondents who supported each policy option, con-ditional on whether and how the foreign country in-tervened.38 Each dot represents the mean level ofsupport, averaging over the other features in the ex-periment. (Later, we test whether these conclusionsdepended on certainty about the identity of the for-eign country.)

The figure reveals several conclusions. First, citizensresoundingly rejected military responses to foreignelectoral intervention. Even in the face of operations tofund candidates, manipulate information, or hack intovoting machines, only 28% wanted to make militarythreats, andonly19%called formilitary strikes. Second,diplomatic and economic sanctions received majority

37 For reference, the online appendix displays the attitudes aboutdemocracy that we used to compute intention-to-treat effects.

38 TheOnlineAppendix shows that responseswere similar regardlessof which foreign country we named and the type of intervention itundertook.


11

support only when the foreign country conductedoperations. This is surprising, given that verbal state-ments triggered public ire and undermined confidencein democracy. It appears that foreign countries couldundertake destructive verbal interventions (with orwithout threats), secure in the knowledge that none ofthe retaliatory measures we studied would attractsupport from a majority of the American public. Fur-thermore, adversaries could carry out electoral attacks

without running the risk that the public would demandmilitary retaliation.39

We also hypothesized that support for retaliationwould depend on partisanship. Figure 6 tests this

FIGURE 4. Effects of Foreign Electoral Intervention on Attitudes about Democracy, By Partisanship

Note: The figure gives a partisan breakdown of intention-to-treat effects of foreign electoral intervention on the percentage of respondentswhosaid theywoulddistrust the resultsof theelection (toppanel), lose faith inAmericandemocracy(middlepanel),oravoidvoting in the future(bottom panel). In all cases, the effect is the change caused by intervening instead of staying out. For each of the six combinations ofpartisanship on the vertical axes, there were between 83–109 observations in the stay out (control) condition. Sample sizes for theintervention conditions were as in Figure 2.

39 This could have important consequences for government policy.Researchers have found that leaders respond to public opinion aboutforeign policy (Tomz,Weeks, and Yarhi-Milo 2020), and rarely wagewar without public support (Reiter and Stam 2002).


12

hypothesis by splitting the sample into Republicans,Independents, and Democrats. Within each group, thefigure shows how each type of foreign interventionaffected support for tough policies (diplomatic sanc-tions, economic sanctions, military threats, or militarystrikes), relative to baseline support for those samepolicies when the foreign country stayed out entirely.The dots in Figure 6 are average intention-to-treateffects, integrating over the other dimensions of theexperiment.

Once again, we found evidence of a partisan doublestandard. For nearly every combination of mode ofintervention and method of retaliation, Republicansreacted more strongly to “Country Favored D” sce-narios than to “Country FavoredR” scenarios, whereasDemocrats did the opposite. Although the differenceswere almost always in the expected direction, only a fewwere statistically significant at conventional levels.40

These patterns suggest what kinds of retaliatorypolicies might—or might not—be politically feasibleafter an electoral intervention. According to Figure 6,electoral interventions may be less likely to spur re-taliatory sentiment by members of the winning partythan by members of the losing party. This means, forexample, that if a Republican candidate rode to victoryin the context of pro-Republican interference, Demo-crats might demand retaliation, but members of thenewly elected president’s own party would be lesswilling to go along. Knowing this, foreign countriesmight feel even more confident that they could in-tervene with relative impunity.

Finally, we conjectured that support for hostile for-eign policies would increase with the level of certaintyabout which country was culpable. To test this possi-bility, we compared support for retaliation whenrespondents were 50%, 75%, 95%, or 100% certain ofthe identity of the country that carried out operations.(Recall, fromTable 1, thatwedid not raise doubts aboutthe identity of the country in the endorsement andthreat conditions, because such expressions are bydefinition overt, leaving no ambiguity about who madethe statement.)

We found some evidence for this hypothesis, but lessthan expected. Figure 7 shows how support for eachpolicy varied by the level of certainty, averaging overthe other dimensions of our experiment. To our sur-prise, citizenswerenearlyas likely to supportdiplomaticand economic sanctions when they were only 50% sureabout the identity of the perpetrator, as when theywerecompletely certain that the named country had con-ducted the operation. Support for nonmilitary re-taliation rose steadily with certainty, but the differencesbetween 50%and 100%certaintywere relatively small:66–545 12 points for diplomatic sanctions, and a 73–685 5 points for economic sanctions. The patterns formilitary threats were similar, albeit with lower baselinelevels of support. Finally, certainty had no appreciableeffect on support for military strikes.

Overall, our findings about uncertainty have a sur-prising political implication: although investigationsinto electoral intervention might increase clarity aboutthe identity of the perpetrator, the accumulation ofevidence may not result in substantially higher publicsupport for international retaliation.

In summary, our data revealed an American re-luctance to retaliate against even themostobjectionableforms of foreign intervention in U.S. elections. Amer-icans overwhelmingly rejected military responses toforeign interference, even when a foreign country un-dertook operations such as funding candidates, ma-nipulating information, or hacking into U.S. votingmachines, and even when the identity of the foreign

FIGURE 5. Support for Foreign Policies, ByMode of Foreign Electoral Intervention

Note: Each panel in the figure shows the percentage ofrespondents who supported a given foreign policy, by treatmentcondition. Sample sizes were as in Figure 3.

40 As the online appendix shows, the double standards became largerand more statistically significant when we excluded inattentiverespondents from the sample.


13

FIGURE 6. Effects of Foreign Electoral Intervention on Support for Foreign Policies, by Partisanship

Note: Each panel of the figure gives a partisan breakdown of the effect of foreign electoral intervention on support for a given foreign policy.Sample sizes were as in Figure 4.


14

attacker was known with certainty. Voters were moresupportive of nonmilitary responses such as diplomaticand economic sanctions, but majorities endorsed thesemeasures onlywhen the foreign country had engaged inoperations. Thus, countries that aspire to interfere inU.S. elections might take comfort in knowing that theiractions, however detrimental to American democracy,might not provoke unified support for retaliation.

CONCLUSION

Despite the growing importance of election in-terference for contemporary politics (Bubeck andMarinov 2017, 2019; Levin 2016, 2019b), we knowrelatively little about how Americans judge foreignmeddling in U.S. elections. In this article, we usedsurvey experiments to investigate three fundamentalquestions about how Americans would respond torevelations of foreign electoral intervention.

First, when would U.S. citizens tolerate foreign in-volvement in American elections, instead of con-demning external efforts to tip the scales? In ourexperiments, American tolerance of intervention wasconditional on the intended beneficiary. Both Demo-crats and Republicans exhibited a clear double stan-dard, disapproving more strongly of foreign efforts tohelp the opposition thanof otherwise identical efforts tohelp a candidate from their own party.

The polarizing effects of foreign electoral in-tervention are, therefore, more widespread than pre-viously appreciated. In a seminal article, Corstange andMarinov (2012) found that when foreign countries tooksides in the Lebanese election of 2009, the domesticpublic split along partisan lines. Corstange andMarinovexpected this type of reaction in “fragile and un-consolidated” democracies such as Lebanon, butspeculated that sucha responsemight notbeplausible in“consolidated democracies,” where they anticipatedstronger bipartisan pushback. On the one hand, con-sistent with this prediction, our experiments in theUnited States showed that bipartisan majorities dis-approved of foreign operations to influence U.S. elec-tions.On theotherhand, theAmericanpublic reacted ina polarizedway to all three types of foreign interferencewe studied. Hence, the divisive effects of electoral in-tervention are not limited to fragile democracies; theyarise in highly established democracies, as well.

Our findings about partisan polarization shed newlight on past interventions, while also portending sharppolitical cleavages in the future. Following the 2016election, Republicans were far more likely than Dem-ocrats to deny that Russia intervened, or to acknowl-edge Russian meddling but dismiss it asinconsequential. Our experiments suggest that suchreactions arose not fromprincipleddifferences betweenthe parties, but instead from a pervasive tendency toapply politically biased standards when judging thebehavior of other countries. Having randomized whichparty the foreign country favored, we found that bothDemocrats and Republicans were far more willing totolerate foreign support for their own party than for the

opposition. Partisan differences were smaller when thecountry carried out operations such as spreading liesabout candidates or hacking into voting machines, butpartisan bias persisted even then. Thus, if a foreigncountry took the Democratic side in a future election,the political reaction might be the reverse of 2016, withRepublicans denouncing andDemocrats condoning theforeign interference.

Our experiments further showed that public toler-ance depended not only on the intended beneficiary butalso on the typeof intervention.Althoughmany citizensdenounced foreign endorsements, they were more

FIGURE 7. Support for Foreign Policies, byCertainty about the Identity of the ForeignCountry

Note: Each panel of the figure shows the percentage ofrespondents who supported a given foreign policy, for each of thefour levels of certainty about the identity of the foreign country.Samplesizeswere194 for50%chance,180 for75%,200 for95%,and 177 for 100%.


15

likely to condemn threats, and they objected most tooperations such as donating money to a campaign,disseminating embarrassing information about a can-didate, or hacking into voting machines. These findingsunderscore the importance of distinguishing differenttypes of electoral interventions, to see how Americanswould respond to the wide variety of tools countriescould employ in U.S. elections.

We also used experiments to investigate a secondfundamental question: when would news of foreignelectoral intervention undermine confidence in demo-cratic institutions? In our experiments, voters wholearned of foreign intervention were substantiallymorelikely to distrust the results of the election, lose faith inAmerican democracy, and abstain from participating infuture elections. Reactions were not symmetric acrossthe population, however. Instead, foreign interferenceled to partisan splits about the state of U.S. democracy.Thus, foreign involvement can have profoundly nega-tive and divisive effects on confidence in Americandemocracy.41

Finally, we studied how election interferencechanged public attitudes about foreign relations. Re-gardless of the form of electoral intervention, citizens inour experiments rejected military threats and militarystrikes. Although many prominent observers charac-terized theRussian intervention of 2016 as anact ofwar,the electoral equivalent of 9/11, and a direct attack oninstitutions at the heart of American democracy, ourexperiments suggested that even operations such ashacking the vote tally would not spur the Americanpublic to resort to military force.

We did find majority support for diplomatic andeconomic sanctions, but only when the foreign gov-ernment interfered thorough operations such as fund-ing parties, manipulating information, or hacking intovoting equipment. Foreign endorsements and threats,on the other hand, failed to generate majority supportfor any reprisals. Finally, consistent with our findingsabout a partisan double standard, both Democrats andRepublicans were less willing to retaliate when theforeign power favored their own party than when theforeign power backed the opposition. These patternssuggest that foreign countries could interfere inAmerican elections without triggering bipartisan publicdemand for tough retaliation.

Future studies could extend our approach in a severalways. One could, for example, introduce ambiguityabout whether foreign countries meddled in elections.Our experiments asked howAmericans would respondto clear revelations of foreign intervention. To findout, we presented the existence or absence of foreignintervention as undisputed, while allowing some un-certainty about the identity of the perpetrator. Follow-up experiments could present inconclusive evidenceabout whether a foreign country intervened, describe

partisan disagreement about whether external in-tervention took place, or both. One could comparereactions in those scenarios with what we found in ourexperiments.42

Future experiments could also vary other features ofthe intervention. We focused on partisan interventionsin which the foreign country promoted one candidate atthe expense of another. Foreign countries could insteadintervene evenhandedly by undermining or supportingboth sides in the contest. Using experiments, one couldcompare public tolerance for partisan versus even-handed interventions, assess how the two types ofincursions would affect confidence in democracy andwillingness to retaliate, and study whether Democratsand Republicans would unite against foreign meddlingthat did not try to give one side an advantage. It wouldalso be instructive to conduct experiments about mul-tipronged interventions that not only take sides but alsoattempt to modify electoral procedures (Bubeck andMarinov 2017, 2019).

In addition, future research could explore the effectsof election interference in other countries. For example,we found that Americans opposed retaliation, despiteliving inoneof themost powerful countries in theworld.Citizens inweaker countrieswould presumably be evenless willing to retaliate, especially against global or re-gional superpowers. Similarly, future studies couldexplore whether the divisive effects of foreign electoralintervention depend on the political context. We foundthat foreign interference split the American publicalong partisan lines. How would citizens react incountries with stronger (or weaker) levels of partisanidentification, or in countries where parties are pro-grammatically more (or less) distinct than the Demo-cratic and Republican parties in the United States?Would reactions depend on the age of the democracy orthe nature of the electoral system? This article couldserve as a blueprint for follow-up experiments to assessresponses toelection interferencenotonly in theUnitedStates but also in other democracies.

Moreover, future research could study how theidentity of the intervening country shapes public per-ceptions. In our experiment, the name of the countrymentioned had little effect on disapproval, faith indemocracy, and support for retaliation. Future researchcould examine how American and foreign publicswould respond to interference by other countries, bothfriendly and unfriendly.

In the meantime, our findings suggest that foreignelectoral intervention represents a significant threat todemocracies such as the United States. Previous re-search has shown that foreign powers can use electoralintervention to boost the chances of their favoredcandidate or party (Levin 2016). Our experiments add

41 Our findings also could provide a new mechanism explaining howforeign interventions could weaken democracy (Levin 2019a): ifenough voters react to news of foreign intervention with diminishedfaith in democracy and lowered levels of political participation, thiscould weaken democratic norms and practices.

42 Uncertainty could affect not only levels of party polarization butalso aggregate levels of public tolerance. Shulman and Bloom (2012)hypothesized that highly salient (“transparent” and “intelligible”)interventionswouldprovokemorepublic resentment thanambiguousinterventions. One could test this hypothesis experimentally byvarying salience while holding the partisanship of the interventionconstant.


16

that electoral intervention can polarize the electorateand diminish faith in democratic institutions withoutprovoking the kind of public demand for retaliationprompted by conventional military attacks.

SUPPLEMENTARY MATERIAL

To view supplementary material for this article, pleasevisit https://doi.org/10.1017/S0003055420000064.

Replication materials can be found on Dataverse at:https://doi.org/10.7910/DVN/E3BAO5.

REFERENCES

Ahlquist, John S., Nahomi Ichino, Jason Wittenberg, and DanielZiblatt. 2018.“HowDoVotersPerceiveChanges to theRulesof theGame? Evidence from the 2014 Hungarian Elections.” Journal ofComparative Economics 46 (4): 906–19.

Anderson, Christopher J., André Blais, Shaun Bowler, ToddDonovan, and Ola Listhaug. 2005. Losers’ Consent: Elections andDemocratic Legitimacy. New York: Oxford University Press.

Anderson, Christopher J., and Christine A. Guillory. 1997. “PoliticalInstitutions and Satisfaction with Democracy: A Cross-NationalAnalysis of Consensus and Majoritarian Systems.” American Po-litical Science Review 91 (1): 66–81.

Anduiza,Eva,AinaGallego, andJordiMuñoz. 2013.“TurningaBlindEye: Experimental Evidence of Partisan Bias in Attitudes TowardCorruption.” Comparative Political Studies 46 (12): 1664–92.

Beaulieu, Emily. 2014. “From Voter ID to Party ID: How PoliticalParties Affect Perceptions of Election Fraud in the U.S.” ElectoralStudies 35: 24–32.

Birch, Sarah. 2011. Electoral Malpractice. New York: Oxford Uni-versity Press.

Bolsen, Toby, JamesN.Druckman, andFayLomaxCook. 2014. “TheInfluence of Partisan Motivated Reasoning on Public Opinion.”Political Behavior 36 (2): 235–62.

Brancati,Dawn. 2014.“BuildingConfidence inElections:TheCase ofElectoral Monitors in Kosova.” Journal of Experimental PoliticalScience 1 (1): 6–15.

Bubeck, Johannes, and Nikolay Marinov. 2017. “Process or Candi-date: The International Community and the Demand for ElectoralIntegrity.” American Political Science Review 111 (3): 535–54.

Bubeck, Johannes, and Nikolay Marinov. 2019. Rules and Allies:Foreign Election Interventions. New York: Cambridge UniversityPress.

Bush,SarahSunn,andLaurenPrather.2017.“ThePromiseandLimitsof Election Observers in Building Election Credibility.” TheJournal of Politics 78 (3): 921–35.

Bush, SarahSunn, andLaurenPrather. 2018.“Who’s There?ElectionObserver Identity and the Local Credibility of Elections.” In-ternational Organization 72 (3): 659–92.

Bush, Sarah Sunn, and Lauren Prather. 2019. From Monitoring toMeddling: How Foreign Actors Influence Local Trust in Elections.New Haven, GT: Yale University Manuscript.

Bush, Sarah Sunn, and Lauren Prather. 2020. “Political Side-Takingby Outside Powers and Mass Attitudes Towards InternationalEconomic Engagement.” International Organization 74 (3).

Campbell, Angus, Philip E. Converse, Warren E. Miller, and DonaldE. Stokes. 1960. The American Voter. Chicago: University ofChicago Press.

Chanley, Virginia, Thomas J. Rudolph, and Wendy M. Rahn. 2000.“TheOrigins and Consequences of Public Trust in Government: ATime Series Analysis.” Public Opinion Quarterly 64 (3): 239–56.

Corstange, Daniel, and Nikolay Marinov. 2012. “Taking Sides inOther People’s Elections: The Polarizing Effect of Foreign In-tervention.” American Journal of Political Science 56 (3): 655–70.

Dahl, Robert A. 1971. Polyarchy: Participation and Opposition. NewHaven, CT: Yale University Press.

Delavande, Adeline, and Charles F. Manski. 2012. “CandidatePreferences and Expectations of Election Outcomes.” Proceedingsof the National Academy of Sciences 109 (10): 3711–15.

Drutman, Lee, Larry Diamond, and Joe Goldman. 2018. Follow theLeader: Exploring American Support for Democracy and Author-itarianism.Washington,DC:Democracy FundVoter StudyGroup.https://www.voterstudygroup.org/publication/follow-the-leader.

Esaiasson, Peter. 2011. “Electoral Losers Revisited—How CitizensReact to Defeat at the Ballot Box.” Electoral Studies 30 (1):102–13.

Goodman, Ryan. 2017. “International Law and the US Response toRussian Election Interference.” https://www.justsecurity.org/35999/international-law-response-russian-election-interference/.

Hathaway, Oona A., Rebecca Crootof, Philip Levitz, Haley Nix,AileenNowlan,WilliamPerdue, and Julia Spiegel. 2012. “TheLawof Cyber-Attack.” California Law Review 100 (4): 817–85.

Hetherington, Marc. 2004. Why Trust Matters. Princeton, NJ:Princeton University Press.

Hyde, Susan. 2011. The Pseudo-Democrat’s Dilemma: Why ElectionObservation Became an International Norm. Ithaca, NY: CornellUniversity Press.

Hyde, Susan D., and Nikolay Marinov. 2014. “Information and Self-Enforcing Democracy: The Role of International Election Ob-servation.” International Organization 68 (2): 329–59.

Kelley, Judith. 2008. “Assessing the Complex Evolution of Norms:The Rise of International Election Monitoring.” InternationalOrganization 62 (2): 221–55.

Kreps, Sarah E., and Debak Das. 2017. “Warring from the Virtual tothe Real: Assessing the Public’s Threshold for War over CyberSecurity.” Research and Politics 4 (2): 1–8.

Kreps, Sarah E., and Jacquelyn Schneider. 2019. “Escalation Fire-breaks in theCyber,Conventional, andNuclearDomains:MovingBeyond Effects-Based Logics.” Journal of Cybersecurity 5(1):1–11.

Leeper, Thomas J., and Rune Slothuus. 2014. “Political Parties,Motivated Reasoning, and Public Opinion Formation.” Advancesin Political Psy

Public Opinion and Foreign Electoral Interventionto know whether Russian intervention caused public sentiment about democracy to slide, especially since the ... denounced Russian interference

Documents