You are on page 1of 25
EIGHT Dealigning Campaign Effects in Argentina in Comparative Perspective Kenneth F. Greene In 2015, Mauricio Macri of Cambiemos (Let's Change) narrowly defeat! the Peronist standard-bearer, Daniel Scioli, in a stunning come-fur- behind vietory. Maeri’ win was notable not only because Peroist didates had dominated national-level politics for 22 of the 32 years since Argentina democratized in 1983 but also because three successive Ki nerist administrations yielded seemingly insurmountable anno the incumbent party. Indeed, few commentators or ee : Macri victory, and some even expected him to drop out during ae rounds of voting. He trailed Scioli by 14.9 percentage points in oe aa 9 open primary and closed the gap but still lagged behind by 2 ras age points in the October 25 first round. Contrary to oun {ar continue to rise during the remaining four wees of he He snatched victory from the jaws of defeat, besting Scioli by 2 age points in the November 22 runoff, ears Vee This was not the only striking electoral upset in rece al cee Fox’s come-from-behind victory in Mexico's 2000 presi elected 0 ended 71 years of dominant-party rule, Boris Yeltsin 25 * onl residency in Russia's 1996 election after losing the first ™ ‘aco nady Zyuganov, the ultranationalist leader of the reconstitute 162 "| Escaneado con CamScanner Dealigning Campaign Effects in Argentina in Comparative Perspect; Mpective 163 party who advocated a return to socialism. Other ; , : med unlikely just weeks before Electign = For ines ans a Henrique Cardoso overcame a 40-point deficit oc, Brave a me in 1994 Baker, Ames, and Renné 2006); Benign “Ninoy” Aqueeen? similar late surge to win the Philippines’ prt, cy in 2010, made a predicted that Donald ‘Trump would win the n 2016. The outcomes of some high also gone against expectations, including the tr defeat of the Colombian peace process in 2016, Tris tempting to look for deep, structural causes of electoral outcomes, For instance, economic globalization i in different ways, for Donald Trump's win, Brexit’s success, leftists’ victors countries (Baker and Greene 2011), party and party system breakdown in Latin America (Lupu 2016¢; Roberts 2014), the struggling European eft (Kitschelt 1994), and the more recent success of concer eting forces (Kitschelt 1995). In fact, most political science theories ab out which par- fsa forces win are based on influences that change slowly, such arden es. Scholars argue that citizens’ class (Barto- 2000; Evans 2000), employment profile (Kitschelt 1994), and/or ethaie identity (Chandra 2004; Horowitz 1985) determine theorists go so far as to argue that democracy ise concession to the 2006; Boix 2003; definition of electoral con sand, of course, United States’ Profile referendums have iumph of Brexit and the vote choices. Regime If represents a policy poor in developing economies (Acemoglu and Robinson see also Meltzer and Richard 1981). Even though the “liberal democracy” almost always highlights uncertainty in testation, our theories are designed to take uncertainty out of the causal story. ' Without uncertainty, it should be possible to forecast election out- RuNSS well ahead of time. Although some of the surprising outcomes veatigned above may be due to forecasting errors itis also plausible that “ote choices respond not just to long-term influences that bind es __ Petcular partisan options but also to shorter-term incense Sao ih the partisan campaigns. Candidates in most competitive ioe ‘vest massively in advertising their platforms, competence, a 7 ne [SS while decrying their opponents’ plans and attributes. Sucl pe Guilt align voters with their “natural” candidate (ie. the can ee cals4 with their pre-campaign attitudes and seep er ones wayPaiga influence would be minimal and outcomes woul ae ib the predictions of structural theories of vote choice email anttisns could dealign voters by drawing them away hoices away from “ndidate. Where significant, such effects can push vote cl ° > Escaneado con CamScanner a 164 Campaigns and Voters in Developing Democracies the predictions of structural theories and make it di comes before the campaigns begin. Analysts of elections and voting behavior in the Un open than are students of comparative polities to the notion paigns cam affect voters’ choices. Nevertheless, research oy) States shows that campaigns typically have a minimal j outcomes (Berelson, Lazarsfeld, and McPhee 1954, C, Converse 1962; Lazarsfeld, Berelson, and Gaudet 1948; Salmo, more 1989) and align voters with their natural candidate (Goi, King 1993). A main reason for this snapback effect is that mos: 9 tethered to a partisan option by deep-seated partisan iden Palmquist, and Shickler 2002). But voters in newer democra firmly established partisanship and thus may be more persuasive effects of campaign rhetoric (Greene 2011), Wher effects may be limited to special “critical” (Burnham 1970) or “reigns” (Sundquist 1983) elections in the United States that seem to oocur ces every couple of generations, they may be the norm in new democrzc In this chapter, I argue that Mauricio Macri won Argentina’ pres because the campaign had a decisive effect in drawing voters to his These effects were large enough that, without the campaigns, Danie! would have won and Peronism’s long run in power would have oon Talso argue that variation in the strength of partisan attachments xcs countries produce different degrees of campaign influence on elec outcomes. Whereas strong partisanship in the United States has limite? campaign effects in the most studied elections since 1940, and weaker = tisanship in Mexico propelled bigger campaign effects in the three 2 democratic presidential elections since 2000, middling levels of px ship in Argentina generated middling degrees of campaign influence 2015 presidential contest. : The first section of this chapter discusses my argument in more" ‘The second section shows that many voters shifted from 20 ote 2 intention to support one of the candidates before the campaigns 9. earnest to a final vote choice for another candidate. The third wn —_ cusses the dynamics of Argentina’ 2015 presidential election The fourth section uses specialized survey data from the ATg® ies Election Study (APES) to demonstrate the impact of oe eo moving vote choices. I also compare these findings to imi Pr on the United States and Mexico. The final section before °F ic argues that voters in Argentina were persuaded by comps jus change their attitudes on key variables. These changes did ™ fficult to forecay Mog, ited States impact on 3, once dea. > | Escaneado con CamScanner Dealigning Campaign Effects in Argentina in Comparative Perspect ive 165 ed vote buying or from priming, whereby voters come to think of i ink of acrempt 5 ant iftaences as more important aspect oftheir vote choices A Theory of Campaign Effects Across New Democracies partisan campaigns are designed to influence what and how voters think vee candidates. The more susceptible voters are to these attempts, the api the campaigns influence their attitudes and, by extension, their 088 movges, OF course, some voters are all but immune to such influence because their partisan attachments discourage them from assimilating new information that is contrary to their beliefs Bartels 2006; Greene 2011; Zaller 1992). For instance, dyed-in-the-wool Democrats in the US context never seriously consider voting for a Republican, nor the reverse, tion about the candidate inclined to choose their ittach- would and out-partisans notably resist positive informal they dislike. As a result, strongly partisan voters are natural” candidate. By contrast, voters with weaker or no partisan a ments are more susceptible to the influence of campaigns. Campaign influence may operate in two main ways. First moveable vorets may be primed to weight certain attitudes as especially important in making dates typically emphasize issues their vote choices (Druckman 2004). Candi on which they have an advantage (Petrocik 1996) and attempt 9 downplay issues on which they are disadvantaged (Hlart 2016). For instanct Repub- licans in the United States routinely emphasize issues of national security and downplay questions of social justice through economic redistribution. If successful, primed voters would follow suit, making national security issues more important element of their vote choice and social justice a less impor tint one. In the extreme, citizens may become single-issue vow Second, campaigns can persuade voters to hold particular ae attitudes that afect their vote choices (Bartels 2006a% Greent 201 on Re 152010). Most issues are presented in positional terms pecause candida Adopt different stances about how to resolve given PP , s omic grow srnblicans in the United States might argue that Orr . lould be enh hhereas Democrat h: es, wherea! g anced through lower tax eas Doetence 5 o writes through public spending. Even s0-CH160 economic everyone typically agrees (Stokes 1963) SUCH TT san mores can jomthis better than weaker growth and Jess ries De ots F Presented in terms of alternative positions: ‘Campaigns P Tosing the g2P en they alter their preferences on these pos! tional issues © and their own platforms: ‘Weer , voters’ preferenc: — Escaneado con CamScanner Although I distinguish between priming and pers cal section of this chapter, it is useful to think of boy as “campaign effects” for the theoretical discussion, broad features of political systems influence how Many strong ya ia partisans there are and thus how susceptible the electonsn ®t to such campaign effects. Scholars debate whether partsns¢*Mk a running tally that summarizes voters’ isue preferences sf ings about the competing parties (Fiorina 1981) or expec but more deeply held social identity (Bartels 2002; Camptell ¢ ee Green, Palmquist, and Shickler 2002). In either case persna 3 increase as party systems age (Converse 1969), conditional ont! partes maintaining fairly consistent profiles (Lupu 2013, 20169 40 rationalist perspective, parties with stable platforms overtime pm voters to learn about their issue positions and form stronger stuck to their governance plans. From a sociopsychological perspective, el, Parties over time permit parents to socialize children and transmit partisan identity across generations (Beck and Jennings 1991). By contrast uns party systems, especially those in newer democracies without a long tah tion of partisan competition, yield fewer partisans who are more sep tible to the influence of the campaigns. Departing from this argument, I previously developed a threfll distinction among new democracies to predict the magnitude of cat paign effects (Greene 2011). Next to long-established democracies lit the United States, cases of redemocratization where the electorate ey rienced one generation or less of authoritarian rule should mantis the strongest partisan identities. In these systems, including Argent Chile, and Uruguay, many parties that held sway in the prior demoowt era return to competition and intergenerational partisan transmision § only partly interrupted. Electorates in new democracies that expe" long episodes of competitive authoritarian rule (Levitsky and Way 21" Schedler 2013), such as Mexico, should have middling levels of 0, ship and experience intermediate campaign effects (Greene 201 77 In these places, competition was managed and manipulated, Dut PIT, options were permitted to compete and establish links to atleast ™ Sroups in the electorate, thus paving the way for stable multip don after democratization (Greene 2007). Finally, counties Jos enced “democracy from scratch” (Fish 1995), where long-te"™ ™ het authoritarian rule banned all partisan competition, a8 in Russ jap Soviet successor states, should leave electorates with the WHET wh moorings at the onset of democratic competition and thus the nerable to campaign influences. Escaneado con CamScanner asi h TON in the m Processes oy ‘ge argue thy he xn ty co! yt pealigning Campaign Efes in crgentna in Comparative Penpeci tive 167 with respeet f0 my typology Aaentina is a case of redemocratization re han a generation of authoritarian rule and where prior part vind continued to command relatively strong identi ie "The Peronist party, under different labels, was founded in yoq7,and one ofits main current rivals, the conservative Unién Civica Red. ical Radical Civie Union—UCR), traces its roots to 1891. Ostiguy (1997) isles that attachment 10 these poles goes fir beyond what we would tly conser to be partisanship to become a deeply held and culturally de set of dispositions. O’Donnell (1973) implied that these strong streets mate Argentine polities into an “impossible game.” The Per- ates he argued, commanded such strong working-class support from the 1040s to the 1970s that it would win under democracy. Lacking similar lev- sof support the Right could not win in open elections but could encour- agethe military to shut down civilian rule. Repeated episodes of authoritar dnnrule reached their zenith in the military government from 1976 to 1983. Following redemocratization in 1983, prior parties reemerged once again {Remmer 1989) and have structured partisan attachments to 2 significant extent (Levitsky 2003; Lupu 2016c). Ifmy argument holds, then Argentina’s campaign influences; however, the degree of should place it between two other cases: On the one hand, voters in Argen- tina should experience more campaign effects than those in the United Stes, where uninterrupted democratic party competition has permitted the emergence of very strong partisan attachments. On the other hand, Voters in Argentina should experience less campaign effects than those in anthoritarian rule before democrs= ‘competition between an als (Greene 2007). tina should afer | reemerged anon vores voters should be susceptible to successful campaign influence ft where long-term competitive aut pat n 2000 featured muted and manipulated ms Jeger dominant party and two niche-oriented rls te degree of suscepsbiity to caunpaga infuence 19 ATEEN $¢ understated because it ean be “Jecisive in close electio Contest between Macri and Scioli. Argentina's 2015 Changes in Vote Intentions during ason Presidential Campaign se predtecton vere Fa Dd “cumenting f whether nd the hing tit final vote choic Shane ghvential. I s important £0 Wii eed intentions to choices Te impactful, further analy changes between citizens gives a sense o Escaneado con CamScanner Campaigns and Voters in Developing Democracies s 8 ‘Survey Wa (dune 27-4 0 0 ° oat ht fe oe oe © 8 : © © © HY 4 SO eS PMP SE EE FEEL Mac -=-sciol) —Nassa Fig, 8.1. Cross-Sectional Poll of Polls during Argentina's Presidential Election es, 2014-15, ‘Source: Leogus, htps//commons.wikimedia.orp\windex php?curidad4SS7404 changes represent ali igning or dealigning campaign effects. I take up hs deeper question ii a later section of this chapter, k ‘The presidential campaign season included three ane distinguished in the poll of cross-sectional polls presented inf ae Before about March 2015, the three leading candidates were Lee ‘ered, Daniel Scioli of the incumbent Peronists running undet i of the Frente para la Victoria (Front for Vietory—FPV), Mauricio bis the Cambiemos (Let's Change) alliance led by his Propuest (Republican Proposal—PRO), and Sergio Massa of the UateF eat Nueva Alternativa (United for a New Alternative UNA) Polled at around a quarter of the vote intentions, with eight ©! soto dates making up the difference. Between March and Octobe So “sacri pulled away, eventually winning 37.1 percent and 3 a the official voting, respectively, in the October 25 first 7" substantially behind, winning 214 percent of the vore am, Out of the race. The final monthlon print to the Novem clection saw Maeri inch ahead of Scioli for the victory Escaneado con CamScanner nd. MSS and nen ining Campaign Efe in Argentina i ' Dealigning ign "gentna in Comparative Perpecive 99 syghough the cross-sectional data are suggest oO a much clearer sense of the consent moron j lying shi, gnd shows which candidates ceded votes to him, Figure 8,2 my Macri oe associated with the August 9 primary elections the ga nyese cana the November 22 ano election In each chica ye a ine aip those of his opponents, whereas cil momentum Aa eth campaigns for the primaries nearly a quarter ofthe eg scented on one ofthe candidates. As they did so, Scio, Maes and Massa coves fatty evenly, though even at this early stage, Mac surpassed sar The most damage was done in the next phase, between the primaries round 1. As voters fled the failing eandldacies of Massa and the other srlleontenders, they flocked to Macri giving him another 8.1 percent fhe electorate compared to Scioli$ lackluster 14 percent gain. (Se the saline appendix for further detail on these shits) Following the round I elections, Macti particularly drew from orphaned Mass voters after their preferred candidate dropped from the rae. Between end of round 1 and round 2, Macri’s share of fleeing Massa voters nearly tiled Scilis take. Undecided voters broke almost evenly across the evo frontrunners after the primaries, and although Macri outstripped Scili in “picking up voters from other candidates in the rae, chese amounted to jas 216 percentage point advantage overall In addition, Macri only took 12 percentage points from Seioli and none went the other way Tes thus not an derstatement to suggest that Macri won the election in Iarge part because Ie absorbed fleeing Massa voters as well as retaining his inal PPON=S uring the latter stages of the campaign season? rransit Campaign Messages during ‘Argentina's 2015 Presidential Election , , tions caused es that the impact of the campaigns on vores Serre In this a rise and Scioli’s relative decline during the a , I describe the leading candidates’ campsisn ® o : cempaignseson ching aoe 2O5 APES sey waves were well dined se 210 7. befgge tudes and preferences. The first wave is the * etn the August 9 primary, and thus captures PFE aXe ees and the meses ber 30. Dropping a a tera the November Tos a aeretgectional surveys FEU Ng clips “Massa drops ovr” measure «this wave captu! ese ave cap! ain parallel those in cr0s Sceur after round 1, In this period, — Escaneado con CamScanner 1 pue sovewn squapuodso1 jaued Mm painseaus se axeIOa}9 Yeo JO WADIOK axe SOM ION " ‘SL0z Sady ‘o2un0s uoseas Ubledwe5 ayy GuuNp ssoVUIN\ NZ'S BLL sionwwprmpepmepun [I] SN MM vee} von] noes El (eajouo eon z puns <- 891049 810% | punci) zpunoy (991049 0104 |. punos <- e914 ojoA Axewid)| (99}049 OR Krew <-YonLOLU g}oA AYE 1 punoy ous wee Escaneado con CamScanner peslgning Compaign Effects in Argentina in Comparative Pepe 8 pective 17] any vorets changed their preferences on the themes that the eam. ign highlighted. - ae i ‘path Scioli and Macri invested heavily in advertising through the During the primary season, Scioli outpaced his rivals by spending mean pesos compared 0 Mac 36 milion, For both candidate, the soi sare ws dedicated to publicity—30 million pesos for Seiol and 20 sar for Macti, spread across television, the Internet, and posters. In ion, because Fee (govermment-sponsored) television airtime is based at each party’ prior vote share, Scioli received about twice Macri’sirtime through round 1. After that, Macri’ strong electoral showing put him on nearly equal footing. Despite his early monetary disadvantage, Macri mounted a sophisticated public relations campaign. He hired a slew of political consultants from domestic and international circles—including Jaime Durin Barba from Feuador, Roberto Zapata from Spain, and Joaquin Mollé from Miami who employed 80 people and produced 300 different spots for television and social media, He also pursued a “frenetic campaign that crisscrossed the country”” making whistle-stop speeches in regions traditionally consid- ered Peronist strongholds.* Scioli also employed an international team, including his longtime Geomign advisor and former Bill Clinton strategist, James Carville, who {hows to be the mose expensive American political consultant.’ When is oe out a video of a girl crying inconsolably at the prospect of a Sug, 1 Scio was quickly inked co Jofo Santana a Brazilian spe- ee emotionaly charged attack advertising who previous ‘ns for Lula da Silva, Dilma Rousse and Hugo Chiver® ange, ee crystalized around the question of continuity La cmnindes a time of the campaigns, voters had come a a Bins haf inistration’s record on the economy Pre-2010 econom = Pred under the weight of 30 percent inflation, annua A Vith internati Percent, poverty was creeping up, and a protract™ aids, Ainge credivors east a datk shadow in some wr ay disredie aes much of campaign content involved artac® i eh ‘cri and responses that sought to neutralize such rhetorie that ™: directed ante= a vt Areracs gasto-d = oe [anacion, com.ar/1826621-scioli-fue ceandidatoque-mas-85" 1 Sitaon a Paso art Rises in "si Ae No Jonathan Gilbert, “Argentina Considers shift as Upstart Ris 2 jp laseson de Ne ork Times, November 20, 2015- pa Politica Ondine, MAY 20 ern ton volvi para ayudar a Scalia ser presidente” Polite bio Satan !Aboliticaonline.com/nota/80796! jnculoscon el de Scioli, neg vi Supuesto experto en eampaias negatl 4a Nacién, November 11, 2015+ SL cilista” Escaneado con CamScanner ON 172 Campaigns and Voters in Developing Democracies o4 03 02 on Zscore oA 02 03 04 toMacti | toSciot | toMacri | toscion | tomacd | wseai APresidentialapproval | A Sciolicompetence _| A Reistibuivepretsrens Fig. 8.3. Shifts in Campaign-Related Variables and Changes in Vote Choices Note: Categories refer toa shift in primary vote intention for another candidat toa runoff tee forthe candidate noted. Bars represent z-scores. Data come from the 2015 APES. In one strategy, Scioli and President Fernandez characterized Mac ri’s economic policy plans as radically out of step with the mainstrea”- Macri discussed ending the cepo cambiario (“exchange rate clamp”) ee on purchasing foreign currency to spur exports, terminating expensi energy transportation subsidies for the middle and upper cass striking a deal with recalcitrant foreign creditors in seeking a lier relations with private banks and the International Monetary (IMF). Scioli referred to these plans as “savage capitalism.” He arg that Macris policies would lead to an effective “megadevaluaton Tepresented a return to the Menem-era economy that facili 2002 economic collapse.’ While campaigning on the outskirs ¢ it nos Aires, Scioli summarized his contrary view: “Your vote". jyate Se” rivate between the state and the public sector, or the markets and p! eth 4d Seni 7. For more details on the candidates issue positions, see Lup, Oliveros" (this vol.) and Schiumerini (this vol), 4 Escaneado con CamScanner esting Campaign Efs in Argentina in ComperstisePepecioe 173 which will - | abandon orkers, pensioners, students, the poor and the regy of heightening the contradictions on policy largek hand, Macri made an explicit pledge to fisep sare nt sure Kirchnerist economic policies, such as the nationalization of the al and airline companies and the conditional cash transfers targeted at poor households. On the other hand, Scioli’s own economic policy plans either perm comparatively vague” or agreed with Macri’ own plans.” ‘tnother strategy to damage Macri’ rise focused more generally on sis image. Scioli and President Fernandez launched blistering attacks in the mass media, characterizing Macri as a technocratic and cold busi- ‘erage Argentine’. In put the stra got. On the om nes magnate whose values were far from the response, Macri sought to soften his tone by avoiding controversial social isues, He designed television spots that showed him mixing with ordi- nary citizens and portrayed him as above corruption. His sober image seems to have hit home as even Scioli attempted to distance himself from President Fernandez’ increasingly aggressive attacks against Macri as the campaigns progressed. Rather than drawing new voters to his side, Scio’ late attempt to become his own man made him appear weaker and softened support among loyal Kirchner allies. During the campaign season, voters polarized on variables that were highlighted by the campaigns and may have been relevant for their vote choices. Figure 8.3 shows voters’ shifting attitudes on presidential approval, distribu- and preferences over economic re -Ineach instance, voters who reported an intention to supportsomeone than Macri in the primaries but voted for him in the runoff shifted attitudes in an anti-Scioli direction. These voters became more cnt of President Fernindez, more pessimistic about Scioli’s competence, less in favor of economic redistribution to benefit the ot support Scioli in the run-up to the primary election, in the runoff, shifted in the opposite direction. ScioliS perceived competence, Scand Daa i ee Lough, “Macri Runs Confident Campaign in Ar 9. La ygamber 13,2015. fol alyzed candidate statements and showed that Macon wed by Macri and then Scioli, herp:/Avww-lanacion q “€N-Sus-propuestas Scic " is axes on W ance, Scioli agreed with Macri about removing exPoTt -_ 7 genta’ Runoff Becton Jearest Pro" ssa made the clearest PY ee omaar/1838540-m55 , Jheat and corn: Escaneado con CamScanner 174 Campaigns and Voters in Developing Democracies Campaign Effects in Argentina's 2015 Presidential Election Shifts in vote intentions away from Scioli and toward Macti dui campaigns as well as changes in voters’ preferences on the isues thew” campaigns highlighted provide circumstantial evidence that the ame themselves may have affected vote choices." In this section, enna statistical models to deepen the case. The models also help show the anes to which the campaigns aligned voters with their natural candidat o dlealigned voters by pushing them toward a candidate they would not hve selected before the campaigns began. T first construct a regression model of vote choice in the November 22 round 2 election. In most ways, this is a standard model because it incorporates the variables typically used to predict vote choices, inching demographics, partisanship, and preferences over major campaign iss, including the incumbent performance criteria, candidate competence TABLE 8.1. Vote Choice in Argentina's November 22, 2015, Round 2 Election Age ~0.010 0.009 Female 0.278 SES od Education 0.089 : FPVID, Jun 0.849 = AFPVID, Jun-Nov 0.750 PJD, Jun 0.594 APJ ID, Jun-Nov oars PRO ID, Jun 1.016 ca ‘APROID, Jun-Nov 0787 - Redistributive prefs index, Jun 0.056 : ‘A Redistributive prefs index, Jun-Nov 0.047 Scioli competence evals, Jun 0.12 2 ‘A Scioli competence evals, Jun-Nov 0.103 : residential approval, Jun 0.202 i ‘A Presidential approval, Jun-Nov 0.169 ” Constant 1424 ere None Na 50,7 = Sh. The dependent variable coved 10 nt. 0 ST he cl and 0 for Macri, Cooicients re logis. Change score variables enonch V2. rile titudes between the panel waves with higher soores associated wit 1 to wave 2. TI, Tewould also be interesting to know how actual campaign Hs eo unlike tracking polls, which can capture specific inflection points income a sneasure public opinion just before and afer specific events the SPICE ANDES makes ita better ft for detecting broader campaign-seson can Escaneado con CamScanner igning Campaign a Dealgrng Compaen Eien Argentina in Comparative Pospecing 175 luations, and, albeit to a lesser extent, th . sal above. the economic policy difference ‘The model also differs from the standard approach pane nate of the data that interviewed the sume nan ut® thecampaign season and then aguin after Election Day ge change-score variables (represented in table 8.1 with Each of these variables indicates changes in votery campaign season. Positive values indicate higher seore election poll compared to the wave I preeleetion pol These variables give a sense of whether attitude change affected final vote choices. If voters’ preferences over key issues were set before the ean psigns began or had they changed haphazardly during the campaigns these variables would not affect final vote choices. As it stands, the results in table 8.1 show that they had an important impact. One way to assess this impact ist entertain the counterfactual question, What would have happened if the campaigns had not existed and all voters had chosen their “natural” candidate as measured by their pre-campaign dispositions in the June wave of the survey? (Note that each voter’s natural candidate is predicted by the model and may differ from her pre-campaign vote intention.) Using this approach yields three ways to appreciate the impact of the campaigns. First, a model just like the one that appears in table 8.1 but without the impact of the campaigns (je., a model that sets all ofthe change-score ‘arabes to zero) correctly predicts the final vote choiess of 73 pert ofrespondents. When we bring the change scores back in so chat the Ml fore ofthe campaigns impacts voters’ choices, the model comrecly PE tics 86 pereent of the final vote choices. In a manner of thinking !mpact ofthe campaigns is to improve predictions or deorei Pr tor) by 13 percentage points. The model does quite We tude Mpact of the campaigns, but it clearly does much better se influences, dl Second, without the impact of the campaigns the m0 is Would have won 4.5 percentage points ae 'gely consequential. Given that the final differs ay, RET was 2.68 percentage points of the POPY Te a have won without the impact f the ined byt cong De 0 argue that Macris win was DOF PTY chments OF : tions in Argentina, the influence of pal ce in office rather, : ere nO of President Feminded’ penton ign seaman camper ted or persuaded (or both) during come. 80 effects had a decisive impact OP ging the 8 carly in the model includes the triangular delta), attitudes during the sin the wave 2 post- el indicates that "This finding Escaneado con CamScanner 176 Campaigns and Voters in Developing Democracies Finally, the counterfactual approach can distinguish a ing campaign effects. To uncover these impacts, dicted natural choice based on her pre-campaigi choice model in table 8.1 to her self-reported ; elections from the June wave I survey an fnal vote choie nna Finkel 1993; Greene 2011, 2015; Lazarsfeld, Berelson, and Gaude ina I represent the various effects in figure 8.4 (see the online appendix to s the same data ina table). The first bar foreach country rereent . aligning effects; the second bar represents dealigning effects. I focus on Argentina here and later compare itto the results of similar analyses for Mexico, United States, which are also represented in figure 8.4, Unsurprisingly, the campaigns aligned the vast majority (734 perm) of voters in Argentina's 2015 presidential election (the first bar). Mon o these voters (39.2 percent, shown in black) decided to support their natusl candidate at the outset and did not waver across the survey waves. A very small group (just 2 percent, shown in white) flirted with the idea of sup porting the other candidate before the campaigns began but then cane home to their natural candidate by Election Day. Another large proportion 2.1 percent, shown in gray) were initially undecided but ultimately voted for their natural candidate. The large size of this later group is probabiy an artifact of the multi-round electoral system. Voters’ initial vote inter tions were measured in advance of the August 9 primary, when voters e choose among 15 candidates, but final vote choice in the November round 2 was limited to just two candidates. Asa result, voters who intended to vote fora candidate other than Scioli or Macr in the primaries wes? necessity, coded as undecided. Had the electoral system fears round withouta runoff, the category of undecided voters likely woul been far smaller and the proportion of voters who remained align: their natural candidate (the black bar) would have been larger op aa a sign Despite the generally aligning effect ofthe campllns * Foy Proportion of voters (26.6 percent) ultimately dealigne' sented candidate contrary to their pre-campaign dispositions, rePrE TT oun second bar. A litte more than half of these voters (15.0 Pi get in gray) came from initially undecided voters or voters ON oe ; choose one of the 13 other candidates in the primary: ST due (Mt began and ended the campaigns at odds with their aoe percent, shown in white) or were converted away from ofsoité = probabil as 12. A.voter is predicted to vote for a particular candidate when ie hihe ha nds is for that candidate (generated from the model in the online ape! other candidate. Escaneado con CamScanner igning and dy are each Votes, roman ts re m disposition using Vote intention in the ys Tom Dealigning Campaign Efects in Argentina in . . Comparative Pe erpective 7 0% ‘gentna201S: " Agontina2015: — Moxloo2000-12: Mole02000-12: United Stats 1960: Unind Stats 1800 ‘NgringEect —DealigningEfloct Aligning Elect DealigringEfect_—Algningfect Dealing Ect Wiitalintention for natural" candidate [E]initially undecided (inital intention against natural” candidate Fig. 8.4. Campaign Effects in Argentina, Mexico, and the United States ‘Note: Aligning effects indicate a final vote choice for a voter's natural candidate. Dealigning effects indicate a final vote choice against a voter’s natural candidate, "Natural" refers to the candidate ‘sociated with each voter's pre-campaign dispositions as determined by a regression model, The model for Argentina appears in table 8.1. The models for Mexico appear in Greene (2015) and inthe ontine ‘penx. The model forthe United States appear in Finkel (1993). 4 date to vote for the other contender after hearing the campaign messages (7.2 percent, shown in black). These dealigning campaign effects proved ‘ . "etined a large proportion of his initially winning coslitc Fae ie More voters into his camp as the campaign season Progresso" Tie Offs he led Scioli by three percentage points among dealigniNé ori, What is Would have naturally voted for Scioli but instead chose Macri. by winning Most striking about this process is that Macri pulled head mot a tially undecided voters but by winning natural Scio WS anof. Stil in the primaries but then “converted away” (0 sah above. Se are the strongest of the campaign effects idenler oe also show that voters converted aw: ina bsotbing the eampaigns,natral Seiler ‘hified in an anti-Seioli direction much mor decisive. Whereas Scioli ition, Macri drew ers who chos u ae Escaneado con CamScanner 178 Campaigns and Voters in Developing Democracies became about half a standard deviation more critical of President Fem « dez and of Scioli’s competence and about a quarter of a standard de Min. less favorable toward the redistributionist policies that Scioli sup a Hypothetically, strategic voting could have played a role in Mies paign season shifts that the statistical model in this section is desgnaj measure." Argentina's two-round system means that voters who pe : preferred a candidate who dropped out after the first round were lined to choosing between Scioli and Macri, if they voted. As a result, some of the two runoff candidates’ support could have come from citizens wo held their noses and selected the candidate they viewed as the least worst. This strategic support could have especially benefited Macri, given the informa. tion T presented above implying that he drew strongly from voters who supported Massa in the first round. If strategic voting accounted for the findings above, such voters would have changed their vote choices simply due to changes in the availabe cn- didates and not, as I have claimed, due to the influence of the campaigns on their issue preferences and candidate evaluations. If this were true,sta~ tegic voters’ underlying preferences and evaluations would have remained the same even though their vote choices changed. The impact would beto bias the results in table 8.1 against my claims of campaign influence Sill, to check for this possibility, I examined campaign season changes in the variables highlighted in table 8.1. For two of them—spen dential approval—such changes were the same, on average, who did and did not support the two finalists in the first round. Change were also the same for this subset of voters when they were further divided vcrove those who said people should always vote for the candidate 8 prefer and those who said people should sometimes vote forthe ands most likely to win (ie., strategic voting). The same pattern emerged ee i further subset of voters who supported Massa in the frst round and in the runoff. ond “The only exceptions were evaluations of Scioli’s competenet py variable, a number of the comparisons did reach statistical sign" = oe | in the direction opposite what one would expect thats the PO tegic voters came to rate him more highly. The only way this fi here support strategic voting would be through complica pear people hold their nose and vote for Scioli but then sationali#e °°. estimate that 610 10 oe A in. Consistent al leeon oie! ding and pres across votes 13. In fact, Weitz-Shapiro and Winters (this vol.) respondents voted strategically in the 2015 presidenti ‘ment of this chapter, these authors show that persuasion ability preceded the decision to vote strategically. regarding _ Escaneado con CamScanner Dealigning Gompaign Eft in Argentina in CompamicePpeace 99 telieving that they must like him more than they did befor ths fisenating but complicated possibility. Short ofthat, tion I have indicates that the campaigns genuinely influ tnd beyond the effects of Argentina’ institutional architecture, nsum, Scioli probably would have won the presidency ifthe campaigns had not occurred. Their effect was to weaken his position and strengthen Macri as voters absorbed his messages. Ultimately, a small but decisive group of voters who appeared to be solid Scioli supporters due to their par, tsan attachments, demographic profiles, and vote intention inthe primary elections dealigned not just by choosing Macri but by becoming genuine supporters of his policy proposals. Argentina's 2015 election thus joins the list of other consequential come-from-behind election outcomes identified in the introduction to this chapter, where structural models of vote choice could not have predicted the outcome. . [cannot assess the best informa renced votes above Campaign Effects in the United States and Mexico Before analyzing why the campaigns exerted a decisive impact, I compare the magnitude of dealigning campaign effects in Argentina to the United States and Mexico. I argued that the campaigns should be more impactful in Mexico and less impactful in the United States due to variation in the Strength of partisan attachments. ' Itis only a slight understatement to say that analysts of the United States have long been obsessed with finding campaign effects. Candidates “vest hundreds of millions of dollars in mounting sophisticated campaigns “signed to segment the electoral market and promote precisely the candi- Ste image and platform issues that will win support. Yet aggregate effects impaigns in presidential elections between 1940 and 1994 caused ie any eB margin shift on average (Bartels 1992, 266; Berelson, Laser M Kun tePhee 1954, 16; Campbell 2000; Finkel 1993, 14 Eee eae 1999. pn Petrocik, and Shaw 2008, 167-68; Markus 1988, 0"; 9 2006),15 ed counted as one estimate- Tange number of studies O” jens, including celevi= the median effect across studies with each study cit yields a 3.4 percent margin shift on average: examine the effects of specific elements of dio advertising, di crtising, dit i te appearances. Some but otal of NE or excellent summaries, see Kav! Escaneado con CamScanner 180 Campaigns and Voters in Developing Democracies ‘The cleanest analysis of campaign inflaence in US pri tions comes from the 1980 American National lectin tld included a specialized panel study much lke the APES dye, Mi described above and used throughout this volume. Finkle the United States data using a method parallel to mine, ant his findings in figure 8.4. Feprodueed ‘The 1980 election featured President Jimm incumbent who oversaw an economy in recesion, concem on ilegal drug problems and urban violence, a seemingly inaetth crisis in Iran, and stil strong Soviet Union with which the Unied $e remained engaged in proxy wars around the lobe. Against this bcking Ronald Reagan offered the optimistic message of “morning in Amerie’ ‘The 1980 elections thus stand out asa potential critical or realigning on Yet even in this context, figure 8.4 shows that the campaigns dedlignd just 15.9 percent of voters. The large majority of these voters were soled Reagan Democrats who broke ranks with their natural candidat to help deliver the Republican to the White House. But the impact of the can- paigns can be overstated, even for these voters. The vast majority of them questioned their support for Carter from the outset, instead telling suvey interviewers that they were undecided. Thus, to the extent the campaigns mattered, they helped uncertain voters validate their choice to defect fom their natural candidate and instead support Reagan. That the 1980 clstion sticks out as a major example of campaign effets inthe United Sat telling. By comparison, Argentina’s election was significantly more ™" able to campaign effects in 2015. aa "The relstivee of American voters to campaign influence coe seemingly “change” election, is not particularly surprising Or be Der tisan identification. In 1980, 87 percent of wotes identified wth cratic Party or the Republican Party (Fiorina 1991, 25). Kei 2, leagues (1992) have long claimed that independent voters ate 20 United States, implying thatthe remaining 13 pereent fT eof sented themselves as independents may have in fact e, jnfortation the major parties. Partisans are naturally less easily swayed By alt that contradicts their existing views (Bartels 2006a; Gre! sway 1992). Thus, the more partisans, the less likely that on es, whe Unlike in the long-established democracy in the USE yoo stability in partisan options helps solidify partisan 2°00") ing va only experienced fully competitive democracy since 2000. at the 20th century, its dominant party authoritarian regi™ 'y Carter as an Unpopalee hostage Escaneado con CamScanner Deagnng Campaign Efecs in Argentina in Compara pe erspective 181 ‘gn competition was manipulated against small eg the incumbent Partido Revolucionario Institucional (Institutional Revol tionary Party—PRD Greene 2007). Although some y her grached tothe PRT, many aligned with it out of nec Ae same time, some voters felt intensely at sppnsiton parties, but they repre MPetitors and in favor of ters felt genuinely sity oF con ached to the niche-ori ented just a tiny slice of the (Greene 2011). Asa result, Mexico began its new demoe genuine partisans than eountr ence, iented ‘ctorate racy with more erm fully closed s that experienced long. authoritarian rule with no partisan competition, such asthe former Serie republics, but less partisanship than Argentina’ experience of redemmers tiation after Tess than a generation of authoritarian rule. By extension, my argument predicts more dealigning campaign effects in Mexico, ‘The findings in figure 8.4 come from analyses of panel surveys fielded before and after Mexico's 2000, 2006, and 2012 presidential clections, The models used to generate these predictions appear in the online appendix and are similar to those presented for Argentina above. Overall, the campaigns dealigned 34.6 percent of voters, more than the 15.9 percent in the United States’ 1980 election and the 26.6 percent in Argentina’s 2015 election. Most striking is the large proportion of vot- ___esswho were converted away from their natural candidate and initial vote intention (12.7 percent, shown in white). The campaigns had such an inci sive impact that they changed the outcome of both the 2000 and 2006 con- tests. Vicente Fox of the conservative Partido Accién Nacional (National Action Party—PAN) won the presidency by 9.41 percentage points over the candidate of the PRI, Francisco Labastida Ochoa, in 2000. Without the impact of the campaigns, the models in the online appendix predict that Labastida would have won 5 percentage points more and Fox would ie Yor 49 percentage points less. Thus, without the campaigns, Labastida Mould have won by 0.5 percentage points and the PRI’ electoral os Would have continued instead of transforming Mexico into @ f'y titive democracy in 2000. re ace of the campaigns was so surprising Id an Felipe Caldersn told me in private thathe et Calder Went on . by 5 percentage points. It is perhaps aan eit Andrés Manuel Lope, ae Win the 2006 presidential eae emocratica (ary © beDeng tor of the Partido de la Revolucion 0.56 percentage points Reade Revolution—PRD) by a nee . raline appendix indi- that cage Campaign effects, The models in points ave ‘alderén gained a whopping 6.9 percentag' that, two days before the at he expected Labastida Escaneado con CamScanner 182 Campaigns and Voters in Developing Democracies campaigns, while Lopez Obrador lost 5.1 Percentage points, der went to Roberto Madrazo, the candidate of the PRI) Thue the impact of the campaigns, Lopez Obrador woul margin and his administration likely would have economic and social policies as well as methods on drugs. (See Greene [2015] for a detailed analys campaigns on these election outcomes.) US, Wri Id have won by Pursued vastly dine for combatting the.” / t war is Of the influence Of the What Caused Campaign Effects in Argentina? As described above, effective campaigns attract voters to a candidge through priming (ie., encouraging voters to consider certain issue a more important in making their vote choices) or persuasion (ie. changing Voters’ attitudes on relevant issues). In this section, I show that peru: sion accounts for the lion’s share of campaign effects in Argentina’ 2015 presidential contest. Macri did not simply beat the Peronist candidate the polls, but his campaign consequentially shifted voters’ preferences o2 key economic policy questions toward his own more market-friendly o:- entation. Thus, the campaigns had a deep impact on voter psychology nd helped cause a significant rejection of Peronist ideas among a smal but decisive group of voters. : Priming, however, did not work. The candidates attempted to SE voters by focusing attention on themes that would advantage ae 7 particular, Macri focused on discontent with President Fernandez’ LE formance in office, casting himself as the sober antidote to her agers" style. Scioli tried to capitalize on his support for public subsidies #&* Macri’s image as a free-market ideologue. seed the wish ‘Testing for priming requires finding out if voters shifte choices they ascribed to the various elements that influence their vore Che ing the campaigns while holding their underlying attitudes cont way to detect changing weights is to compare the coeflicient ign * sion models of pre-campaign vote intentions and oe "Acro choices using the same set of pre-campaign dispositions 3 Pret sth approach ensures that respondents’ attitudes remain constan models detect priming but not persuasion. a ‘Table 8.2 shows no evidence ofpriming of distributive Pr appr would favor Scioli and little evidence of priming of Presi" ined that would favor Macri. The impact of presidential PP act ‘may Hi about 10 percent during the campaigns, indicating that e so Escaneado con CamScanner Dealigning Campaign Effects in Argentina in Com ‘Paratioe Perspective i 4 183 failed to bring President Fernéndez’ performance Altematively, it could suggest that Scoli was mildly eer for voters, questions about her administration from the race, Finally, th, in removin, thc euations of Sil? competence became less imp meh show ofthe campaigns, many voters who thoughe Scio way ne gee oe ertheless planned to vote for him. Presumably these votes liked i ae other reasons, an outeome that may have been due to pense nF ‘To analyze persuasion, I adopt Bartels’ (2006) definition os ai change in prospective voters’ clectrally relevant attiudes oh ceptions” (82). One way to detect persuasion in sample survey date tp obsersing systematic shifts in explanatory variables and vote choieee A, areatsibutable to.an externa stimalus such as campaign advertising Lace ing a good measure of advertising reception, I follow the approach sug gested by Price and Zaller (1993) and use an index of politcal nowledge that includes responses to three factual questions: (1) How many prov- inces does Argentina have? (2) Who is the minister of the economy? and @) Who is the president of Brazil? I interact this index with measures of partisanship with the idea that such dispositions bias the impact of rising Inowledge from the campaigns. Partisans tend to augment the impact of information that reinforces their views (Bartels 2006a; Greene 2011). Asa result, FPV partisans who favor Scioli and PRO partisans who favor Macri should polarize their attitudes more due to the influence of the campaigns & their knowledge of campaign content rises. ABLE 8.2. Priming during Argentina's 2015 Election Season Round 2 Vote Primary Vote November 22 * eee oS 7 Sig SE _ Cocf Sig SE__—Coe < 0.006 0.002 r a 0.009 oom a 0.199 0.205 _ i 0.073 0.079 -0.153 Hee = 0.103 - 0.064 ae 0.029 beti tive prefs ind iS 0.032, 0.02 0077 Speco Seen a i , . 4 Conga *Pproval, Jun oe ae HL Sar . 447g 082 ‘ on i dude addition © ete nary, eight main candidates competed 19 we ion Oe ints are ois model a8 Escaneado con CamScanner 184 Campaigns and Voters in Developing Democracies Table 8.3 presents separate models for predicting shif . . a Mey, of presidential approval and Scots competence aswell asp economic redistribution, Two findings ate notable. These show that partisans polarized in response tothe same gener ot in the environment. Voters who identify with the FPV shifteg heen tudes more strongly in favor of Scols postions with the mages knowledge they acquired, whereas PRO partisans shifted in he eo! direction in favor of Maer. The effets are especialy noabe fp tal approval, where partisans polarize by nearly two standard dere Although partisan filters cleaely biased the reception of parts pi, cal information, the effects did not offset eachother as we mightewesg the United States, where especially strong partisanship makesitdifesh move voters’ underlying attitudes (Hillygus and Shields 2008; Zaller 1972 1996). Instead, the models show that forall three vote-relevant variables rising knowledge is associated with preference shfis that favor Moc Vox ers with the highest levels of political knowledge shifted ther atinude about one-third to one-half a standard deviation farther in favor of the position that favored Macti than voters with the lowest level of poli knowledge. The implication is that Macr’s campaign messages and inves- ment landed better than Scioli’ thetoric. Persuasion in favor of Macri accounts for most of his campaign bump and thus played a primary role in his presidential victory. Filtering the p= lution ICES Over ont TABLE 8.3. Persuasion during Argentina's 2015 Election Season z Presidential approval, Scioli competence evals, Redistributive P Nov Nor a Variable Cocf Sig SE Cocf Sig SE Coef_S a Dependentvar,jun 040 0.08 008" (OOF 024 at Age 0.00 0.00 0.01 . 0.00 001 023 Female 0.05 0.07 -0.07 O1L 008 00 SES 0.03 0.03 -0.06 on Education 0.03 003 003 OOF ay wt OF FPVID, Jun O15 * 008 039 * ee 0.09 He PROID, Jun 0.12 0.18 -030 0 Oe Tot tit Knowledge, Nov —- we) 6m hm FPVIDxKnowledge 033 0.0315 oor Olt 4, ast PROIDxKnowledge -0.36 *** 0.04 -0.10 028 1365 Constant 13 “= 01a 154 7 N 767 731 0.12 _ a 0.43 0.06 cone canals "Note: Each model uses the dependent variable at the top ofits column. ExplanatOny nies: kn dependent variable measured in June. See the main text for the components of th > | Escaneado con CamScanner ealguing Campaign Effects in Argentina in Comparatice Pe rpective 185 le changes through the vote-choice model gut in that the average non-partisan voter was os in table .1 indicates Myote for Macti due to the persuasive eff out 20 percent peer valk ok is eet wi artes mien ese persuasive effets acount for aout 10 peritage pik shift in favor of Macri and thus account for the lion’s share of his ane ene Scioli during the campaign season, which are documented in figure 8 bes paigns- Conclusion tics like to think big, The subfield came into its vn with sweeping historical analyses of themes such as regime type, state puilding, and the politics governing economic development. Researchers fen offered structural arguments for these outcomes that involved tec- tonic processes over significant periods of time. Theories of voting behav- jor in comparative politics largely followed suit, adopting approaches that highlight voters’ interests that are thought to be tied to their demographic profiles, their social class, or the sector of the economy in which they work. The prospect of voters not behaving in ways that can be read off of their sociological or economic characteristics is somewhat unsettling. If voters do not choose according to their long-te™ jnterests, then pa ‘at and distinct ideologies: mandates given sedied, and the bases OF science takes for Scholars of comparative polit ties may not represent cohere! by voters to their governments can become m representative democracy begin to look shaky. . cy begin to haky. Bu et significant hit if we project our NO” desires OF pr ere aesthetically pleasing theory instead of listening ©° what World tell us. - oo is now a significant body of research showing that patt paca have a significant short-term imp : a Messaging can be so influential that 1 converts ore 08 choosing the candidate they initially int y initially a on ould naurly support inthe a ofthe ca 15 president ; : vty tl presidential election, Foe and away from Scioli Scop \Vithou the influence of the TT nl neha Would have won 1s Ope campaigns UN ifed Voters go PPV hada larger proportion ofStTOME. inven ot ee can Yl have been immune the pers be such i 80. As it stands, enough voters were sus — Escaneado con CamScanner 186 Campaigns and Voters in Developing Democracies that Macri not only beat Scioli but persuaded voters to dislike Presi, Fernindez’ performance in office, to think of Scioli as less competens to shift their preferences mildly inst redistribution. Whether this Let shift is durable and helps Macri institute the market-oriented reforme ri secks is another question—and one that pushes us to think abou bliss theories of representative democracy that take campaign effects serious) Escaneado con CamScanner

You might also like