One of the main challenges in the study of human behavior and social phenomena is to reconcile the gap between what people say in surveys that they intend to do and how they finally end up acting. Many of the questions in surveys and interviews force people to rationalize behaviors, overriding emotional or unthinking reactions, which limits the reading of what is observed.
In different contexts, both within and outside our region, we find that electoral polls have enormous difficulties in predicting results, especially when outsiders emerge that generate an imbalance in the status quo. Something similar happened in Argentina, where polls did not accurately portray what happened in the electorate’s behavior. Many of them indeed flee from the conventional methodology manual, since conducting surveys with solid samples is expensive and budgets are limited. In addition, the country does not have updated census data to work with a quality sampling framework. And it cannot be ignored that many agents that emerge as “pollsters” are political advisors, mostly interested in showing data that favor their clients.
But elections also mobilize emotions that traditional polls do not manage to measure adequately. Even more so when, days before the elections, more than a third of the electorate still has high levels of undecided voters. This calls for the testing of new tools that can explain the gap between what people say (rationalized intention) and what they finally do (emotional vote), based, for example, on the model of Daniel Kahneman, winner of the Nobel Prize in economics. For Kahneman, up to 90% of our decisions are not thought out or rationalized (characteristics of System 2), but are quick, intuitive, and implicit (characteristic of the so-called System 1). Tools that work on System 1 complement traditional approaches (generally System 2), seeking to incorporate this implicit and emotional understanding of human behavior.
Beyond the traditional declarative verbalization, it is necessary to understand the voter from an emotional perspective, complementing the more rational one. An experiment that evaluated the role of emotions in the Argentine election allowed us to correctly identify the candidates who participated in the second ballot and to project a 54% voting intention for Javier Milei and 46% for Sergio Massa, values close to the result of the second round. Massa, the candidate of the ruling party, enters the second round of voting for a strategic vote: the majority votes for him for fear of a worse candidate winning. The motivation is purely rational. He also generates a lot of rejection (negative emotions): 45% would never vote for him, compared to 29% who rejected Milei. This meant a ceiling in the second ballot. The vote for the opponent, Milei, was more decisive: 76% voted for him with conviction. This does not mean that the trigger was purely sentimental. The majority chose him for his economic ideas, although some reflected a general disenchantment with politics. In the case of a second round, the studies coincided in indicating that Milei benefited disproportionately from the votes of Patricia Bullrich, candidate of the opposition Together for Change (JxC), who came in third place (and many of the Peronist Schiaretti, placed fourth).
In addition to these usual indicators, other tools made it possible to anticipate Milei’s triumph. One of them is called IRT (Implicit Response Testing), which uses an algorithm to measure the response time and evaluate the level of attachment to a stimulus, in this case, voting intention. Generally, the algorithm yields an average of 70% implicit responses (fast and automatic). In the case of Milei, the implicit response rate was 74%, reflecting a more determined voter base. Voting intention for the rest of the candidates, including Massa, averaged only 56%, a low value reflecting less commitment. The speed of adhesion captures the emotional involvement of voters.
Another tool derived from visual semiology seeks to compare the image of each candidate versus the ideal. Using an image bank with more than 9,000 examples pre-coded in their sentimental valuation, each voter is asked to choose 10 images that express how a certain stimulus, in this case, a candidate, makes them feel. They only select images, whose affective or emotional composition has already been categorized, without having to explain the reason for the choice and thus exempting them from an effort of rationalization. Unconsciously, each individual thus displays a collage of his or her emotions concerning each candidate, and the aggregation of the results of all the individuals makes it possible to compose an emotional map of the candidate. The figures associated with the ideal candidate invoke luminous, positive images, which communicate satisfaction in the face of competence, intelligence, and displays of authority through the candidate’s technical knowledge. But in this second ballot in Argentina, none of the real candidates came close to that ideal, even in their most loyal nuclei. Both Milei and Massa were considered aggressive and with limited dominant authority (that technical authority was somewhat more associated with Milei than with Massa, based on the attribution of expertise in economics). However, the use of this tool revealed that the main difference between the two candidates was the association of Massa with images that conveyed an excess of demands. What voters interpreted with their pictorial referents is that for Massa the presidency embodied a personal project (making the voter feel that he was indebted to their vote in his favor). Massa was detached from a potential contribution to build the future of the country. On the other hand, Milei was linked to an element of vulnerability and closeness, which made him more human.
Electoral processes characterized by high uncertainty, the emergence of an outsider, and the general disenchantment with politics are especially sensitive to the weight of emotions. This makes it difficult to predict results if one starts from an exclusively conventional approach, under the assumption that people react only to a cost-benefit calculation and that their intentions anticipate 100% of their future behaviors. It is time to better understand the global context in which responses at the ballot box occur and to recognize the strength of the emotions that are reflected in the vote.
*Text presented at the SAIMO-CEIM congress in the panel on public opinion studies co-organized by WAPOR Latin America.
Translated by Janaína Ruviaro da Silva from the original in Spanish.