Can we still rely on poll data to help form election scenarios?
Election results in Argentina and Ecuador surprised many analysts. Southern Pulse South America consultant Marco Bastos discusses pollsters, upcoming election scenarios, and lessons learned from yet another “unpredictable” election in Latin America.
Background
In August, both Argentina and Ecuador held national elections and are expecting another round in October.
This is not a new phenomenon - recent elections in Latin America and elsewhere have failed to capture voter sentiment. For example, many pollsters failed to foresee Pedro Castillo’s rise in Peruvian 2021 presidential election. In Chile’s 2021 election, today’s president Gabriel Boric was not considered the favorite within his own left-wing coalition primary. In Brazil in 2022, leading pollsters failed to capture the electoral support of then president Jair Bolsonaro. Still, similar problems have been occurring in the United States (2016 and 2020), in the United Kingdom (2016) and more recently in Spain, where polls envisioned an easy win for conservatives, but instead the election ended almost in a tie.
Reasons that pollsters are failing to predict results primarily include obsolete polling methods, such as calling home phones; a rapidly changing public opinion due to social media and widespread discontent with institutions, or that governments more? broadly are seen as inefficient and political parties with little identity with most of the population -- the rise of the null vote.??
However, even though pollsters seem to generally be failing to capture voter sentiment, they can still be a useful tool for risk analysts so long as they are combined with other methods to design scenarios.
South America analyst Marco Bastos takes a look at upcoming elections in Ecuador and Argentina, how to leverage poll data with a grain of salt, and ways to come up with election scenarios that rely also on other analytical techniques.
Polling inaccuracies
On 13 August, far-right Javier Milei was the most-voted candidate in Argentina's primary elections. Ahead of the primaries, leading pollsters like ESOP, Giacobbe and Opinaia predicted Milei receiving 15%, 18% and 20% of the votes, respectively. Yet, Milei received 30% of the votes, single-handedly beating the center-right coalition Juntos por el Cambio (28.3%) and the incumbent Peronist coalition (27.3%). Most notably, Milei won in the provinces, including those that are typically Peronist strongholds. No pollsters predicted his popularity in the provinces.
A week later in Ecuador, Luisa González led the first round of presidential elections with 33.6% of the popular vote. González, puppet of the former socialist president Rafael Correa, will face lawmaker Daniel Noboa. Noboa, a lawmaker and heir to the country's largest banana producer, received 23.5% of the votes in the first round. Noboa’s father, Alvaro, has a reputation as a corrupt businessman known for his exploitative practices while growing Exportadora Bananera Noboa SA. Yet, that baggage didn’t seem to affect his son, Daniel, who surprised analysts by making it to the second round over pollster and analyst favorite Jan Topic and Otto Sonenholzner.
Milei should have been considered a serious contender since he gained popularity as he protested stay-at-home orders during the Covid-19 pandemic framing government restrictions as socialism. As for Ecuador, our baseline scenario was that González would participate in a run-off. Still, analysts were largely surprised by both Milei and Noboa's results because polls failed to reflect likely candidate performance.
While I anticipated Milei’s strong public support in Argentina, I failed to see Noboa’s rise in Ecuador. Using what I got correct about Argentina and failed to see about Ecuador, I made a list of ten ways to use polls wisely as a risk analyst in Latin America.
1 - Elections are about framing, a.k.a. marketing.?
In politics, whoever delimits the boundaries of what is discussed and how it is discussed has a huge advantage. This approach is widely used by marketers. In post-pandemic Argentina, it has become increasingly more common to see a general dissatisfaction with politicians in general, from all sides of the ideological spectrum. This could suggest that an anti-establishment candidate could garner broad public support. In that sense, Milei’s rise is anything but a black swan.
An anti-establishment figure from the far-right should not be seen as a surprise. The scenario of an unpopular left-wing administration delivering more than 100% annual inflation is, in itself, favorable for the right-wing opposition. In the current case of Argentina, as the previous center-right neoliberal reformist government of Mauricio Macri (2015-2019) also failed to deliver jobs and growth, an anti-establishment option had a fertile ground.?
Moreover, recent signs show that Argentine politics were drifting more to the right. For example, former center-right president Mauricio Macri traveled to the US to be photographed with former US president Donald Trump. Macri’s picture was a piece of evidence that ideology more aligned with the far-right could be accepted by a generally progressive Argentine public. Macri did no favor to his center-right Juntos Por el Cambio coalition by cozying up to Trump. Rather, he enabled the Argentine public to normalize Milei and his far-right ideas.
In the case of Ecuador, my assumption about the growth of the law-and-order candidate Jan Topic was not entirely wrong. Topic, a political outsider, went from being a completely unknown figure to receiving 15% of votes countrywide within just two months on the campaign trail, as security was a relevant issue to most voters. The campaign trail was marked by two high-level assassinations against Manta mayor Agustín Itriago and presidential candidate Fernando Villavicencio. Those crimes reinforced citizen’s perception and real data that security in the country was becoming a major issue.?
Yet, I failed to see alternative scenarios in which other candidates besides Topic could profit out from this public mood. Noboa showed up in the television debate — watched by 82% of Ecuadorians — wearing a bulletproof vest. That likely had a resounding effect on voters and played a role in Noboa’s rapid, final push ahead of voting day. Also, name recognition can’t be ruled out, as the Noboa family is a household brand for Ecuadorians. That might possibly have an effect in older demographics and in Guayaquil’s metropolitan area, birthplace of the family and largest district countrywide.??
2 - Two types of elections
Brazilian scholar Alberto Almeida says that there are two types of elections: “elections of change” and “elections of continuity.” I apply Almeida’s framework to all elections in the region, and not just those in Brazil. A scenario more prone to change is when most voters are unhappy with their country - most of Latin America at this time. On the other hand, elections of continuity happen when citizens are happy with the status quo. Examples of this might include upcoming elections in Mexico in 2024; or El Salvador in 2025, though concerns amount around the democratic process there. This is a basic but still relatively effective predictor, and can be measured by the president's image and qualitative surveys on people's moods toward public affairs.
Argentina’s 2023 election is set to be one of change. In April of this year, 75% of Argentinians were pessimistic about their country. In elections of change, the candidate who can sell himself to the public as the mark of change has the upper hand. In Argentina, Milei is best positioned as the “change,” reinforced by his referring to most other candidates as members of the “political caste.” That is not the case for other favorites such as opposition figure Patricia Bullrich, a cabinet member during the unpopular Mauricio Macri administration, and Sergio Massa, current economy minister.
Ecuador 2023 is also an election of change, as president Guillermo Lasso had a disapproval rating of 64% among respondents in the most recent survey prior to his invoking of muerte cruzda to dissolve congress after multiple impeachment attempts, and later called for snap elections. Contrary to the Argentine scenario, there was no incumbent in this election because the incumbent president’s party did not appoint a candidate.
3 - Elections without incumbents are more prone to surprises
Elections without incumbents are more prone to being elections of change, especially when a mood of social discontent is present. This has to be incorporated into the scenario assumptions. Examples include Colombia in 2022, Peru in 2021, Ecuador in 2021, Chile in 2021 and Brazil in 2018.?
In Ecuador 2023, Noboa presents himself as a third option between the left-wing correismo camp led by former socialist president Rafael Correa’s pick Luisa González.?
Correa finished his 10-year terms in 2017 as one of the most popular leaders in his country’s history. He governed during an unprecedented boom of commodity exports led by Chinese economic growth, which boosted state coffers and allowed for an expansion of social policies. Correa’s support was vital for the victory of his then-ally in the 2017 presidential election, Lenin Moreno. Moreno soon departed from the Correismo camp. Since then, Correa’s public support has fallen due to corruption charges; but his base, primarily made up of those who directly benefited from the expansion of social policies, remains. Any candidate with Correa’s stamp of approval benefits from his popularity.
It surprised me that Noboa’s alignment as a moderate figure who did not enter into the pro- or anti-Correa polarization, gained more traction than Topic’s law-and-order message. The lesson for the future is to leave room for alternative scenarios, especially in elections without incumbents.?
4 - Look for standards
Looking at past Ecuadorian elections, the Correismo camp (whose focal point is now the Revolución Ciudadana (RC) party) has always had at least 30% of votes countrywide in recent history. This is the minimum base that its candidate, Luisa González, was likely to get in the first round of the elections.
领英推荐
However, Milei-backed candidates have performed poorly in the dozens of elections that Argentine provinces have held since February. This fact led some analysts to believe that Milei was losing ground. However, the standard in Argentina is that local elections are about local issues, with people looking for solutions that have an immediate impact on their daily lives. More explicitly, local elections in Argentina are generally less about political parties and more about proposed changes and proven successes. Voters cast their ballots worrying about how well the roads are paved in their neighborhoods, as well as the quality of the local schools and hospitals.??
The analysts who correctly read the historical standard of provincial elections in Argentina, did not extrapolate those local results to determine a national pattern. Voters will not behave the same way in every future election, but standards matter. To gauge the public's mood and how that might affect elections, we might look at how society is framing issues (point 1).
5 - How could the candidate become more popular?
One exercise to consider is how a candidate’s popularity might grow, whether it be through grassroots, territorial support, the backing of a popular union/politician/business group, a successful marketing campaign to capture public opinion trends, or simply a generally good and engaged social media team.?
In Ecuador 2023, Noboa’s fast growth is explained by a mix of having a successful social media strategy and a support network on the ground. The Noboa family is also one of the richest in Ecuador, a fact that cannot be discounted in the gray areas that circle elections in the region. Thus, it was a mistake to rule out that support for his candidacy could grow exclusively due to his stance as an outsider and his father’s reputation. Money can buy both fine-tuned political advisors, and support on the ground.
6 - Social media and Google metrics
Social media metrics have to be followed closely by risk analysts. The number of followers and engagement on various social networks must be measured periodically (weekly, for example). This way, it is possible to diagnose changes in public attention.?
Also, following Google Trends is crucial. With the privilege of hindsight, we can see that Google searches for Noboa increased significantly in Ecuador in the week before the election. Ecuador's electoral law, which prohibits publishing polls 10 days before voting day, makes it impossible for polls to detect these variations. Paying closer attention to the online environment would have helped analysts to better predict Noboa’s rise.
In Argentina, Google Trends shows an enormous public interest in the scandal involving the high-profile Peronist operator and provincial official Martin Insaurralde, a key ally of both Sergio Massa and Vice President Cristina Kirchner. On 30 September, photos of Insaurralde in a luxurious yacht off the Spanish coast with alleged girlfriend and famous digital influencer showing off Cartier bracelets, Louis Vuitton leather goods, and a Rolex watch quickly went viral. Insaurralde has never been anything other than a government official, which makes any lawful explanation of the wealth difficult to come by, which has led to his resignation and the opening of an investigation into the matter. The heated popular interest in that case is almost certain to hurt Massa’s candidacy. Milei, who accuses politicians of being “the caste” is well-positioned to be the beneficiary of the scandal.???
My sources within the milieu of political marketers from across the region say that, in the past few years, social media has influenced voters to make their final decisions closer to election day. In Argentina, 9% decided who they would vote for in the primary elections within a week before voting day, while 13% decided on the morning of the election.?
That makes the task of gathering information more complex for a wide array of professionals, including pollsters, campaign strategists and risk analysts. Once again, leaving room for alternative scenarios is a strategy to fine-tune the analysis.
7 - Check poll methodology
There are multiple different methodologies for polling, each of them with its pros and cons:?
As there is no such a thing as an infallible method, analysts must familiarize themselves with poll methodology for any pollster data they intend to include in their scenario generation.?
8 - Poll aggregators
The most effective way to reduce uncertainty, while still using the polls is to consult poll aggregators. In the first round of Ecuador’s presidential election this year, Calculo Electoral predicted there would be a 72% chance of a run-off between Gonzalez and another candidate.?
In Argentina, poll aggregator La Política Online gave the exact figure for the center-right candidate Patricia Bullrich between polls and primary results. The difference between estimation and real-life results was not far off for Finance Minister Sergio Massa (3,9%), but they were for the anti-establishment Milei (10,7%). Either polls failed to capture voters’ support for Milei, or he gained public support in the final week, or morning of the election. This hypothesis is likely to be correct given the uptick in Google searches for his name in Argentina in the days before the election. In the Brazilian 2022 elections, Southern Pulse successfully predicted a razor-thin election based on data of poll aggregators.
9 - Do not expect polls to predict the future
Election polls must be interpreted as a snapshot of the moment. Comparing the results to previous surveys with similar polling methods helps to see the moving picture of how the public views the issues and candidates at hand. The observation of polling data has to take into account qualitative data about the framing of the election in the view of voters.?
Many countries do not allow polls to be released for a certain number of days before the election — 10 days in both Argentina and Ecuador. This kind of legislation requires analysts to work with little information as elections draw closer and provides a reinforcement for point 6 — monitor social media and Google trends.
10 - Charisma matters
Finally, charisma still matters. Watch how candidates present themselves in public debates, speeches, and on camera in general. If you find yourself captivated by what they said, whether you agree or disagree, it’s indicative that voters might feel similarly. If you find yourself reaching for your phone or struggling to follow the main point of their commentary, it might be indicative that voters find them forgettable or do not believe they are fit to be the leader of their nation.?
Looking back, the US elections in 1960 saw the first televised presidential debates, with John F. Kennedy (JFK) going up against Richard Nixon. JFK looked fresh, young, and most importantly he presented well - both his ideas and physically. Nixon on the other hand, appeared old, tired, and struggled to sound as coherent and convincing on camera. The election was close, but Kennedy won.
The same holds true today. In Ecuador, Noboa is relatively charismatic, as exhibited by his debate performance. Gonzalez, while less charismatic, benefits from Correa’s backing. In Argentina, a team member attended an event in which the top three candidates spoke to the private sector. He noted that when Milei and Massa spoke, you could hear a pin drop, but when Bullrich spoke, she did not command nearly the same attention. This is exhibited in their most recent debate performance on 1 October.
Charisma is not something that can necessarily be measured by data or methodology. It requires an analyst to go with their gut instinct and put themselves in the shoes of a voter that’s on the fence about how they want to vote. Data and media monitoring continue to be great tools, but analysts must remember that it’s only part of the entire, far more complex picture.
Whether it’s Argentina, Ecuador or elsewhere in Latin America, Southern Pulse has the experience, network, and relationships to simplify this challenging region with honest, direct answers to your most complicated questions.
Want to learn more? Let’s chat.
CC BY-ND