The initial narrative emerging from Brazil’s first-round presidential election results paints a picture of a triumphant far-right resurgence and a surprisingly weak performance from the incumbent left-wing president, Luiz Inácio Lula da Silva. With Flávio Bolsonaro, son of the polarizing former leader, securing 47% of the vote against Lula’s 45.1%, the race now heads to a runoff, a scenario many on the left had hoped to avoid. While Lula himself termed the result "unexpected," a deeper dive into the numbers suggests less of a polling catastrophe and more of a predictable tightening in the final days of a highly polarized contest.
The immediate reaction from some quarters has been to declare yet another failure of pre-election polling. However, a dispassionate examination reveals a more nuanced reality. Many reputable polls had indeed shown Lula with a lead, often placing him in the low to mid-40s, and Bolsonaro in the high 30s to low 40s. A 2-to-3-point shift, while significant enough to push the election into a second round, is well within the margin of error for many national surveys, particularly when considering the dynamic nature of undecided voters and late deciders in a highly charged environment. To frame this as a profound "miss" is to misunderstand the inherent limitations and probabilistic nature of polling itself.
What the results underscore is not the death of polling, but rather the stubborn persistence of certain political currents, even when obscured by public sentiment that leans towards change or stability. The narrative of a decisive Lula victory in the first round was, in many ways, an optimistic projection rather than a firm data-driven certainty. While Lula consistently led in head-to-head matchups, the outright majority required for a first-round victory is always a high bar, especially when a significant portion of the electorate remains deeply entrenched in a populist, often anti-establishment, mindset.
It is precisely this segment of the electorate, often harder to reach and sometimes less transparent with pollsters, that can swing outcomes at the margins. The "shy voter" phenomenon, while often overused as a post-hoc explanation, cannot be entirely dismissed in contexts where social desirability bias might influence responses regarding support for controversial figures. Moreover, the inherent difficulty in accurately modeling turnout in such a politically febrile climate adds another layer of complexity. If the Bolsonaro camp successfully mobilized a segment of the electorate that was underrepresented in polling samples or that decided to participate at the last minute, that alone could account for a percentage point or two.
The notion that Lula’s performance was "unexpected" may stem more from campaign enthusiasm and perhaps a touch of wishful thinking than from a cold assessment of the data. While the left’s hope for a swift resolution was understandable, the numbers always pointed to a tight race, with a runoff a plausible, if not probable, outcome. To express surprise at a 2-point deviation from a central tendency that itself had a 3-point margin of error is to misinterpret what polling promises. Polls offer snapshots, not prophecies.
Looking ahead to the runoff, the dynamics shift. The 47% secured by Bolsonaro is a formidable base, suggesting that his support is far from a fringe phenomenon. Lula’s task now is to energize his base and peel off votes from third-party candidates, a challenge that will require more than just appealing to a broad anti-Bolsonaro sentiment. The coming weeks will test the resilience of both campaigns, but one thing is clear: anyone predicting a landslide based on early projections or even historical trends is once again mistaking a probabilistic indicator for an absolute truth. The data remains stubbornly, and appropriately, humble.