The metamorphosis of power in Alfredo Chaves:we no longer live like our parents

We no longer live like our parents

Five losses and the turnaround

One election, same political base

TSE's Open Electoral Data Repository

• votacao_secao_YEAR_ES.csv: votes by voting precinct, 2004 to 2024 (from 1,229 rows in 2004 to 2,089 in 2024, after the municipal filter)
• votacao_candidato_munzona_YEAR_ES.csv: consolidated result per candidate, 2004 to 2024 (45 to 73 records per election, adding mayoral, vice-mayoral and council candidates)
• votacao_partido_munzona_2020/2024_ES.csv: party-line vote totals, 2020 and 2024 only (9 and 11 records)
• detalhe_votacao_munzona_YEAR_ES.csv: turnout, abstention, blank/null votes, 2004 to 2024
• prestacao_de_contas_eleitorais_candidatos_YEAR.csv (revenue and paid expenses, 2004 to 2024): extracted via HTTP range requests from a ~1.3 GB national file, without downloading the whole thing
• bem_candidato_YEAR_ES.csv: assets declared by candidates, 2008 to 2024
• perfil_eleitor_secao_YEAR_ES.csv: electorate profile by precinct (gender, age bracket, education), 2008 to 2024
• consulta_cand_YEAR_ES.csv: candidacy registry, 2004 to 2024 (gender, date of birth, education and occupation; race/color is only recorded from 2016)

Additional sources

Several sections of this study cross-reference or cite sources outside the TSE repository, always identified where they appear:

• IBGE, SIDRA: table 6579 (estimated resident population, section 06) and tables 9514 and 9605 of the 2022 Demographic Census (sex and race/color, section 06).
• IBGE, Territorial Meshes: the municipality's official boundary, used in the polling-location map (section 09).
• Alfredo Chaves city hall: the official list of districts and localities, used to assign each voting precinct to a district (section 04).
• TSE, pesqele‑divulgação: the election-poll registration system (Law 9,504/97, Article 33), consulted for the 9 polls from 2024 (section 08).
• Alfredo Chaves City Council, SPL: the legislative record consulted for Commendation Motion No. 003/2024 (section 06).
• The candidacy's official Instagram (@hugoluiz11_): the 2024 post history, the only source in this study that isn't public, auditable data (section 07, with caveats detailed there).
• IPESES: a post-election opinion poll from March 2026, cited as additional context, outside the TSE's electoral scope (section 08).

Data-processing pipeline

The raw extracts go through five steps before any comparative analysis: a standard ingestion, cleaning, and load pipeline, implemented in Python (pandas), with the analytical layer in SQL (SQLite). A SQLite database gathers all six elections (votes by precinct, candidates, turnout and districts) and five SQL queries reproduce the study's central results; a script checks each result against the pandas pipeline. The financing, assets, profile and IBGE analyses read the filtered CSVs directly in pandas, with the same cleaning rules. The interactive charts illustrating the study are written in TypeScript (React and Recharts) and styled with CSS.

Cleaning and normalization (pandas)

-- precincts won by the group, by year (sql/02_secoes_vencidas.sql, 2004 to 2024)
SELECT ano, COUNT(*) AS secoes,
       SUM(CASE WHEN votos_grupo > 0 AND votos_grupo = max_secao THEN 1 ELSE 0 END) AS secoes_vencidas
FROM v_prefeito_secao   -- valid votes: excludes blank (95) and null (96) ballots
GROUP BY ano
-- result: 2020 = 1 of 36 precincts; 2024 = 42 of 42

Comparative analysis (SQL)

-- winner of each precinct, by year (highest nominal vote count for Mayor)
SELECT NR_SECAO, NM_VOTAVEL, SUM(QT_VOTOS) AS votes
FROM secao_2024
WHERE DS_CARGO = 'Prefeito' AND NR_VOTAVEL > 0
GROUP BY NR_SECAO, NM_VOTAVEL
-- (identical for secao_2020; winner = MAX(votes) per NR_SECAO)

Joining 2020 ⋈⋈ 2024 by precinct number allows comparison of only the 36 precincts present in both elections (6 new precincts were created by 2024, reflecting electorate growth, and enter the analysis only as a 2024 result).

From a broad defeat to a win in every precinct

In 2020, this political base's majoritarian candidacy won just 1 of 36 comparable precincts (precinct 9), taking 40.0% of the valid vote citywide (3,681 votes), against 56.4% for the winning candidate. In 2024, the same base, now under the coalition “Renovação e Juventude” (“Renewal and Youth”; Republicanos / PP / MDB / PSDB‑Cidadania federation), won all 42 precincts in the municipality, with 58.1% of the valid vote (5,779 votes).

Of the 36 comparable precincts, 31 moved from a minority (<50%) to an absolute majority (≥50%) of the valid vote between the two elections, a swing averaging 17.9 percentage points in those precincts.

A different party every election

The consulta_cand data shows the group's candidate never ran under the same party twice in twenty years: PT in 2004, PSDB in 2008, PSD in 2012, PMDB in 2016, Republicanos in 2020, PP in 2024. The ideological spectrum shifted too, from a left-wing party in 2004 to center and right-wing parties from 2012 on, tracking the national realignment of Brazilian politics over the period. It doesn't look like an isolated local strategy choice.

Coalition size, though, doesn't explain the win on its own: in 2016, the candidate ran with the largest coalition of the whole series (9 parties) and still lost, with 48.0% of the vote. In 2020, with the smallest coalition (3 parties), he also lost. The 2024 coalition (5 parties: Republicanos, PP, MDB and the PSDB‑Cidadania federation) is mid-sized within the series, neither the largest nor the smallest.

On the opposing side, the pattern is the reverse. In the five elections this group lost, between 2004 and 2020, the winner was always PSB: Dr. Fernando took 2004, 2008, 2016 and 2020, and Roberto Fiorin, also PSB, took 2012, four terms and two different names under the same party label, twenty years straight. That streak only breaks in 2024, when Rolmar Boteccia tries to keep PSB in power and loses to Hugo Luiz's PP. In other words, while the losing side changed parties every election, the winning side had a stable label for two decades. It was precisely that stability that broke in 2024, along with the result.

Where the group was strong, where it was weak

Alfredo Chaves has 7 official districts: the seat and six rural districts (Crubixá, Ibitiruí, Matilde, Ribeirão do Cristo, Sagrada Família and São Bento de Urânia). We cross-referenced each precinct's polling location, and through it its district, with the group's candidate's vote share in each of the six elections, to see whether the turnaround was even across the territory or concentrated in one corner of the municipality.

Source: votacao_secao, TSE, all six elections. Each precinct's district determined by its 2024 polling location, cross-referenced with the official locality list from the Alfredo Chaves city hall. Precinct numbers are stable over time in this municipality (only new precincts are added, none are renumbered), so each precinct's 2024 location applies to every year it existed. Percentages weighted by each precinct's total valid votes. District boundaries: IBGE, 2022 Census district mesh. The division into seven districts (Alfredo Chaves, shown here as “Sede”, the district seat; Crubixá, Ibitiruí, Matilde, Ribeirão do Cristo, Sagrada Família and Urânia, listed as São Bento de Urânia by the city hall) appears in the Municipal Basic Sanitation Plan, item 5.1 (Municipal Law No. 633/2017, repealed by Law No. 668/2019), based on IBGE.

What about the council race?

The same precinct→district mapping, now for the group's full council slate (every candidate in the coalition, elected or not — 35 names in 2020, 30 in 2024, identified by registered party), as a % of valid council votes in each district. Only possible for 2020 and 2024: the only two elections where the family's candidacy also ran a council slate.

Source: votacao_secao and consulta_cand (fields SG_PARTIDO, SQ_CANDIDATO), TSE, council race, 2020 and 2024. Same precinct→district mapping as Figure 6. The group's slate is defined by each year's mayoral coalition parties (Republicanos/PTB/Patriota in 2020; Republicanos/PP/PSDB in 2024), matched by SQ_CANDIDATO — not by name, which varies between the candidate and precinct extracts.

In absolute numbers

The percentages in Figures 6 and 6.1 hide the size of each district. The charts below show the total mayoral votes in each election from 2004 to 2024, with a line for the group's candidate, another for the main opposition and dots for the other candidates; the list above the charts names everyone. Each district has its own scale: Sede holds about two thirds of valid votes, and the other six districts are too small to share the same axis.

Source: votacao_secao, TSE, Mayor race, 2004 to 2024, blank and null votes excluded. Same precinct→district mapping as Figure 6.

The political flip in the City Council

Alfredo Chaves' City Council has 9 seats. In 2020, Hugo Luiz Picoli Meneghel's political base, then a council candidate for PTB on the majoritarian coalition “Mudança com Transparência” (PTB / Republicanos / Patriota), elected only 3 council members against 6 from opposing parties (the coalition that won the mayoral race: PP / PSB / PSDB, among others): a clear minority. In 2024, with campaign coordination extended to every council candidate on the “Renovação e Juventude” coalition (Republicanos / PP / PSDB), the slate's bloc grew to 5 of 9 seats, a simple majority, if a tight one.

Adding up the nominal vote for every council candidate on the slate (elected and not), the growth shows up in the vote count too, not just the seats:

Who ran each election

The mayoral vote pulled the council vote, but not equally in both elections

Mayor and council are separate votes on the same ballot: a voter can pick the group's candidate for mayor and an opponent for council, or the other way around. We cross-referenced, precinct by precinct, the share of valid votes for the group's mayoral candidate with the combined share of valid votes for the group's council candidates, in 2020 and in 2024.

Source: votacao_secao, TSE, 2020 and 2024. “Group,” for council races, defined by the same parties as the methodological note below. Pearson correlation (r) on the share of valid votes per precinct; includes straight-party votes (a vote for the party with no named candidate), attributed to the party voted for.

What it cost, and who showed up to vote

Data from the TSE's electoral campaign finance filings (revenue and expenses declared by every candidate, opponents included) and its turnout breakdown (attendance and abstention) round out the picture: the turnaround didn't come from a bigger electorate, and the revenue raised by the group's candidacy was only the largest among the competitors in 2024. In the five earlier defeats, the winner declared 1.5 to 4.4 times more revenue than the group's candidate. Even so, the group's campaign converted the money into votes more efficiently than any other candidate for the office in 2008, 2012, 2016, 2020 and 2024; in 2004 it was the second most efficient. The two figures below show every mayoral candidate of the six elections.

Turnout doesn't explain the 2024 turnaround, but it's been falling for twenty years

Between 2020 and 2024, the registered electorate grew just 4.6% (12,294 → 12,860) and turnout stayed practically flat: 79.15% in 2020, 79.74% in 2024. Since the slate's candidate grew 57% in absolute votes (3,681 → 5,779) while the total electorate barely moved, the turnaround was predominantly conversion of voters who already voted, not mobilization of new ones.

Across all six elections, though, turnout isn't flat: it fell steadily, from 91.8% in 2004 to the 79-80% range of 2020 and 2024. The registered electorate grew 39% over that same period (2004: 9,247; 2024: 12,860), but the number of people who actually voted grew far less, 20.8% (2004: 8,492; 2024: 10,255), because a growing share of registered voters stopped showing up.

Source: detalhe_votacao_munzona, TSE, mayoral race, all six elections (2004 to 2024).

Electorate profile: almost nothing changed

The electorate's demographic profile (perfil_eleitor_secao data, TSE) also stayed stable over time. The gender split has been slowly converging toward parity every election: 51.7% men / 48.2% women in 2008, 51.1% / 48.8% in 2012, 50.6% / 49.3% in 2016, 50.2% / 49.7% in 2020, and 49.9% / 50.0% in 2024 (the TSE never published this figure for 2004). Practically fifty-fifty all along, with a steady trend toward a slight female majority.

Cross-referencing with IBGE: the electorate grew, the population didn't

To understand whether the electorate's growth (9,247 registered in 2004 → 12,860 in 2024, +39%) reflects the municipality growing, we cross-referenced TSE data with IBGE's annual population estimates (SIDRA, table 6579, municipality code 3200300). The answer is no: Alfredo Chaves' estimated population stayed practically flat over the period (14,113 in 2004, ~14,373 in 2024), with dips in some years. The electorate didn't grow because the municipality grew. It grew because an increasing share of the population already living there became registered to vote.

Sex, race and income: do candidates look like the people who live here?

IBGE's 2022 Demographic Census (SIDRA, tables 9514 and 9605) makes it possible to go beyond total population and cross-reference sex and race/color with who actually runs for office. Alfredo Chaves' population is 49.3% women and 50.7% men, 61.4% white, 34.3% mixed-race (“pardo”) and 4.2% Black, numbers very close to the registered electorate's, which is also moving toward gender parity. Among candidates, though, the picture diverges: of the 69 council candidates in 2020, 37.7% were women; in 2024, out of 70 candidates, that share fell to 32.9%, the opposite of the electorate's trend, which keeps moving closer to 50%. Mayoral candidates are left out of these percentages: there are only three per election, and in both 2020 and 2024 all three were men.

The racial composition of the council candidacies follows a similar pattern. In 2020, white candidates were 59.4% of the total, close to the municipality's population share of 61.4%. By 2024 that share rose to 68.6%, while mixed-race candidates fell from 30.4% to 22.9%: a drift away from Alfredo Chaves' racial composition, even as the share of Black candidates grew slightly (7.2% → 8.6%). Among mayoral candidates, all three in 2020 were white; in 2024, two of the three were white and one was mixed-race.

What the posts show about the race

This section steps outside the TSE dataset and uses a different source: the post history of the candidacy's official Instagram account (@hugoluiz11_), from the day of party affiliation through the election result. It isn't public, auditable data like the rest of this study. It's the campaign's own archive, reviewed post by post from screenshots exported from the app, with date, views, likes, comments and shares for each publication. We treat these numbers the way the campaign team saw them in real time: a measure of organic reach and engagement, not voting intention.

The timeline covers six milestones: affiliation to Progressistas (PP) on March 27, 2024; the pre-candidacy launch on June 7; the running-mate announcement for Cleber Bianchi, at the August 3 party convention; the public dispute over a suspended poll on September 22; the win on October 6; and the thank-you post for 5,779 votes, published hours after the result: the highest-engagement post of the entire campaign.

We classified every one of the 166 posts identified by content type, to understand not just when reach grew, but what the campaign was actually posting. Street events and walks are the most common format (nearly a third of posts), followed by policy proposals, coalition posts, and routine administrative communication from the council-member term (City Council session notices, which the campaign posted in pairs: an invite beforehand, an explainer video after). But volume isn't the same as engagement: voter testimonials, which appear far less often, engage well above average, more than four times routine administrative communication.

Who the candidates were, beyond the votes

consulta_cand and bem_candidato data (TSE) bring age, education, occupation and declared assets for each candidate: public information, required by law at every election. Table 8.1 gathers the 16 mayoral candidates of the six elections, from 2004 to 2024, and Figures 23 and 24 compare them. The contrast with Hugo Luiz is the sharpest in the entire study.

Hugo Luiz was 25 on election day in 2024, holds a law degree from UFES, and, per the TSE's records, is white and single. The other two mayoral candidates that year were 61 and 71.

The 9 registered election polls

The TSE maintains its own system for registering election polls (pesqele‑divulgacao.tse.jus.br), required under Article 33 of Law 9,504/97. For Alfredo Chaves in 2024, 9 polls are on record, from 5 different firms, between April and September. We opened and read all 9 registration filings in PDF. It's worth noting a real limit of the public data: the TSE doesn't publish the result (voting-intention percentage per candidate) in these filings, only formal confirmation of registration (firm, protocol, date, scope). It's the commissioning firm that releases the result, typically through a local press outlet.

We located the original document for 6 of the 9 polls: the full reports each institute produced for its client, with methodology, sample size and margin of error. Two of them (Apr 16 and Sep 21) had also been released to the press; the other four are real data, but restricted in circulation, never published by a press outlet:

• Apr 15–16, 2024: Instituto Solução (TSE ES‑09808/2024), 403 respondents, 4.9 p.p. margin, commissioned by AQUI NOTÍCIAS.COM (released).
• Aug 29–31, 2024: Instituto Veritá (TSE ES‑08806/2024), 352 respondents, 5.5 p.p. margin, commissioned by Marataízes Piúma Iriri Internet Ltda. (not publicly released).
• Sep 17, 2024: no institute or TSE registration number appears on the document (an atypical result, see the note below).
• Sep 20–21, 2024: Inove Consultoria (TSE ES‑04825/2024), 574 respondents, 4.0 p.p. margin, commissioned by H.A. Comunicações Ltda. (released by Montanhas Capixabas).
• Sep 27–29, 2024: Ipopes Pesquisa de Opinião Pública, 400 respondents (not publicly released).
• Oct 1–2, 2024: I9‑Inove Consultoria (TSE ES‑06358/2024), 373 respondents, 5.0 p.p. margin, commissioned by Portal ES Agora (not publicly released).

The six real-data points don't share one calculation basis: the Aug 29–31 poll and the official Oct 6 result sum to 100% among the three named candidates, but the Sep 20–21, Sep 27–29 and Oct 1–2 polls (the three 3-way polls with a gap) leave 9.6 to 23.75 percentage points unassigned, a sign those three report the share of the total sample (undecided included), not the share among named candidates. We normalize those three to sum to 100% among the names polled. The Apr 16 poll is different: it only had 2 names (Boldrini wasn't a candidate yet), and normalizing a 2-way race would artificially inflate both existing names, so that point appears with its raw value, as reported in the original document.

After the election: approval of the administration

This data comes from opinion polling, not the ballot box, and so falls outside the official TSE-based pipeline that underlies the rest of this study. Still, it helps round out the picture. In March 2026, already in the second year of the term, the Espírito Santo Institute for Research and Statistics (IPESES) surveyed 405 voters across the seat and districts of Alfredo Chaves (Mar 11–12, 2026, margin of error 4.9 p.p., 95% confidence, quotas by sex and age).

Sources: Aqui Notícias, Imprensa Capixaba and Informe Capixaba, all from March 2026, reporting the same IPESES poll.

Precinct by precinct, with polling location

The complete table of the mayoral candidacy's valid-vote share in every voting precinct, across both elections, plus the physical polling location of each precinct in 2024 (school, community hall, or church, per the TSE's official registry). Rows highlighted in green mark precincts that moved from a loss to an absolute majority. Raw vote counts and totals per precinct are in the repository's underlying data; here, the percentage already tells the story.

The 18 polling sites of the municipality

Alfredo Chaves' 42 voting precincts are spread across 18 physical polling locations, between the municipal seat and rural districts, all identified in the TSE's official registry:

Map of polling locations

The numbers on the map correspond to the # column in the table above. Because most rural addresses in the TSE registry have no street number (just “Main Street, no number”), the sites in the municipal seat (markers clustered close together, at the center of the map) are approximate, lightly spread apart just so they don't overlap; the rural-district sites correspond to the identified locality.

Click a number on the map to see the polling location.

Base map: © OpenStreetMap contributors. Coordinates from geocoding (Photon/Komoot) the location and district names in the TSE's registry.

What the data shows

What the numbers allow us to imagine for 2028

# analysis/cenarios_2028.R (base R, no extra packages)
m_aptos <- lm(aptos ~ ano, data = compar)            # linear trend of the electorate
pred    <- predict(m_aptos, data.frame(ano = 2028),
                   interval = "prediction", level = 0.80)
votos_validos <- aptos * comparecimento * validos_sobre_comp

share_grupo    <- share_2024 + oscilacao              # uniform swing, in p.p.
share_oposicao <- 100 - share_grupo - terceiro        # third candidate: 0, 5 or 10%
margem <- (share_grupo - share_oposicao) / 100 * votos_validos

What this study cannot show

Skills, one by one, and where they show up