Public opinion
The polls, with their technical notes
What each polling firm measured about Yamandú Orsi's government, by what method and on what dates. Within the same eleven-week window, presidential disapproval read anywhere between 49% and 65% depending on who asked. That gap is the finding.
What this is
This page does not evaluate the government. It collects the public-opinion surveys Uruguayan firms published about Yamandú Orsi's administration, each with its technical note: who ran it, between which days it was in the field, how many people it interviewed, by what channel and with what margin of error.
We do this because the figure in circulation usually arrives with none of that. And because when the numbers are placed side by side, something no headline shows becomes visible: over the same window, about the same country, the firms are not measuring the same thing.
Last reviewed: · Our calculation over figures published by third parties. No polling firm publishes this range.
The main finding
Five firms measured the same thing and got different numbers
Surveys fielded between 22 May and 11 August 2026, all on the president's performance.
Five firms asked about the president's performance within an eleven-week window. Approval barely moved between them: from 20% to 26%, six points. Disapproval moved a lot: from 49% to 65%, sixteen points. The net balance — approval minus disapproval, our own arithmetic — runs from −25 to −45: twenty points apart.
June is the cleanest case. Cifra was in the field from the 1st to the 17th and Factum from the 6th to the 20th: near-overlapping windows, the same country, the same week. Cifra reported 65% disapproval and Factum 56%. Nine points.
There is no saying which of the five measures better. We have no way to decide it and we will not insinuate one. What can be said, and what this page publishes, is that a reader who sees 26% in one outlet and 20% in another is not looking at a contradiction: they are looking at two different instruments.
| Firm | Fieldwork | Cases | Mode | Approve | Middle block | Disapprove | Balance |
|---|---|---|---|---|---|---|---|
| Usina de Percepción Ciudadana | 22 May – 26 May 2026 | 500 | automated WhatsApp protocols with telephone monitoring | 24 | 27 | 49 | -25 |
| Cifra | 1 Jun – 17 Jun 2026 | 800 | telephone, landline and mobile | 20 | 15 | 65 | -45 |
| Factum | 6 Jun – 20 Jun 2026 | 900 | mobile telephony | 24 | 20 | 56 | -32 |
| Equipos | 23 Jun – 8 Jul 2026 | 705 | face-to-face, in households | 26 | 21 | 53 | -27 |
| Opción | 16 Jul – 11 Aug 2026 | 806 | telephone | 22 | 28 | 50 | -28 |
Our calculation
What changes between firms is not how many approve: it is how many are left in the middle
Ranked by the size of the middle block — those who neither approve nor disapprove, plus those with no opinion — the five surveys fall in almost the exact reverse order of disapproval: Cifra leaves 15 in the middle and reports 65 disapproval; Factum 20 and 56; Equipos around 21 and 53; Usina 27 and 49; Opción 28 and 50. Approval, meanwhile, barely moves between 20 and 26.
Put another way: the firms that leave less room for a middle answer turn undecideds into disapprovers, not into approvers. The number that gets the headline — disapproval — is the one most sensitive to questionnaire design. The one that varies least — approval — is the one least often headlined.
This is an arithmetic observation of ours over five published points. It is suggestive and it is not a demonstration: with five firms there is no way to submit it to a statistical test, and no firm asserts it.
- The window is not homogeneous: the Usina survey closed its field period on 26 May, before the presidential-pickup episode; the other four were fielded afterwards. It is visible in the dates column, which is why we do not call this a trend.
- Every balance and every range in this section is our own calculation. The firms publish approval and disapproval, not the balance.
The record
Every survey, with its technical note
One row, one survey, ordered by last day of fieldwork. None enters without firm, exact field dates and sample size.
Where a field was not published, the cell says so: we neither estimate it nor leave it blank. The "middle block" column adds those who neither approve nor disapprove to those with no opinion, because not every firm separates the two; the breakdown, where the firm publishes it, sits below the table.
| Firm | Measured | Fieldwork | Cases | Mode | Universe | Margin | Weighting | Approve | Middle block | Disapprove | Balance | Question |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Opción~ | president | 16 Jul 2026 → 11 Aug 2026 | 806 | telephone | national | ±3,5% (95%) | not published | 22 | 28 | 50 | -28 | not published |
| Opción~ | government | 16 Jul 2026 → 11 Aug 2026 | 806 | telephone | national | ±3,5% (95%) | not published | 19 | 24 | 57 | -38 | not published |
| Equipos | president | 23 Jun 2026 → 8 Jul 2026 | 705 | face-to-face, in households | residents across the national territory with 2,000 inhabitants or more | ±3,7% (95%) | sex, age, region, education, employment status, dwelling type and previous vote | 26 | 21 | 53 | -27 | published |
| Factum | president | 6 Jun 2026 → 20 Jun 2026 | 900 | mobile telephony | residents of the country, urban and rural, aged 18 and over; sampling frame the 2023 INE population census | ±3,3% (2 sigmas) | by the October 2024 vote | 24 | 20 | 56 | -32 | published |
| Cifra | president | 1 Jun 2026 → 17 Jun 2026 | 800 | telephone, landline and mobile | national, urban and rural, 18 and over | ±3,4 pp (95%) | not published | 20 | 15 | 65 | -45 | not published |
| Usina de Percepción Ciudadana | president | 22 May 2026 → 26 May 2026 | 500 | automated WhatsApp protocols with telephone monitoring | not published | not published | declares no weighting; sample selected by sex and age strata from national population projections | 24 | 27 | 49 | -25 | published |
| Equipos | president | 21 Apr 2026 → 4 May 2026 | 704 | face-to-face, in households | residents across the national territory with 2,000 inhabitants or more | ±3,7% (95%) | sex, age, region, education, employment status, dwelling type and previous vote | 27 | 25 | 48 | -21 | published |
| Opción~ | president | 21 Apr 2026 → 27 Apr 2026 | 869 | mobile telephony | national | ±3,5% (95%) | not published | 26 | 28 | 43 | -17 | not published |
| Opción~ | government | 21 Apr 2026 → 27 Apr 2026 | 869 | mobile telephony | national | ±3,5% (95%) | not published | 20 | 30 | 48 | -28 | not published |
| Equipos | president | 19 Feb 2026 → 5 Mar 2026 | 704 | face-to-face, in households | people aged 18 and over, resident in the country with mobile telephony (verbatim from the note; see the caveat) | ±3,7% (95%) | sex, age, region, education, employment status, dwelling type and previous vote | 33 | not published | 40 | -7 | published |
| Cifra | president | 18 Feb 2026 → 28 Feb 2026 | 801 | telephone, landline and mobile | national, urban and rural, 18 and over | ±3,4 pp (95%) | not published | 31 | 23 | 46 | -15 | not published |
| Factum | president | 8 Feb 2026 → 22 Feb 2026 | 900 | mobile telephony | residents of the country, urban and rural, aged 18 and over; sampling frame the 2023 INE population census | ±3,3% (2 sigmas) | by the October 2024 vote | 37 | 21 | 41 | -4 | not published |
| Equipos | president | 18 Nov 2025 → 2 Dec 2025 | 1109 | mixed: 706 face-to-face and 403 by telephone | not published | ±2,9% (95%) | not published | 36 | not published | 36 | 0 | not published |
| Cifra | president | 24 Oct 2025 → 3 Nov 2025 | 801 | telephone, landline and mobile | national | not published | not published | 40 | 20 | 40 | 0 | not published |
| Cifra | president | 18 Sept 2025 → 28 Sept 2025 | 803 | telephone, landline and mobile | national | not published | not published | 43 | 24 | 33 | +10 | not published |
| Cifra | president | 9 Apr 2025 → 19 Apr 2025 | 1115 | telephone, landline and mobile | national | not published | not published | 41 | 41 | 18 | +23 | not published |
| Cifra | president | 17 Mar 2025 → 30 Mar 2025 | 800 | telephone, landline and mobile | national | not published | not published | 37 | 48 | 15 | +22 | not published |
- Opción, 16 Jul 2026 → 11 Aug 2026. Fieldwork split in two stretches, 16-20 July and 6-11 August, 403 cases each.
- Opción, 16 Jul 2026 → 11 Aug 2026. Published breakdown: very good 3, good 16, neither 24, bad 27, very bad 30, no answer 1; it sums to 101 because of the source rounding. This 57% is the figure many August headlines presented as presidential approval. It is not: in the same survey, disapproval of the president is 50%.
- Equipos, 23 Jun 2026 → 8 Jul 2026. Middle block: 20 neither approve nor disapprove and 1 does not know or does not answer. The firm publishes the middle block inside the chart, not in the prose of the release.
- Factum, 6 Jun 2026 → 20 Jun 2026. Secondary press reported the field period as "2 to 20 June". We use the house's own note: 6 to 20.
- Cifra, 1 Jun 2026 → 17 Jun 2026. Middle block: 12 neither approve nor disapprove and 3 give no opinion.
- Usina de Percepción Ciudadana, 22 May 2026 → 26 May 2026. Middle block: 22 "are indifferent" and 5 "choose not to answer". Verbatim question: "What is your opinion of the President's performance?". The same report declares a second fieldwork round, 29 May to 2 June, for other questions. Usina does not declare who funds its political studies and is not a member of the industry chamber.
- Equipos, 21 Apr 2026 → 4 May 2026. Middle block: 23 neither approve nor disapprove and 2 do not know or do not answer.
- Opción, 21 Apr 2026 → 27 Apr 2026. Two outlets differ on this wave's middle block: 28 per El Observador and 29 per Búsqueda. We publish the former and leave the discrepancy visible, because the firm's own site does not open to automated access and there is no way to settle it.
- Opción, 21 Apr 2026 → 27 Apr 2026. "Approve" is the sum of very good (4) and good (16); "disapprove", of bad (26) and very bad (22). Both sums are ours.
- Equipos, 19 Feb 2026 → 5 Mar 2026. This wave declares face-to-face methodology and, two lines below, a universe of people "with mobile telephony". It is a carry-over from the mixed design the firm used until 2024, when it declared both universes separately; the April and June waves already show it corrected. We quote it verbatim because it is what was published.
- Cifra, 18 Feb 2026 → 28 Feb 2026. Middle block: 17 neither approve nor disapprove and 6 gave no opinion.
- Equipos, 18 Nov 2025 → 2 Dec 2025. The only Equipos wave with a mixed mode. Not strictly comparable with its own 2026 waves, which are purely face-to-face.
- Cifra, 24 Oct 2025 → 3 Nov 2025. Middle block: 15 neither approve nor disapprove and 5 give no opinion.
- Cifra, 18 Sept 2025 → 28 Sept 2025. Middle block: 17 neither approve nor disapprove; the rest give no opinion.
- Cifra, 9 Apr 2025 → 19 Apr 2025. Middle block: 15 neither approve nor disapprove; more than a quarter had no formed opinion.
- Cifra, 17 Mar 2025 → 30 Mar 2025. Middle block: 19 neither approve nor disapprove; the rest, close to three in ten, had no formed opinion.
Method
Why they differ
Three differences documented by the firms themselves, and one observation of ours.
First: they do not all ask about the same thing. Opción rates the government and the president separately, in the same sample and the same questionnaire. In its August 2026 survey, disapproval of the government is 57% and of the president 50%. Seven points apart, ten on the balance. Placing Opción's 57% next to Factum's 56% compares different objects.
Second: they do not all ask the same way. Opción uses a five-point scale — very good, good, neither, bad, very bad — and the other three an approval question. A scale with two negative boxes and two positive ones does not behave like an approve-or-disapprove.
Third: they do not talk to the same people or by the same channel. Equipos interviews face-to-face in households in localities of 2,000 or more inhabitants; Factum by mobile phone, including rural population; Cifra combines landline and mobile; Usina uses automated WhatsApp protocols. Each mode leaves out a different group, and the international guidelines require declaring those exclusions precisely for that reason.
And fourth, this one is ours: the size of the middle block orders disapproval almost perfectly. It is an observation over five points, not a demonstration.
None of these differences is anyone's defect. They are different design decisions, all declared by the firms.
Trajectory
Each firm's series, on its own
Four series, four charts, one axis. Never a single line.
Overlaying the four series on one chart would manufacture a single trend that no firm published. They go separately, each with its note, on the same vertical axis so they can be read in parallel without the scale deceiving anyone.
What all four do agree on is the direction and the mechanism: the block of people without a formed opinion, which in March 2025 was enormous, tipped towards disapproval rather than approval. In Cifra's series, approval runs 37, 41, 43, 40, 31 and 20, while disapproval climbs steadily from 15 to 65.
They agree on the shape of the curve. They do not agree on the level. Both things are true at once.
Cifra
- 17 Mar – 30 Mar 2025: Approve 37%, Disapprove 15%
- 9 Apr – 19 Apr 2025: Approve 41%, Disapprove 18%
- 18 Sept – 28 Sept 2025: Approve 43%, Disapprove 33%
- 24 Oct – 3 Nov 2025: Approve 40%, Disapprove 40%
- 18 Feb – 28 Feb 2026: Approve 31%, Disapprove 46%
- 1 Jun – 17 Jun 2026: Approve 20%, Disapprove 65%
Factum
- 8 Feb – 22 Feb 2026: Approve 37%, Disapprove 41%
- 6 Jun – 20 Jun 2026: Approve 24%, Disapprove 56%
Equipos
- 18 Nov – 2 Dec 2025: Approve 36%, Disapprove 36%
- 19 Feb – 5 Mar 2026: Approve 33%, Disapprove 40%
- 21 Apr – 4 May 2026: Approve 27%, Disapprove 48%
- 23 Jun – 8 Jul 2026: Approve 26%, Disapprove 53%
Opción — president
- 21 Apr – 27 Apr 2026: Approve 26%, Disapprove 43%
- 16 Jul – 11 Aug 2026: Approve 22%, Disapprove 50%
- The Equipos wave of November and December 2025 is mixed — 706 face-to-face cases and 403 by telephone — while the 2026 ones are purely face-to-face. That point is marked differently.
- Opción's series has two points. Two points are not a trend; they are shown anyway because they are the two surveys with a technical note it published.
- A truncated axis would turn two points of difference into a visual collapse. The charts run from 0 to 100.
Causality
What can and cannot be said about the pickup truck
The case became public on 26 May 2026, through an investigation by the Radio Carve programme Así nos va into the president's asset declaration before the transparency board: an invoice from February 2025 shows the pickup billed at about 54,000 dollars against a list price near 79,000, eight days before he took office. Much of the coverage chained that fact to the fall in approval.
It is worth dating both things properly, because we got it wrong ourselves first: the date we had on file was that of an article published on 1 June, not that of the event. Confusing the date of a story with the date of what the story reports is the easiest error to make on a page like this one.
The firms themselves do not chain them. Cifra warned that its survey was carried out at a particularly sensitive moment. Opción, measuring afterwards, said explicitly that there is a temporal coincidence but that the data do not allow a causal relationship to be confirmed.
Nor do the numbers support it. The only survey whose fieldwork is not entirely after the case is Usina's, which closed its first round on 26 May, the very same day: it was already reporting 49% disapproval. Equipos measured five weeks later, with the case fully in the open, and reported 53%: twelve points below Cifra, which measured before Equipos. If the episode explained the level, those numbers would have to fall in a different order.
On the transparency board, precision matters too: it did not rule on the substance. On 5 June it processed six anonymous complaints bundled into one file and assigned it to a lawyer for study; in late June it requested information from the presidency. It resolved nothing, and headlines saying it "is investigating him" run ahead of the procedure.
We go no further than the firms that did the measuring.
How to read the numbers
What an approval poll does not measure
Written for someone who never studied methodology. None of this is opinion: these are limits of the instrument.
- It does not measure how people will vote.
- Approving of a president's performance and voting for their party are two different questions and give different results. None of the surveys on this page asks about voting intention.
- It does not measure whether a government governs well.
- It measures how many people say they approve. An administration can fall in approval for reasons that have nothing to do with what it did, and rise for reasons just as external.
- The margin of error is a floor, not the total uncertainty.
- The ±3.3% or ±3.7% each firm declares captures only one source of error: having interviewed nine hundred people instead of everyone. It does not capture who could not be reached, who declined to answer, or how question order weighed. A study of 4,221 polls from the last three weeks of campaigning, across 608 US statewide elections — presidential, Senate and governor — between 1998 and 2014, found the real average error was close to twice what the declared margins implied. That is another country and another kind of poll: the logic carries over, the magnitude is not measured here.
- A difference inside the margin is not a difference.
- If a firm moves from 26% to 24% and its margin is ±3.7%, you cannot say it fell. The same holds between two different firms.
- What is done with the undecided changes the headline.
- The share who do not know or have no opinion is not a detail: the reading of one and the same result changes entirely depending on whether that block is 5% or 25%. Here it runs from 15% to 28% depending on the firm.
- Subgroup breakdowns carry more error than the total.
- When an 800-case survey reports what one party's voters think, that percentage is computed over a fraction of the 800. None of the firms publishes the size of those subgroups, so there is no saying how much more error they carry.
- It is not a snapshot of the day you read it.
- Days or weeks pass between the last day of fieldwork and publication. That is why the column that governs this page is the field date, not the publication date.
- And it measures none of what the rest of this site measures.
- A poll does not say how much a public body spent or whom it bought from. That is in the public-procurement record and can be checked contract by contract.
Breakdowns by vote
How it splits by who people voted for
Three firms published breakdowns by previous vote in 2026. None published the size of the subgroups.
These breakdowns are the most quoted and the most fragile part of any poll: they are computed over a portion of the sample, and none of the three firms publishes how large that portion is. They are shown because the firms published them, with the warning up front rather than in a footnote.
Factum
- Frente Amplio voters: 72 approve, 10 disapprove.
- Coalición Republicana voters: 7 approve, 71 disapprove.
- The lowest approval is recorded among 18 to 33 year olds.
Cifra
- Frente Amplio voters in 2024: 41 approve, 34 disapprove.
- Coalition voters: 94 disapprove.
- Non-politicised: 9 approve, around 75 disapprove.
Opción
- Frente Amplio voters, on the government: 39 approve, 28 disapprove, 33 neutral. On the president: 46 and 22.
- Coalition voters, on the government: 4 approve, 81 disapprove, 15 neutral. On the president: 6 and 72.
The three breakdowns are not comparable with one another: the reference vote is defined differently and the object measured changes, being the government in one case and the president in another.
Sympathy and antipathy
Leader image
Two firms measured leader image in June and July 2026. Neither published a table we can reproduce.
What we left out
There is no leader table here, and that is deliberate. The nine-pair table that circulated in July is published by Cifra nowhere: it lives in two outlets that contradict each other — one attributes 34% antipathy to Lacalle Pou and the other "barely 1%" — and building a table naming nine people out of that would be inventing it. The label naming who commissioned that study fell for the same reason: a single outlet supports it, and the technical note for the same fieldwork describes it as multi-client.
What can be published from Cifra is the Orsi series, which the coverage reproduces consistently and which the firm itself returns to in its reports.
Equipos rated around thirty leaders in its 23 June to 8 July field period — 705 cases, face-to-face, ±3.7% — but published the results only inside a chart. Its prose yields ranges, not figures. That is why it has no table either: a table of numbers the firm did not publish would be an invented table.
Orsi's series, according to Cifra
- October 2025 — sympathy 56, antipathy less than a third
- February 2026 — sympathy 51, antipathy 36
- 1 to 17 June 2026 — sympathy 34, antipathy 52
Equipos
- The verbatim question: "I am going to name some public figures, and I will ask you to tell me, first, whether you know them or not, and if you do, your sympathy towards them, on a scale from 0 to 10."
- Recognition above 90%: Lacalle Pou, Orsi, Cosse and Topolansky. Around 90%: Delgado, Bordaberry, Ojeda and Manini Ríos.
- Sympathy: Lacalle Pou above 40% and, according to Equipos, the only one of the thirty with a positive net balance. Orsi and Topolansky just above 30%. Cosse, Bordaberry, Rodríguez and Ojeda above 20%.
- Among Frente Amplio supporters, Orsi and Topolansky draw six out of every ten positive judgements. Among Coalition voters, Lacalle Pou almost three out of four.
- The two firms use different scales: Cifra asks about sympathy, antipathy and indifference; Equipos uses a 0-to-10 scale. They are not comparable point by point.
- The Orsi series is reproduced by the press from Cifra's reports, not read from a report we could open: the firm's site does not answer automated requests.
One survey with a technical note
Ministers
Cifra, fielded 1 to 17 June 2026, 800 cases, landline and mobile telephone, ±3.4 percentage points at 95% confidence.
It is the only 2026 cabinet survey we found with a published technical note. The "no opinion" column is the one to read first: Lubetkin's 23% disapproval does not compare with Negro's 64%, because more than a quarter of respondents do not rate Lubetkin at all.
| Minister | Portfolio | Approve | Middle block | Disapprove | No opinion |
|---|---|---|---|---|---|
| Carlos Negro | Interior | 21 | 6 | 64 | 9 |
| Cristina Lustemberg | Public Health | 39 | 8 | 41 | 12 |
| Mario Lubetkin | Foreign Affairs | 38 | 11 | 23 | 28 |
| Gabriel Oddone | Economy and Finance | 35 | 8 | 43 | 14 |
Nómade published a cabinet rating in October 2025 with percentages to one decimal, but with no field dates, no sample size, no mode and no margin. Under this page's rule it has not a single publishable figure.
One year into the term
How it compares with previous governments
The least bad comparison is one firm against itself, at the same point of the cycle.
Comparing different presidents through different polling firms stacks every problem on this page at once. The cleanest comparison available is Cifra against itself: the same firm, the same question, one year into each government.
It has to be read with a warning up front: only two of those six surveys have a technical note of their own. The other four are figures the firm mentions in later releases. They have a firm and an approximate date, but no field dates, no sample size and no margin. They are marked differently and not drawn as if they were the same thing.
And there is a gap that cannot be filled: Orsi's eighteen-month point is September 2026 and has not happened yet. For Vázquez in 2006 and Mujica in 2011 we found no published, dated survey in that window. The only well-documented year and a half is Lacalle Pou's in September 2021, with three firms.
| President | When | Approve | Disapprove | Balance | Evidence |
|---|---|---|---|---|---|
| Jorge Batlle | 2001 | 27 | not published | — | quoted, no technical note |
| Tabaré Vázquez (I) | March 2006 | 56 | 17 | +39 | quoted, no technical note |
| José Mujica | March 2011 | 39 | not published | — | quoted, no technical note |
| Tabaré Vázquez (II) | March 2016 | 39 | not published | — | quoted, no technical note |
| Luis Lacalle Pou | 18 to 28 February 2021 | 64 | 25 | +39 | with technical note |
| Yamandú Orsi | 18 to 28 February 2026 | 31 | 46 | -15 | with technical note |
The only well-documented eighteen-month point
| Firm | When | Cases | Margin | Approve | Disapprove | Balance |
|---|---|---|---|---|---|---|
| Equipos | 22 September to 1 October 2021 | 500 | ±4,3% | 52 | 33 | +19 |
| Cifra | 26 August to 3 September 2021 | 707 | not published | 56 | 32 | +24 |
| Factum | 1 to 14 September 2021 | 900 | ±3,3% (2 sigmas) | 61 | 35 | +26 |
What the firms themselves say
The three statements that follow belong to the polling firms, not to us, and they differ from one another because they start from different numbers.
Equipos · June 2026 report
«as a rule, presidents have reached their second year in office already with negative balances, with the exceptions of Vázquez in 2006, Mujica in 2011 and Lacalle Pou in 2021»
Cifra · June 2026 report
«only President Batlle, during the 2002 crisis and its aftermath, recorded lower approval»
Opción · on its April 2026 survey, via Búsqueda
«the level of support for the Orsi administration is the lowest recorded by a Frente Amplio government»
No continuous Factum line can be drawn through 2021: there its middle block is two to four points, while in 2016 and 2026 it is twenty to twenty-eight. Either the question or the base of calculation changed, and we could not determine which. Until that is resolved, its series is cut.
Context
What people call the country's main problem
Three firms, three different question formats. Shown separately and never charted together.
These three surveys are not comparable with one another, and the reason is in plain sight: Equipos counts security as a first or second mention and the other two do not state that criterion; each firm also groups the categories differently. A bar chart putting them together would be manufacturing a comparison.
Cifra
Insecurity, crime and drug trafficking 46; the economy around 33; unemployment around 19; everything else below 5.
Equipos
Security 58 as first or second mention; unemployment around a third; then the economy, social problems and education.
Factum
Security 53; the economy 10; work 9. It is a single-mention question about first place, not a concern scale: the percentages split the top spot. The full list exists only inside images and we could not recover it.
Performance against expectations
| Answer | % |
|---|---|
| Above expectations | 1 |
| As expected | 27 |
| Below expectations | 49 |
| Expected nothing | 22 |
Empty module
Voting intention
No surveys published in 2026
This module is empty on purpose. In 2026 none of the firms we follow published a national voting-intention survey.
Everything search engines return with voting-intention figures belongs to the 2024 election cycle. One concrete, checkable case: an article dated 14 August 2026 republishes a Nómade survey with data from 8 August 2024. The publication date deceives; the data date does not.
Another case, this one from an automated search: a summariser returned a supposed third two-month period of 2026 for departmental voting intention in Montevideo, citing Factum pages that are in fact from 2025. That figure does not exist.
When some firm publishes a 2026 survey with a technical note, it goes here. In the meantime this section shows less rather than filling the space.
Criteria
How we choose what to show, and what we leave out
The criterion is published before the data, and the exclusion list carries names and reasons.
We did not invent our own criterion: we apply the same technical-note standard the polling firms grouped in CEISMU, the industry chamber, committed to in 2017, cross-checked against the ESOMAR and WAPOR guidelines and AAPOR's disclosure standards. It is worth saying what that commitment is and is not: private self-regulation, not law — no sanction, no enforcing authority — and it binds only member firms. Several houses that publish polls in Uruguay are not members, among them Usina de Percepción Ciudadana, which appears on this very page. Even so it is the floor the industry set for itself, and if we publish what it asks to be published, we are not the ones deciding what counts as a serious poll.
That union yields the note we ask for: polling firm; who funded or commissioned it; start and end field dates; publication date; sample size; universe and coverage exclusions; sampling frame; sampling type; collection technique; margin of error with its confidence level; verbatim question wording; weighting variables; the share of don't-knows; and the link to the full note at source.
Of those fields, three are a condition of entry: firm, exact field dates and sample size. The rest are displayed and, when missing, say "not published".
If you find an error on this page, write to us. Correcting in plain sight is part of the criterion.
What we left out
| Survey | Reason |
|---|---|
| Nómade Consultora, its entire output | It publishes no technical note in any coverage we found: no field dates (only "winter"), no sample size, no mode, no margin. All that can be stated is that its last political release is from October 2025. |
| Grupo Radar | We found no 2026 political survey of theirs in open sources and their news section returned an access error. That does not license writing that Radar did not measure: all that can be stated is that we did not find it. |
| Opción, first quarter of 2026 (government 29/33, president 26/43) | They circulate with no field dates and no sample size. The Opción series that holds up starts in April 2026. |
| Factum, June 2025 wave (46/31/22) | It did not publish exact fieldwork dates. |
| Factum, Montevideo mayor, second two-month period of 2026 (Bergara 23/30/46) | We could not recover the technical note; it also measures a different office and does not belong to this series. |
| The nine-leader sympathy and antipathy table attributed to Cifra | Cifra did not publish that study on its site and the two outlets that carried the table contradict each other: one attributes 34% antipathy to Lacalle Pou and the other "barely 1%". A table naming nine people built on that would be an invented table. The label "commissioned by the Partido Nacional" fell too: a single outlet supports it, and the technical note for that same fieldwork describes the study as multi-client. |
| CB Global Data, presidential image, August 2026 | It is the most recent measurement in circulation (fielded 5 to 10 August, 37.5% positive image and 62.8% negative) and we name it so as not to hide it, but it does not qualify: it publishes no sample size, it measures image rather than approval, and it is a regional online panel across eighteen countries. Two outlets also place Uruguay at different ranks in the same table. |
| Equipos, leader image person by person | The house published the results only inside a chart. We show the ranges it wrote in prose, not a table reconstructed by eye. |
| Equipos, breakdown among Frente Amplio voters (61→45 and 13→28) | The release does not say which two surveys those points belong to. |
| Any press figure that contradicts the house's own note | A real case: a radio station reported the Equipos sample as "2000". That number is the universe threshold — localities of 2,000 or more inhabitants — not the sample size, which is 704 cases. |
| Any reconstructed verbatim question | One of our searches returned a Cifra question labelled "reconstructed from context". That is not a source. |
| opuy, the aggregated Uruguayan public-opinion data package | It is the best historical baseline available (1989-2023, six firms, MIT licence), but it carries no sample size, margin, mode, question or sponsor, its date field mixes last day of fieldwork with publication date, and its last change predates this government. |
How to read this page
Everything here was measured by others. We do not poll, weight, correct or average: we transcribe the figures each firm published alongside the technical note it published them with, and we link to the original.
The only things we add are two calculations, labelled every time they appear: the net balance — approval minus disapproval — and the ranges between firms within one field window. No firm publishes those two numbers.
We cannot say which firm measures better, and we do not insinuate it. We cannot claim a fall is due to one particular event: none of these surveys measures that, and the firms themselves warn against it.
Nor does this page evaluate the government. That is what the rest of the site is for, working on the public-procurement record, checkable contract by contract.
If a figure here does not match the source, the source is right and we are wrong. Write to us and we will correct it with the date in plain sight.
Sources
Each row links to the firm's own release where we could open it, and to the press that reproduces the technical note where we could not.
The polling firms
- Factum — aprobación, tercer bimestre de 2026 (ficha completa)
- Factum — segundo bimestre de 2026
- Factum — primer bimestre de 2026, con cortes por voto
- Factum — evaluación contra expectativas (24/07/2026)
- Equipos — junio-julio de 2026, con la pregunta textual
- Equipos — abril-mayo de 2026
- Equipos — febrero-marzo de 2026
- Equipos — noviembre-diciembre de 2025 (ola mixta)
- Equipos — imagen de unos treinta dirigentes (29/07/2026)
- Cifra — junio de 2026
- Cifra — febrero de 2026
- Cifra — Lacalle Pou al año de mandato, febrero de 2021
- Opción — tercer trimestre de 2026, vía la diaria
- Opción — segundo trimestre de 2026, vía El Observador
- Usina de Percepción Ciudadana — mayo de 2026, vía UyPress
The criteria
- Protocolo de ficha técnica de CEISMU, publicado por la propia cámara
- CEISMU — nómina de empresas asociadas (consultada en agosto de 2026)
- La misma reproducción del protocolo en el portal de Factum, con fecha 23/11/2017
- ESOMAR/WAPOR — Guideline on Opinion Polls and Published Surveys (2014)
- AAPOR — Disclosure Standards
- Shirani-Mehr, Rothschild, Goel y Gelman (2018) — Disentangling Bias and Variance in Election Polls
- opuy — datos de opinión pública uruguaya, 1989-2023