Is Catalan Pro-Independence Sentiment at a Low? What the CEO Microdata Say, Corrected
The 31% behind this week's "lowest in a decade" headlines comes from the October 2025 barometer of the CEO (Centre d'Estudis d'Opinió, the Catalan government's polling institute), and it answers a different question from the one that gave 48% in 2017. Comparing the same question and correcting the sample with real recalled vote, support for independence has fallen by just over four points since 2017: from 41% to 37% in October 2025. Its low point came between 2022 and 2025, and in the latest barometer, from September 2026, it stands at 39%.
22 October 2026 · Miguel · Leer en español
This week the Institut d’Estudis de l’Autogovern (IEA), the Catalan government’s institute for self-government studies, presented its Informe Autogovern 2025, and several outlets reported that support for independence has fallen to 31%, “the lowest point of the last decade”, against 48-49% in 2017. We went to the microdata of the survey those figures come from and reanalysed them.
Where the 31% comes from
The report does not run its own survey. Its figures (31% for an independent state, 33% for the current autonomous community, 23% for a state within a federal Spain, and 13% for “region of Spain” plus those who do not answer) match, to the decimal, the October 2025 Baròmetre d’Opinió Política of the CEO (Centre d’Estudis d’Opinió, the Catalan government’s polling institute): 30.8%, 33.2%, 22.9% and 13.0%, according to the microdata the CEO itself publishes.
So this is a year-old figure. The CEO has since published two more barometers, in June and September 2026.
Two different questions
The CEO asks two questions about independence, and they do not measure the same thing:
- “Do you want Catalonia to become an independent state?”, answered yes or no.
- “What should Catalonia be?”, with four options: a region of Spain, an autonomous community, a state within a federal Spain, or an independent state. Here some people prefer the federal state but, forced to choose between yes and no, say yes.
The 48-49% of 2017 is the yes to the first question. The 31% of 2025 is the fourth option of the second. Using the CEO’s own weights, this is the comparison between like questions:
| CEO barometer | Yes to an independent state | “Independent state”, among four options |
|---|---|---|
| October 2017 | 48.7% | 40.2% |
| October 2025 | 39.4% | 30.8% |
| September 2026 | 43.4% | 32.9% |
Before correcting anything, the fall between 2017 and 2025 is nine points on both questions, not eighteen.
The sample has too many voters of some parties and too few of others
Every survey reaches some profiles more than others. A common way to detect this is to compare respondents’ recalled vote with the actual result. In the October 2025 barometer, among those who say they voted in the 2024 regional election:
| Party | Recall in the CEO sample | Actual result in 2024 |
|---|---|---|
| PSC (Catalan Socialists) | 33.2% | 28.6% |
| ERC (pro-independence) | 22.1% | 14.0% |
| Junts (pro-independence) | 15.9% | 22.1% |
| Comuns (left-wing, Sumar’s Catalan partner) | 7.3% | 6.0% |
| Vox (far right) | 6.0% | 8.1% |
| CUP (left-wing pro-independence) | 5.7% | 4.2% |
| PP (conservative People’s Party) | 5.7% | 11.3% |
| Aliança Catalana (far-right pro-independence) | 2.8% | 3.9% |
Percentages of votes cast for a party list. In addition, among those who remember what they did, only 30% say they did not vote, whereas abstention among residents was 42%.
The sample over-represents ERC and PSC voters and under-represents those of the PP and Junts. Because support for independence differs sharply between electorates (from 80% among ERC voters to 8% among PP voters), this composition moves the result.
The corrected series
We reweighted every barometer since 2015 so that its recalled vote reproduces the actual result of the preceding regional election (each party’s votes and abstention), without touching its breakdown by sex, age and province. We also computed it correcting only the split between parties, without adjusting declared abstention.
| Barometer | Yes: CEO | Yes: corrected | Independent state: CEO | Independent state: corrected |
|---|---|---|---|---|
| October 2017 | 48.7% | 41.4% | 40.2% | 34.0% |
| October 2018 | 47.2% | 39.8% | 38.9% | 32.7% |
| October 2020 | 43.6% | 36.9% | 34.9% | 29.9% |
| October 2022 | 41.6% | 35.7% | 33.5% | 28.7% |
| October 2023 | 40.8% | 37.3% | 30.8% | 28.7% |
| October 2024 | 39.6% | 35.8% | 29.8% | 27.2% |
| February 2025 | 37.6% | 35.1% | 27.5% | 25.9% |
| October 2025 | 39.4% | 37.1% | 30.8% | 29.4% |
| June 2026 | 45.0% | 41.4% | 35.5% | 33.1% |
| September 2026 | 43.4% | 38.7% | 32.9% | 29.7% |
Margin of error of the corrected yes: about ±2.3 points in each barometer (±3.3 in 2017, with a smaller sample).
Three conclusions:
- The correction always lowers support for independence, by between two and seven points, because the CEO sample contains more pro-independence voters than turned up at the polls. Correcting only the split between parties has a somewhat smaller effect, in the same direction.
- The fall since 2017 is real, but smaller: just over four points in the corrected yes (from 41% to 37% in October 2025) and about five in the independent-state option (from 34% to 29%).
- The low point is not now. The corrected yes bottomed out between 2022 and February 2025, at around 35%. It rebounded in June 2026 (41%) and in September stood at 39%, almost two points above October 2025, a difference within the margin of error.
Who wants independence
In the September 2026 barometer, corrected, the yes to an independent state is 47% among people born in Catalonia, 18% among those born in other regions of Spain and 25% among those born abroad. By age, support is lowest among 25-34-year-olds (31%) and highest from age 50 (42%); among 18-24-year-olds it is 39%, though with a small sample (187 interviews). By recalled vote, 89% of CUP voters say yes, as do 87% of Junts voters, 80% of ERC voters, 24% of Comuns voters, 21% of PSC voters, 8% of PP voters and 30% of those who did not vote.
How we calculated it
- Data. In-person Baròmetre d’Opinió Política of the CEO, cumulative microdata file (61,838 interviews since 2014). Barometers from October 2015, each with between 1,300 and 2,000 interviews.
- Weights. “CEO”: the CEO’s own weighting coefficient. “Corrected”: additionally, raking so that recalled vote in the latest regional election (a question asked of the electoral roll, including abstention) reproduces the official result (2015, 2017, 2021 or 2024, depending on the barometer’s date) and the approximate turnout of residents on election night (77%, 82%, 53% and 58%), keeping the sex, age and province margins of the CEO-weighted sample. Those who do not know, do not answer or could not vote keep their weight. Margin of error: 95% interval using the Kish effective sample size of each barometer.
- Limitations. Recalled vote also has errors: some people do not remember or do not say how they voted, and some side with the winner. Correcting for recall is standard practice in Spain, but it is not the only option; that is why we give both series. The 2015-2024 votes are the official ones as collected on Wikipedia, and resident turnout is the election-night figure, rounded.
- Code.
articulo_ceo_independencia.py, in the altnalysis repository. Percentages include those who do not know or do not answer; without them, the corrected yes for September 2026 is 41%.