Why We Vote and When We Decide: Decisions Are Coming Later and Later

In 1996, 87% of those who voted in general elections had made up their minds before the campaign. In 2023, 61%. Among people under 35, only 43%. Thirty years of post-election surveys by the CIS (Spain's state polling institute) also show why people vote: two out of three Spaniards see it as a right more than a duty, and the most-cited reason for choosing a party is ideological closeness, followed in 2023 by voting against another party.

30 October 2026 · Miguel · Leer en español

The data

After every general election, the CIS asks several thousand people whether they voted, for whom and when they decided. Sometimes it also asks why. We have unified the ten post-election surveys from 1996 to 2023, with 61,428 interviews, to follow those answers over time. The questions are not always worded the same way, so some series are more comparable than others; we flag this in each case.

Decisions are coming later and later

When voters decided how to vote, 1996–2023Percentage of voters who say when they decided. CIS post-election surveys, weighted. In 2004 the start of the campaign and the day of the vote were not distinguished.

The series shows a clear trend:

In 1996, 87% of voters had decided before the campaign.
In 2008, 80%.
In 2023, 61%. One in six (16%) decided in the last week, and one in ten the day before the vote or on election day itself.

There are exceptions that prove the rule. In 2015, with the arrival of Podemos and Ciudadanos, only 64% had decided in advance. In 2016, a rerun of that election with the same candidates, it rose again to 77%. When the menu of options changes, decisions are delayed. In 2026 there are new or reshuffled parties (Frente Amplio, Se Acabó La Fiesta, Podemos running separately), so late decisions are to be expected.

Young people decide the latest

Voters who had decided before the campaign, by agePercentage of voters in each age group. CIS post-election surveys of 1996, 2008 and 2023.

18–34

199680.7%

200873.0%

202343.3%

35–54

199688.3%

200879.5%

202356.8%

55 and over

199692.4%

200886.9%

202373.5%

Age makes the biggest difference. In 2023, three out of four people over 55 (74%) had decided before the campaign, compared with fewer than half of those under 35 (43%). In 1996, even young people decided early (81%). The drop has been across the board, but much sharper among the young. They are the group with the least party loyalty and the most alternatives to hesitate between.

2023 voters who had decided before the campaign, by partyPercentage of those who recall voting for each party. CIS post-election survey of 2023.

2023 voters who had decided before the campaign

Vox72.7%

PP70.1%

PSOE60.5%

Nationalist parties59.7%

Sumar48.7%

In 2023, Vox (73%) and PP (70%) voters were the ones who had decided earliest. PSOE voters were at 61%, and Sumar voters below half (49%). The campaign matters more for parties whose voters decide late.

Voting: duty or right?

“Voting is above all a duty”Percentage of respondents who choose “duty” over “right.” CIS post-election surveys, 2008–2019 (the question was not asked in 1996–2004 or in November 2019 and 2023).

When the CIS asks whether voting is above all a duty or a right, around two out of three answer “right.” The proportion has barely changed: civic duty hovered between 34% and 39% from 2008 to April 2019. This view of voting as a right helps explain why turnout rises and falls with what is at stake, more than with a sense of obligation.

Why people choose a party

The CIS has asked in different ways why people vote for a party. That is why the figures for each year are not strictly comparable, but some patterns recur:

Ideological affinity, or “being a supporter of that party,” is always the most-cited reason. It ranges from 25% to 45% depending on the year and the wording.
The tactical vote, or voting against another party, is the second major motive in the most polarized elections: 30% in 1996 and 29% in 2023. In the others, between 5% and 16%.
The government’s record weighs more when there is a government to judge: 39% in 2000, after Aznar’s first term, and around 25% in 2016 and April 2019.
The leader is the main reason for between 6% and 17% of voters, less than is usually assumed.
Habit (“I always vote for them”) is chosen by around one in five when it is offered as an answer.

And those who do not vote

Among those who did not vote, the decision to abstain is also made later and later. In 2023, 52% had already decided before the campaign and 14% said they never vote. One in six (16%) decided not to vote the day before or on election day. In earlier elections, “I never vote” ranged from 25% to 52% of abstainers. That it fell in 2023 fits with what we reported in the article on systemic abstainers: there is more and more intermittent abstention.

What it means for Nov. 29

With four in ten voters deciding during the campaign, today’s polls do not fix the result. That is why our forecast leaves a wide margin and gives more weight to the most recent polls as the election approaches.
Young people and voters of the smaller parties are the ones who decide most during the campaign. That is where more votes are in play.
PP and Vox voters arrive at the campaign more decided than those on the left. For the PSOE and Frente Amplio, the campaign has more room for gains and also for losses.

How we calculated it

Surveys. CIS post-election surveys of the general elections of 1996, 2000, 2004, 2008, 2011, 2015, 2016, April and November 2019, and 2023 (studies 2210, 2384, 2559, 2757, 2920, 3126, 3145, 3248, 3269 and 3420), with their weighting (cis_poselectorales.py).
Timing of the decision. Answers from those who voted to “When did you decide to vote for that party?”, grouped into before the campaign, at the start of or during the campaign, in the last week, and the day before or the day of the vote. In 2019 and 2023, people who do not recall a party (blank, void, don’t know) are not asked.
Duty or right, and motives. The wording changes between studies. In 2023 the motives are multiple-response and the first is taken in a fixed order, so that figure is approximate. The mapping of each question is in docs/cis_poselectorales.md.
Limits. These are self-reported memories collected after the election. Some people credit themselves with a firmer decision than they actually made. The surveys also capture less abstention than the real figure.
Code. cis_poselectorales.py and articulo_decision.py, in the altnalysis repository.