4.4 Million Viewers, How Many Votes? What We Know About Politicians on Entertainment Shows

Pedro Sánchez's interview on La Revuelta, a late-night show on Spanish public TV, was the programme's most-watched episode since its launch, with almost three times its usual audience. Twenty years of research and eleven Spanish precedents since 2014 point the same way: these appearances reach people who hardly follow politics and improve the guest's image, but their effect on the vote is small, short-lived and hard to separate from the rest of the campaign. Spanish polls show no pattern: on average, the guest's party moved three tenths of a point in the two weeks that followed.

23 October 2026 · Miguel · Leer en español

La Revuelta’s most-watched night

On Thursday 22 October, an average of 4,387,000 people watched David Broncano’s interview with Pedro Sánchez on La 1 (Spain’s public TV channel), a 33.9% audience share. It is the show’s record: in its two years on air its best night had been the visit of Rosalía (2,706,000 viewers and 20.4%). The peak minute, at 22:59, reached 4,761,000 viewers and a 40.4% share. At the same time, El Hormiguero, the prime-time talk show on Antena 3, stood at 11.4% (Vertele, using provisional Kantar data).

According to Barlovento, 7,129,000 people watched at least one minute of the interview (Lanza Digital). It is the most-watched interview with a Spanish prime minister in more than fifteen years: it beats the 4,334,000 viewers of Mariano Rajoy at Bertín Osborne’s home in 2015, and was surpassed only by the two editions of «Tengo una pregunta para usted» with José Luis Rodríguez Zapatero, in 2007 and 2009 (FormulaTV).

To measure the jump, it has to be compared with what is normal. Since its launch, La Revuelta has averaged 1,511,000 viewers and a 12.4% share, and the previous Monday, 19 October, it had 1,284,000 and 10.5% (FormulaTV). The interview multiplied the average audience by almost three: about three million people who do not usually watch the show watched it that night.

The question that matters for an election forecast is a different one: how many votes does a night like this move? To answer it we reviewed what the research says, the precedents since 1968 and, using our own poll series, what happened after each major appearance of a Spanish leader on an entertainment show since 2015.

A tradition that began with five seconds

On 16 September 1968, Richard Nixon appeared for five seconds on Rowan & Martin’s Laugh-In to say the show’s catchphrase, “Sock it to me?”. His advisers had told him not to; he won the election seven weeks later (History.com; Nixon Foundation).

On 3 June 1992, a day after securing the Democratic nomination, Bill Clinton played “Heartbreak Hotel” on the saxophone, wearing sunglasses, on The Arsenio Hall Show. He was third in the polls, behind George Bush and Ross Perot, and was looking for young, urban voters who did not watch the news (History.com). Critics called it unbecoming of a president; Clinton won in November.

Since then, a visit to an entertainment studio has been part of the calendar of any US campaign. In 2000, George W. Bush and Al Gore went on the shows of Oprah Winfrey and David Letterman. In 2014, Barack Obama sat “between two ferns” with the comedian Zach Galifianakis to promote enrolment on healthcare.gov: the video became the site’s top source of visits and traffic rose almost 40% overnight (CBS News). But only about 19,000 of the roughly three million people who had watched it clicked the link, 0.6% (Mediaite). And in October 2024, Donald Trump’s almost three hours on Joe Rogan’s podcast passed ten million YouTube views in a day and fifty million weeks later (Inside Radio).

The Obama case sums up the problem well: a huge audience does not automatically turn into behaviour. That millions of people watch something does not say how many change their minds.

What the research says

Serious research on the subject began with the 2000 US election and reaches four fairly stable conclusions.

1. They reach people who do not follow politics. Matthew Baum, of Harvard, studied the 2000 campaign with the American National Election Studies (ANES) survey. Voters who got most of their information from entertainment programmes were more likely to find the other party’s candidate likeable and to vote for him than better-informed voters (Baum, 2005, American Journal of Political Science). That is why candidates go: it is the only way to talk to people who watch neither the news nor the debates.

2. They improve the image, not the platform. Patricia Moy, Michael Xenos and Verena Hess, using more than 11,000 interviews from the 2000 Annenberg survey, found that people who watched Bush on Letterman came to judge him more by his personal traits. In the week after the interview, his rating rose from 56.2 to 59.0 points (out of 100) among viewers of the show and fell from 58.5 to 55.5 among those who did not watch; the difference disappeared the following week (Moy, Xenos and Hess, 2006, International Journal of Public Opinion Research; University of Washington summary).

3. The effect is brief and depends on who is watching. Baum and Angela Jamison showed that interviews on daytime shows such as Oprah’s helped poorly informed voters vote more consistently with their own opinions, not change them (Baum and Jamison, 2006, Journal of Politics). In an experiment on John Kerry’s 2004 interview on Letterman, Michael Parkin found that the interview engaged viewers without leaving them worse informed or focused only on image, and that it barely changed the criteria they used to judge the candidate (Parkin, 2010, Political Research Quarterly).

4. What moves votes is repeated exposure, not a single night. The most recent work with a causal design, not yet peer-reviewed, uses the holding of mixed martial arts (UFC) events as a natural experiment. Where events took place, young men listened more to Joe Rogan’s podcast and the Republican vote rose in the primaries in which Trump ran, with the effect concentrated in that group (Ash, Galletta, Masera and Zhang, working paper). It is an effect of media diet, not of a single interview.

In short: an appearance like this reaches voters who would otherwise learn nothing, improves for a few days the goodwill towards the guest among those voters and can help them vote what they already thought. There is no evidence that it persuades voters of other parties in numbers large enough to move an election.

The Spanish precedents: from Bertín to Broncano

In Spain, the habit took hold in the 2015 campaign, the first with four major parties. Soraya Sáenz de Santamaría danced on El Hormiguero on 6 October; Pablo Iglesias (Podemos, on the left) sang and played the guitar on the same show on 3 November, with the best share in its history until then (21.7%); Albert Rivera (Ciudadanos, Cs, centre-right liberals) and Pedro Sánchez (the Socialists, PSOE) went on Pablo Motos’s set, and Sánchez and Mariano Rajoy (conservative People’s Party, PP) visited Bertín Osborne’s home on La 1, in his celebrity home-interview show “En la tuya o en la mía”. Rajoy drew 4,334,000 viewers and a 23% share.

Then came Santiago Abascal’s (Vox, far right) interview on El Hormiguero on 10 October 2019, with 4,049,000 viewers and a 23.5% share (the third most watched in the show’s history); those of Rivera (2,458,000) and Pablo Casado (PP, 2,049,000) in the same campaign (The Objective); and the remote head-to-head of June 2023: Sánchez on 27 June (2,922,000 and 22.8%) and Alberto Núñez Feijóo (PP) on the 28th (3,079,000 and 25.9%, the show’s record share). That July, Sánchez also appeared on the podcast La pija y la quinqui, hosted by two 23-year-olds and with a very young audience.

This week’s La Revuelta has the highest share of the whole series. The figures need one caution: linear TV audiences have fallen since 2015, so viewer numbers are not comparable across years; the share, which measures what part of the TV sets switched on was tuned to the show, is a better comparison.

Do the polls move afterwards?

To find out, we did the following with all polls published since 2014 (excluding the CIS, Spain’s state polling institute). For each appearance, we compare the average of the polls for the guest’s party with fieldwork ending in the previous two weeks against that of the polls conducted entirely in the following two weeks. Beforehand, we correct each poll for its pollster’s usual lean in that election cycle, so that a party does not seem to rise only because in that fortnight the pollsters most favourable to it published more.

The result on its own says nothing unless one knows how much polls move in an ordinary fortnight. So we repeated the calculation on all the other days of the same cycle, away from any appearance (what is called a placebo), and compared each change with the range that contains 90% of those normal changes.

Ten appearances have enough polls before and after. On average, the guest’s party rose 0.3 points: seven times it rose and three times it fell. Only one change falls outside the normal range: that of the PSOE after Sánchez’s interview on El Hormiguero in June 2023 (+1.7 points). And that case has a catch.

The PSOE was already rising before the interview: from 25.0% in early June to 26.6% on 25 June, after Sánchez called the general election early following the municipal elections of 28 May. In the two weeks before the show it had gained 1.1 points; in the two after, it gained 1.7. It is hard to attribute that difference to one interview rather than to the trend, to the formation of Sumar (a left alliance) or to the PP’s regional pacts with Vox, which were sealed in those same weeks. Feijóo had a larger audience than Sánchez and the PP moved little (+0.6, within the normal range). At the ballot box, the PSOE got three points more than the last polls gave it and the PP almost one less.

The second-largest rise is Vox’s after Abascal’s visit in October 2019 (+1.6 points, within the normal range for a campaign in which Vox kept growing).

Vox stood at 9.6% on the day of the interview and at 10.3% six days later. The big rise came afterwards, with the “procés” sentence of 14 October (the Supreme Court ruling on the Catalan independence leaders), the riots in Catalonia in the following days and the debate on 4 November: 11.4% on 25 October, 13.4% on 6 November and 15.1% at the ballot box. With these data, the night of 10 October cannot be separated from everything that came after.

The 2015 cases point in the same direction. After going to Bertín Osborne’s home on 25 November, the PSOE fell 1.3 points in the following two weeks, having risen 1.2 in the previous two. Rajoy’s PP, with the largest audience of that campaign, moved 0.4 points. Sáenz de Santamaría’s dance was followed by a drop of seven tenths of a point for the PP.

Audience is not support

That 4.4 million people watch an interview does not mean that 4.4 million people agree with the interviewee. There are at least three kinds of viewer on a night like this:

• The show’s regular audience: between 1.3 and 1.5 million people on any night, who already know and choose the show. For them, the interview is part of their usual appointment.
• The event audience: the extra three million. They watch the interview because it is news, and among them are supporters, the curious and critics. Feijóo had a larger audience than Sánchez on El Hormiguero in 2023 and then got less than the polls gave him: audience measures interest, not preference.
• Those who do not watch it but see the clips, in the news and on social media, over the following days. This second wave reaches more people than the broadcast and the research on it is still scarce.

Data by age and by region, published the next day, fit that reading. By age, the audience share was 51.1% among 25-to-44-year-olds and 48.7% among 13-to-24-year-olds, far above the average; the average viewer was 52 years old (Barlovento, according to Lanza Digital and The Objective). By region, the interview was watched most in Madrid (43.2%), the Basque Country (41.7%) and the Valencia region (40.0%), and least in Andalusia (27.5%) and the Canary Islands (21.1%) (Infobae).

If audience reflected support, the interview would have been watched most where the PSOE is strongest. It was rather the opposite: across the 14 regions with data, the correlation between audience share and the vote our forecast gives the PSOE is negative (−0.43), though weak and based on few cases. Each region’s schedule also matters: in the Canary Islands, for example, the broadcast started at 21:03 local time. And without the show’s usual share in each region we cannot refine this any further.

If the interview has any electoral effect, the research suggests where to look: in the turnout of voters with little interest in politics, rather than in vote switching. In our model, people who did not vote in 2023 are the group that most overstates its probability of voting (it is the strongest correction in our calibration with the CIS), followed by first-time voters: they are the hardest to mobilise. If Sánchez gets some of the young left-wing voters who planned to stay at home to go and vote, it will show up more in turnout than in declared voting intention.

What to expect for 29 November

• In the polls. None of those published so far did its fieldwork after 22 October. The latest from IMOP for El Confidencial, published today, was conducted between 13 and 16 October. The first with later fieldwork will arrive next week and will enter the forecast like any other.
• In the model. We do not add an effect of our own for the interview: there is no empirical basis to quantify it, and our analysis finds no pattern. If anything moves, the polls will pick it up.
• In the press. We have added the interview as a milestone in our momentum index, which measures the tone of coverage of each party. There we will see whether the conversation of the coming days favours or harms the PSOE.

Eleven Spanish precedents and twenty years of research do not allow us to rule out an effect, but they do bound it: if it exists, it is small, brief and concentrated among people who hardly follow politics. That an interview breaks audience records is a story about television; whether it is also an electoral story will be told by the polls of the next two weeks.

How we calculated it

• Audiences. Average audience and audience share from Kantar according to the linked sources: FormulaTV (El Hormiguero, 2014–2015), Wikipedia (En la tuya o en la mía, 2015), El Español (Abascal, 2019), El Independiente and The Objective (2023) and Vertele (2026, provisional data). Audience by region: Dos30 with Fifty5Blue data (Infobae); age profile and unique viewers: Barlovento with Fifty5Blue data.
• Polls. Those compiled on Wikipedia for the 2015 to 2026 general elections, excluding the CIS. Before: polls with fieldwork ending in the 14 days before the broadcast. After: polls that started fieldwork after the broadcast and ended it within the following 14 days. At least three are needed in each window; Sánchez’s 2014 El Hormiguero interview does not have them and La Revuelta’s does not yet.
• House correction. In each cycle and party, each pollster’s mean deviation from the 15-day rolling median of all polls, shrunk with four polls of pseudo-weight and centred on the industry mean.
• Placebo. The same calculation on each day of the cycle within one year of the event, excluding the 14 days around any appearance and the election day onwards. The normal range is the interval between the 5th and 95th percentiles of those changes.
• Limitations. Two-week windows capture everything that happens in them, not just the interview, and some 2015 appearances are less than two weeks apart. With ten cases, the analysis can detect large effects, not effects of a few tenths.
• Code. articulo_tele.py, in the altnalysis repository.