The fact that the Moderate Party’s campaign leadership wants to call out the polling firms after the election is both remarkable and irresponsible. Instead of trying to understand voters and accept basic principles regarding margins of error and current trends, they choose to blame the map. Here, I explain why historically high voter mobility during the final days of the campaign determined the election outcome, why Swedish opinion polls remain among the most accurate in the world, and why political attempts to put pressure on independent research firms have no place in a democracy.

The Moderates seem to have a bit of a hard time understanding voter trends, margins of error, and perhaps opinion polls in general. Several of us were surprised to see SVT’s report that the Moderates’ campaign manager, Martin Borgs, was going to “call in” the polling firms so that we could explain why we had underestimated the Moderates in the last three elections.

I can only recall one other instance in a democratic society where a political party has publicly raised objections to opinion polls in this way, and that was by Trump in the United States. But I could never have imagined that this would happen in Sweden. For that matter, the United States is classified as a “flawed democracy,” a far cry from Sweden’s strong democracy.

But my thought is that we can view this as an opportunity to explain how polls work.Opinion polls provide a snapshot of the current situation, not a prediction of the future. 

Voter mobility has increased over the past 12 years. In 2022, 19% of voters made up their minds over the weekend; this year, the figure was roughly the same. In 2018, after Jimmie Åkesson said “they aren’t Swedes” during SVT’s final debate, 4% of Swedes changed their minds—about 2% switched to the Moderates (M) and 2% to the Social Democrats (S). Voters mainly went to the two governing parties, at that time to ensure that the Sweden Democrats (SD) would not gain influence. It may have been forgotten, but the parties had both a December agreement in 2014 and a January agreement in 2019 to ensure that the SD did not gain influence. There have not been two blocs since the Alliance, which was also a historical exception. Tidö was formed AFTER the 2022 election; there was no “conservative bloc” prior to the 2022 election. The Liberal Party (L) and the Sweden Democrats (SD) couldn’t even ride on the same bus during that election campaign.
Not only I, but also Sören Holmberg noted after the 2018 election that voter turnout was historically high. It hasn’t decreased since then. 

Why has voter mobility increased? It’s due to an ever-greater focus on government formation and party lineups, while ideological differences have taken a back seat. In this election, political maneuvering has taken up so much space that even the substantive issues have been overshadowed. The question of who will govern with whom has overshadowed everything else.
The Liberals are a clear example of this. When they succeeded in establishing the narrative that the Tidö coalition cannot survive without them, they experienced a historic surge just two weeks before the election. That surge was first detected by Novus and subsequently confirmed by Verian and Ipsos.
Overall, the polling firms have presented a consistent picture. While it is certainly easy to get fixated on isolated percentage point differences, the margin of error always plays a role. Rapid shifts in voter sentiment in the final stages—which, moreover, continue even after the polls have closed—naturally mean that the final results may differ, even though, on the surface, it appears the numbers should be identical. The picture nevertheless remains fundamentally consistent, with entirely natural variations, and one really cannot ask for more from opinion polls.

We also deliberately stop conducting the poll on Thursday evening and present the results on Friday, so as not to create a false impression that the election has already been decided. The times when polls have influenced elections around the world are when it has been incorrectly claimed that the outcome is already a foregone conclusion. In those cases, voter turnout has dropped. The most clearly documented effect is precisely that voters stay home because they believe everything has already been decided and that their vote isn’t needed. That is why we are very careful to emphasize that this is not a forecast, and we conclude our polls a few days before Election Day. That is our way of taking responsibility.

It is therefore particularly troubling when the Moderate Party’s campaign manager claims that we are wrong and that we are doing things wrong. When the party that has led the government openly questions all surveys on SVT, nothing good can come of it. It undermines trust in the knowledge-based society and is simply irresponsible.

It is also noteworthy that the Moderate Party does not understand how voters think, and instead chooses to attack the methods and surveys rather than trying to understand the voters.

The 2018 Election: “They’re Not Swedish”

If we also look at these three elections and the voter polls that were released, let’s start with 2018

During SVT’s final debate in 2018, Jimmie Åkesson said that “they are not Swedes,” which led to both the Moderates and the Social Democrats gaining nearly two percentage points each. Whenever Åkesson makes statements that could be interpreted as xenophobic, it tends to hurt their voter support.

The Sweden Democrats upset the (power) balance between the Moderates and the Social Democrats
The Sweden Democrats had already upset the balance between the two major parties in 2014. At that time, the Social Democrats campaigned on the message that “Sweden is falling apart,” believing that a weakened Moderate Party would automatically benefit them, just as Håkan Juholt’s earlier collapse had previously given Fredrik Reinfeldt a boost. But during the 2014 election campaign, they failed to recognize that the Sweden Democrats were tapping into the lingering anxiety in the wake of the financial crisis.

Although the Sweden Democrats lost some of those protest votes during the final debate in 2018, that election also marked the beginning of the high voter mobility we have seen since then. This has become particularly evident in the discrepancies between the latest opinion polls and the actual election results.

The margin of error is shown in the graph above. All polls were below the Moderate Party’s final election result, but the gap was small when the margin of error is factored in, with the line representing the election result. At the same time, the graph provides a somewhat simplified picture because Demoskop and Ipsos do not include margins of error, as their methods are not based on independent random samples. However, if they had used a random sample, the margins of error would have been precisely at these levels. They are included in the comparison nonetheless, since these were the figures reported in the media. It is possible that they used random samples specifically in this election year, given the methodological changes made over the years, but in 2022 and 2026, they did not use random samples at all.

The 2022 Election, Russia's Invasion, the Energy Crisis, and Emporia

Here, we do not see the same pattern at all—namely, that the polling firms’ pre-election results for the Moderates fell outside the margin of error compared to election day. However, we did see the same trend regarding the “weekend effect” for the Moderates as in the previous election. In this election campaign, the Social Democrats led the government with Magdalena Andersson as prime minister, while Ulf Kristersson was the challenger. During the final weeks, the battle was between the Moderates and the Sweden Democrats over which party would become the second-largest. The wave of violent crime caused great concern, and the shooting at Emporia gave the Sweden Democrats a boost shortly before Election Day. Russia’s full-scale invasion of Ukraine and skyrocketing electricity prices dealt a heavy blow to the Social Democratic government and benefited the Moderates. Right around the turn of the month from August to September, the Moderates regained voter support while the Social Democrats suffered a sharp decline. Although the Social Democrats managed to make up for the loss toward the end, attention increasingly focused on the three major parties. The Moderates once again managed to win over voters during the election weekend, just as they did in the previous election, although the shift this time was not quite as dramatic.

2026 Game Theory Election

If you factor in the margin of error, the deviation from the election results wasn’t particularly dramatic this time either for Novus and Ipsos, even though the other three were significantly lower than Ipsos and Novus. As always, this is a snapshot of the current situation and not a forecast.

As I mentioned earlier, based on Novus’s post-election survey, 17 percent of Moderate Party voters made up their minds during the election weekend itself, which corresponds to 3.4 percentage points. Overall, this is the election in which we have seen the most voter shifts in an extremely short period of time, both before and during the election weekend. Such late shifts naturally affect the polling firms’ average margin of error relative to the final election results. For Novus, that margin of error was just over 1 percentage point in this election—slightly higher mainly due to increased voter volatility—but from an international perspective, this is extremely good.

Here, too, we can largely attribute the Moderates’ rise over the weekend to the Sweden Democrats’ sharp decline. Analysis is still ongoing, but there is strong evidence that the Sweden Democrats were perceived as being far too xenophobic. When Jimmie Åkesson makes statements that can be interpreted as xenophobic or as attacks on the media, the party loses supporters, and those voters shifted to the Moderates over the weekend. The Liberals’ recovery was primarily due to support votes from the Moderates. Those votes had already shifted before the weekend and, as we know, remained that way even on Election Day. The Moderates thus lost voters to the Liberals, but were saved in the final days by the collapse of the Sweden Democrats.

At the same time, the Social Democrats ran a weak campaign, which benefited the Moderates toward the end. The prime ministerial debate appears to have strengthened Ulf Kristersson and the Moderates, while the Social Democrats actually lost some voters over the election weekend, although that loss was nowhere near as large as that of the Sweden Democrats. The Moderates won on the strength of their narrative that the “Tidö parties” formed a united front, while the Social Democrats lost ground due to the Left Party’s unwillingness to compromise.

There is much more to say about voters’ reasoning during this election campaign, and anyone reading this text who appreciates it should thank Martin Borgs for the fact that I’m sitting here writing this analysis today instead of having the day off…

If you really want to understand, it’s not too late to buy the Novus post-election survey

We see no systematic errors in the voter barometers. Swedish opinion polls based on scientific methods are among the most accurate in the world. Polls are becoming increasingly important as voters find it harder to navigate—not only a turbulent political landscape but also social issues in general. Moreover, the Voter Barometer is far from the most interesting survey we conduct. Novus’s commercial surveys guide companies and organizations in their strategic decisions, which in turn strengthens Sweden’s competitiveness and creates jobs. Without the knowledge provided by scientifically grounded surveys, decision-makers are groping in the dark.

The Voter Barometer’s overall figure, in and of itself, says very little beyond providing a snapshot of the parties’ battle for voters’ trust. What’s truly interesting lies in the differences between various groups and the shifts over time. It’s not about pitting different polling firms against each other or compiling a “poll of polls,” but rather about analyzing the enormous amounts of data we actually have at our disposal. Our real challenge is rather that our work is reduced to a single percentage point for the Moderates, which is called into question simply because someone dislikes what it shows.

If I had been the campaign manager, I would have asked questions instead of being critical. Unfortunately, Martin has now, on repeated occasions, chosen to attack those he believes are disrupting his circles, rather than trying to understand the actual situation.

I never thought I’d be called out by a political party in Sweden. Nothing good can come of such behavior. Novus is completely independent of any political party, and our entire mission is based on providing an accurate picture of the current situation.

If you get lost because you can't read the map, blaming the map is hardly going to help you find your way.

The fact that the Moderate Party’s campaign manager also doubts the party’s own ability to win over voters in the final stretch contains an irony that I’m sure I’m not the only one to notice.

As a polling firm, we have made a conscious decision not to publish polls that capture the final debates. Neither the party leader debates on TV4 and SVT nor Saturday night’s debate on TV4 were included in the survey period. Voters entered the election weekend with a clear understanding of the balance of power in public opinion, and armed with that knowledge, they were better equipped to cast their votes in a well-informed manner.

We do not predict the future, and no one should think that is what we do. This is our way of taking responsibility for the role we play in a democratic society. Nor will we allow political parties to try to control us or question our independence. Even suggesting that we should be called in raises the question of what the real purpose is. Are we going to be scolded for not fitting into their internal narrative?

If you really want to understand, it’s not too late to buy the Novus post-election survey

Torbjörn Sjöström

Chair of Novus, Founder of Globeknot

Globeknot isa nonprofit, completely independent think tank whose purpose is to disseminate knowledge to the media, the public, and those in power.It is funded by donations; if you'd like to support us, click here.