We looked at the Novus Voter Barometer for July to see how voter preferences are distributed based on religious affiliation.

As part of the Novus Voter Barometer, eligible voters in the parliamentary election are asked the following question: “Regardless of how religious you are, Do you belong to any religion?”
Novus surveys a random sample of the Swedish population via Kivra and text message; in July, 5,726 interviews were conducted, making it—with a good margin of error—Sweden’s largest voter barometer, conducted monthly based on a random sample. Because of this method, we can discuss margins of error while also focusing on smaller groups. Generally speaking, the minimum number of interviews required to draw meaningful conclusions is about 100; below that, the margin of error becomes extremely large. And, of course, this assumes that you actually have a random sample of the Swedish population. Otherwise, it’s impossible to make any statements at all about the population as a whole. Self-recruited panels do not, even when considered as a whole, provide a basis for making statements about the entire Swedish population. There is a lack of scientific support for that—both in terms of the overall sample and, of course, in terms of specific subgroups like these. However, Novus’s surveys meet the standards that allow us to analyze the data. However, one should keep in mind that when the number of interviews in a group drops to just over 100, the margins of error become larger than if there were 1,000 or 5,000 interviews. Nevertheless, it provides a clear indication of the relative sizes of the parties even within the smallest groups.

Among those initially eligible to vote in the parliamentary election, religious affiliation breaks down as follows:

The dominant group is Christians, at nearly 64%; next are those who do not consider themselves to belong to any religion at all, at about 30%. Muslims account for just under 3% and Jews 0.2%; all these figures have margins of error—for 0.2%, the margin of error is about 0.2. For 9%, the margin of error is about 0.8; for 2.6%, it is about 0.4. It’s also important to remember that these figures reflect self-identification; according to the Church of Sweden’s own statistics, 4.9 million people aged 16 and older are members of the Church of Sweden. This is consistent with the proportion of respondents who identify as Christian in this survey. The proportion of Jews, according to official statistics, is a bit more difficult to pinpoint, but statistics suggest approximately 0.2%, which is roughly within the margin of error. The proportion of Muslims, according to the SOM Institute, is in line with the Novus survey.

Read more about language proficiency and religious affiliation in this Novus report, but it doesn’t just look at registered voters, and it’s a few years old—plus, as mentioned, the margin of error has an impact: Demographics, Religion, and Language Proficiency – Novus – Knowledge and Surveys

The survey was conducted in July, with data collected from July 6–9, 2026, from a random sample of the population via Kivra and text message. You can read it here on the Novus website, including my analysis of voter sentiment: L at an all-time low and support for V declining – Novus – Knowledge and Surveys

I have chosen to factor in party affiliation in relation to the size of the various groups so that, when combined, they total the voter support they receive in the voter barometer. For example, for the Social Democrats (S), which stood at 31.8% in July, Christians account for just over 20%; the Social Democrats are the largest party even among this group, just as they are the largest among those who do not belong to any religion, as well as among those who are Muslim. But as you can see in the chart, those who are Muslim represent a relatively small group of the total population, and their impact on the election outcome is very small.

If we look at each group individually, the picture seems more dramatic, but the relevant representation is the graph above

Torbjörn Sjöström
Founder of Globeknot, Chairman of Novus