How the scores are calculated

Short version: for every vote where the big parties disagreed, we check whether a place sided with one camp more than Switzerland did. We average these gaps over a period, compare periods, and test the yearly numbers for a clear trend.

1. The data

• Vote results: the yes and no votes of every municipality for every federal vote since 14 June 1981. Source: Federal Statistical Office (BFS), on opendata.swiss.
• Party advice: the official yes or no advice of the SP, the FDP and the SVP for each vote. Source: Swissvotes (Année Politique Suisse, University of Bern).
• Mergers: the BFS list of all municipality changes since 1980.

2. Two scales: left–right and liberal–conservative

The page uses the same two scales as the Swiss "political map" of Michael Hermann and Heiri Leuthold (Sotomo):

• Left ↔ Right is about money and the state. How much should the state share out, through pensions, health care and public services? Who pays, through taxes? Left = more state and more sharing. Right = lower taxes, less state and more private solutions.
• Liberal ↔ Conservative is about openness and society. How open should Switzerland be to Europe and to migrants? Should society change or keep its traditions? Liberal = more open and more change. Conservative = more tradition and more national control.

The two scales are independent. A place can be right and liberal (low taxes and pro-Europe, as in many rich suburbs). A place can also be left and conservative (for more welfare, but sceptical of opening up).

A word on "liberal". In Swiss politics, "liberal" often means the FDP, which is on the right on money questions. On this page, "liberal" only means open on society questions, as in Sotomo's map.

How the votes are sorted

Sotomo finds the two scales with statistics run on all votes. This page uses a simpler rule that anyone can check: who stood against whom, based on the official advice of the three biggest parties.

Why not three separate scores for "left", "liberal" and "conservative"? A vote result is one number, the yes share, so each vote can only push a place in one direction. If every vote counted for all three labels, the scores would just mirror each other. Sorting the votes onto two scales avoids this.

3. The score for one vote

For one vote, we compare the place with Switzerland as a whole:

If the plus side advised YES: score = local yes % − Swiss yes %
If the plus side advised NO: score = Swiss yes % − local yes %

So a plus score always means "sided more with the plus side than Switzerland did". We compare with Switzerland so that easy and hard votes do not distort the picture. An initiative that everybody rejects still shows who rejected it less.

Example, Bern, 3 March 2024:

• 13th AHV pension (SP: yes; FDP and SVP: no). Bern 64.1% yes, Switzerland 58.2% yes. Score = 64.1 − 58.2 = +5.9 left.
• Higher retirement age (SP: no; FDP and SVP: yes). Bern 23.2% yes, Switzerland 25.2% yes. Score = 25.2 − 23.2 = +2.0 left.

4. The score for a period

You pick a period length: 2, 5 or 10 years. The periods count back from the latest year with votes (2026, with votes up to 27 September). For 5 years this gives 2022–2026, 2017–2021, and so on back to 1982–1986. If the oldest period is shorter than the others, it is kept only when it covers at least 60% of the length, and it is marked with a *.

period score = average of the vote scores of all votes in that period on that scale

"Initiatives only" uses popular initiatives. "All votes" adds referendums and counter-proposals. Tie-break questions are never used.

5. The change

change = period score − period score of the period before

Example, Bern, initiatives, left–right scale: 2017–2021 = +19.1 (11 initiatives). 2022–2026 = +21.0 (9 initiatives). Change = +1.9 toward left.

Because each score is measured against Switzerland, a change means the place moved compared with the rest of the country. If all of Switzerland moves left together, nobody "changes" on this map. To see how the whole country moved, use the Switzerland level (section 6).

6. Levels: municipalities, cantons, language regions, Switzerland

• Cantons and language regions: we add up the yes and no votes of all their municipalities (as they are today), then use the same formula as for a municipality. Big towns count more because they have more voters. The language region of each municipality is the BFS classification (German, French, Italian or Romansh). The Swiss abroad are not in any canton or region total.
• Switzerland: compared with itself, Switzerland would always be 0. So for this level we show the share of Swiss voters on the plus side instead. For example, "43.3% left" means that in 2022–2026, on average 43.3% of voters sided with the SP in left–right initiatives. The official Swiss result is used, including the Swiss abroad.

Note: at the Switzerland level the numbers depend on which topics came up for a vote. A period with many hard left proposals will show a lower left share, even if people did not change their views.

7. Popular votes: the "Trend since 1981" view

The Popular votes tab has two views. Compare periods uses the period length and the slider (sections 4 and 5). Trend since 1981 ignores periods and looks at every year at once.

In the trend view, the chart shows one point per year, from 1981 to 2026, for the place you picked and the scale you picked. Bigger dots mean more votes that year. Years with no vote on the scale are skipped. Click a dot to list the votes of that year; click a title to open that vote on the map.

We then fit the best straight line through the points. Years with more votes count more (weighted least squares). We only draw this trend line when it is clear, which means a standard t-test on the slope gives p < 0.05. In plain words: if the place had not really moved, a slope this steep would appear by chance less than 1 time in 20. If the trend is not clear, the page says so and draws no line.

Examples (all votes, liberal–conservative): Bern moves 3.3 points per 10 years toward liberal, a clear trend. Schwyz moves 0.8 points per 10 years, which is not clear, so no line is drawn. The whole of Switzerland moves 4.2 points per 10 years toward liberal, a clear trend.

The map in the trend tab colours every place by its trend, in points per 10 years. Only clear trends get a colour. Places with no clear trend are pale. Places with fewer than 5 years of votes are grey.

The test assumes that the yearly ups and downs are independent of each other. Changing topics from year to year break this a little, so treat a clear trend as a strong hint, not as proof.

7b. Looking at a single vote

Under the map, The votes behind lists every vote that went into the chosen period on the chosen scale: date, title, topic, the Swiss yes share, the yes share of your place, and your place's score for that vote. Open a vote to see:

• Why it counts for the scale: the advice of the SP, the FDP and the SVP, and which answer counts toward which side. The advice of the Mitte (CVP before 2021), the Greens and the GLP is shown for context; it does not change the scale.
• How people voted: yes and no shares for your place, its canton, its language region and Switzerland.
• Its own map: every place compared with Switzerland on that one vote. This is exactly the vote score from section 3; the period score is the average of these maps.

Titles and topics come from Swissvotes. The topic is the first-level policy area that Swissvotes assigns to each vote (for example "Social policy" or "Foreign policy").

8. Merged municipalities

Results are always shown for today's municipalities (state 1 January 2026). For years before a merger, we add up the yes and no votes of all the former municipalities that are now part of it. Then we calculate the yes share. So a merged town has one unbroken history.

Four old codes could not be matched because they were split between several municipalities (Schlosswil BE, Caneggio, Casima and Monte TI). Their votes are left out. A few tiny municipalities report together with a neighbour and have no own results.

9. Limits

• Few votes in short periods. With 2-year periods, a period often has only 1 to 3 votes on a scale. The page warns when a period has fewer than 3 votes. Before 1996, almost no initiative split SP and FDP against the SVP, so the liberal–conservative scale needs "All votes" there.
• Small places jump around. In a village with 100 voters, a few people change the result a lot. The movers list hides places under 1,000 voters by default.
• Parties change too. The SVP of the 1980s was a different party from the SVP today. The FDP and the centre parties also moved over time. Party advice is a practical label, not a perfect one.
• Different topics in each period. One period may have many pension votes, the next many environment votes. A place can look like it moved only because the topics changed.
• Who votes changes. Turnout and the population change over time. The map shows how the voters who voted behaved, not what all residents think.
• The latest year is not complete. 2026 has votes up to 27 September. The last period can change after the next vote day.

10. Votes per period

Number of votes used per 5-year period:

11. Town and country

At the Switzerland level, a chart shows the gap between town and country. It uses the urban-rural typology of the Federal Statistical Office (BFS, 2020) and adds up all urban and all rural municipalities, like a canton. Gap = urban score − rural score, per 5-year period; above the line, cities are more left (or more liberal) than the countryside. The trend test is the same as in section 7, run on the yearly gaps. Result in October 2026: cities are about 12 points more liberal and about 6 points more left than the countryside, the same as in the 1980s; no clear change.

12. Comparing two places

In the trend views, Compare with adds a second place as a dashed line. The list follows the level: municipalities, cantons or language regions. At the Switzerland level there is nothing to compare.

13. Download

The footer of the map page offers the computed numbers as CSV files: vote scores per 5-year period, vote trends and party shares. They use the same formulas as this page. Licence CC BY 4.0; name Polatlas and the sources (BFS, Swissvotes). The BFS vote results may not be used commercially without the BFS’s permission.

1. What the Elections tab shows

The Elections tab shows National Council elections from 1971 to 2023 (every 4 years). It does not mix with the vote scores of the Popular votes tab. It shows plain party shares: which share of the party votes each party got.

Compare elections has four map views:

• Share of votes: the share of the party or bloc you picked.
• Change vs election before: its gain or loss in percentage points.
• Strongest party: the biggest of the six main parties (SP, Greens, GLP, Mitte, FDP, SVP) in each place.
• Biggest winner: the main party with the largest gain since the election before. If all parties lost, it is the one that lost least.

Compare elections shows the chosen election next to the one before: one bar per party for the selected place. Everything over time (the party mix chart, the line for one party and its trend line) is in the Trend since 1971 view.

The party mix chart stacks all six parties plus "other" to 100% for every election. It shows how a place moved from one party to another. Trend since 1971 colours the map by the trend of the chosen party, with the same test as in section 7 of the Popular votes tab (each election counts the same).

What this cannot show: which voters switched from which party to which. Results by place only show the net change. Real voter flows need surveys of voters, such as the Selects election study.

2. Data and method

• Source: BFS, "Nationalratswahlen: Parteistimmen und Parteistärke seit 1971" for every municipality, and the BFS table for cantons and Switzerland.
• Parties: SP, Greens, GLP, Mitte (with CVP and BDP before their 2021 merger), FDP (with the Liberal Party before 2009) and SVP.
• Blocs: Left = SP, Greens and small left parties (PdA, Solidarités, POCH, PSA, FGA). Centre = Mitte, CVP, BDP, GLP, EVP, CSP, LdU. Right = FDP, LPS, SVP and small right parties (SD, EDU, Lega, MCR, Republicans, Freedom Party). Other lists are counted in the total but in no bloc.
• Adding places up: each canton elects a different number of seats, so each voter casts a different number of party votes. To add municipalities across cantons, we first turn party votes into "fictitious voters" (party votes ÷ the canton's seats, from the official BFS figures). Cantons, language regions and Switzerland are then added up from these. Check: our total for Switzerland in 2023 is SVP 28.0% and SP 18.3%; the official result is 27.9% and 18.3%.
• Mergers: party votes of former municipalities are added to today's municipality, as for the votes.
• Trend line: the same straight-line test as in section 7 of the Popular votes tab, with each election counting the same. A line is drawn only when p < 0.05.

3. Limits

Small cantons with only one seat (for example Uri, Obwalden, Nidwalden, Glarus and the two Appenzells) often have a contest between people more than between parties. Their party shares can jump a lot from one election to the next. Moutier moved from Bern to Jura in 2026; its older results are counted with Jura's seat factor.

1. The data

The Polls tab shows national opinion polls over time. All figures come from the study series «Sicherheit» of ETH Zürich (Military Academy and Center for Security Studies). Each year, a survey of Swiss voters asks the same questions, mostly in January. The figures on this site are copied from the published report (edition 2025); we do not have the raw survey data.

• EU: two statements. “More economic cooperation with the EU, without joining” and “Join the EU without reservations”.
• Neutrality: “Switzerland should keep its neutrality”, and the “differential” view: take a side in political conflicts abroad, stay neutral in military ones.
• Army: the share who find the army “absolutely necessary” or “rather necessary”.
• Trust in institutions: five of the institutions in the study: Federal Council, Federal Parliament, courts, army and political parties.

2. How the numbers are shown

• Shares in %: the share of people who “agree” or “somewhat agree”, as published and rounded to whole percent.
• Trust: the average on a scale from 1 (no trust) to 10 (full trust). The chart only shows 4 to 8, so the lines can be told apart.
• Two surveys in one year: in some years a second survey was made, for example in 2022 after the war in Ukraine began. It is drawn as a second point in that year.
• Hover or tap a year to see the exact values of all lines.
• Event marks on the EU chart show votes and events that may help to read the line. They are context, not proof that the event changed opinions.

3. Limits

• National only. A survey asks about 1,000 to 2,000 people. That is enough for Switzerland, not for a canton or a municipality. So the Polls tab has no map.
• Error margin. A single year can be off by about 2 to 3 points. Look at the direction over several years, not at one step.
• Wording and method changed. Some questions were worded differently in earlier years (the notes under each chart say how), and the survey method changed over time.
• A poll is not a vote. People answer a question in a survey; in a vote they decide on a concrete proposal. The two can differ.
• Not an official dataset. The authors of the study are not involved in this site. If they ask, the figures will be removed.