Party identity

Most people do not match the party they say they support

Where people land when their answers and their party label disagree, and what that has to do with how easily wording moves them.

Published 23 September 2026 · Figures updated 23 September 2026

At the end of every country test we ask one optional question: which party do you feel closest to? Separately, the test works out which party your answers actually sit nearest to. Of the 2,558 people who answered that question for a party the test scores, 55% landed closest to a different party from the one they named.

Matched a different party55%1,412 of 2,558 people
Better than chance by3.1×chance would be about 14%
Extra framing bias when mismatched+2.3points on a 0 to 100 scale

That sounds like a verdict on voters, and it mostly isn't. Party identity is built from far more than policy: family habit, a leader you trust, a local candidate, a sense of who a party is for, and the plain arithmetic of which party can actually win where you live. The test only measures one of those things, agreement on policy. What the gap shows is how often those other reasons outweigh it.

Measured against chance

The raw match rate is not a fair way to compare countries, because matching is harder when there are more parties to choose from. A test with 5 parties gives a random answer a one in 5 chance of landing on yours. So the chart below shows each country's match rate against the rate you would get by picking a party at random.

Share who matched the party they named, by countryBars show the match rate with a 95% interval. The tick shows what random picking would produce.
Matched own party95% intervalExpected by chance
Show the data
CountryPeopleMatched own partyParties scoredChanceTimes chance
Finland1,10752.0%812.5%4.2×
United States22946.7%520.0%2.3×
Germany12343.9%812.5%3.5×
Australia17641.5%520.0%2.1×
United Kingdom54234.1%714.3%2.4×

Every country clears the chance line comfortably, which is the first thing worth saying: people are not guessing. Finland stands out at 4.2 times chance. Australia is lowest at 2.1 times. Finland is the largest group in our data by some distance, which is why the key result below is also checked without it.

Where people land instead

Splitting by the party people named shows that the gap is not spread evenly. Some groups answer almost exactly like their party. Finns Party identifiers in Finland matched their party 83% of the time. At the other end, only 10% of people who named Centre Party in Finland matched it.

Share who matched the party they named, by partyGroups of at least 50 people. Bars show the match rate with a 95% interval.
Show the data
Party namedCountryPeopleMatched itMost common alternativeShare on alternative
Finns PartyFinland43583%Christian Democrats9%
Reform UKUnited Kingdom5871%Restore Britain12%
RepublicanUnited States6469%Independent13%
DemocratUnited States6367%Green17%
Social DemocratsFinland10957%Green League19%
Green LeagueFinland9342%Social Democrats39%
LaborAustralia6237%Teal Independents32%
ConservativeUnited Kingdom8336%Reform UK35%
Green PartyUnited Kingdom10936%SNP37%
Liberal DemocratsUnited Kingdom7935%Labour20%
National CoalitionFinland22934%Finns Party35%
LabourUnited Kingdom11223%Green Party25%
IndependentUnited States8218%Democrat30%
Left AllianceFinland13316%Green League42%
Restore BritainUnited Kingdom8014%Reform UK73%
Centre PartyFinland7910%Finns Party43%

In 7 of the 16 groups, more people matched a different party than the one they named:

Look at where people go when they leave, and a pattern appears. For each group, the most common alternative is a neighbour: a party on the same side of the spectrum, usually one competing for the same voters. In none of these groups is it a party from the opposite side. People drift to the party next door, which suggests the label they chose is close to right, just not exact.

The most consistent groups in our data are Finns Party, Reform UK and Republican. The least consistent are Centre Party, Restore Britain and Left Alliance. One plausible reading is distance. A party whose platform sits well away from its nearest rival is easy for the test to tell apart, so its supporters' answers rarely land in between. A party that shares much of its platform with a neighbour leaves far less room for error, and small differences in how someone answers a handful of questions can tip them across.

The mismatch goes with being easier to sway

Every CoreOpinion test asks key positions twice, once framed sympathetically and once critically, and measures how far your answer moves between the two. That movement is your framing bias score, on a 0 to 100 scale.

People whose answers matched their party scored 21.9 on average. People who landed on a different party scored 24.3. That gap of 2.3 points has a 95% interval of 1.1 to 3.5, so it is very unlikely to be a fluke of the sample.

Because Finland makes up so much of the data, we checked the same comparison without it. The gap does not shrink, it grows: 22.0 against 25.1, a difference of 3.1 points (95% interval 1.4 to 4.8, 1,451 people).

Average framing bias, matched against mismatchedBy country. Higher means answers moved more when the wording changed.
Matched own partyMatched a different party
Finland
United States
Germany
Australia
United Kingdom
Show the data
CountryMatched own partyMatched a different partyDifference
Finland21.922.9+1.0
United States20.725.1+4.4
Germany27.226.8-0.4
Australia21.324.8+3.5
United Kingdom20.623.5+2.9

The gap shows up in 4 of the 5 countries large enough to report, but not in Germany, where the two groups score about the same. We would rather show that than smooth it over: country samples here are small, and a pattern that holds in most places but not all is a pattern to keep testing, not a law.

We cannot say which way the relationship runs. It could be that people who are more responsive to wording are also less settled in their views, so their answers wander further from the label they chose. It could equally be that people who hold a party label loosely answer each question more on its own terms, and so respond more to how it is phrased. What the data does say is that the two travel together.

What this does and does not tell you

It does not mean people are voting wrongly. Choosing a party because of its leader, its record in government, or its chance of beating a party you dislike more are all perfectly reasonable, and none of them show up in a policy questionnaire. It also does not measure how people vote: the question asks which party someone feels closest to, not how they voted.

What it does suggest is that if you care about policy most, your party label is worth checking now and then. More than half of the people in this data would find the closest match on the issues is a party they did not name.

Want to see where you land? Take the test or pick your country from the menu above.

About these figures