How this works
Vancouver Vote Compass compares your answers to what each candidate has said in public. When a candidate hasn’t answered a question directly, we use an AI estimate based on what they have said, and we label it clearly.
Where positions come from
Every position comes from a public source: a candidate’s platform or statements, their party’s platform, council votes, interviews, questionnaire answers or news coverage. Every position links to its source. If we couldn’t find a source for something, we left it out.
Some sources count for more than others. A candidate’s direct answer to our questionnaire comes first, then their campaign material, then interviews, public records and news coverage. If several sources say the same thing, we show the strongest one, and every position tells you what kind of source it came from.
Sometimes a candidate answers a question directly. Other times we match something closely related they’ve said to the question. Those are labelled inferred, show how confident we are, and still link to the source.
AI estimates are different. If a candidate has said something related but hasn’t answered a question directly, we may use an AI model to make a careful guess. Each estimate names its source, explains the reasoning and says how confident it is. It isn’t the candidate’s own answer, but it does count toward your match score.
One position per candidate, per question
We treat a candidate’s own material and their party’s material the same way. The source link is always right there, so you can see who said it and decide how much to trust it.
A sourced position always wins. If there isn’t one, an AI estimate fills the gap. If the source doesn’t lean either way, the estimate is neutral. Either way, it’s labelled as an estimate, not as something the candidate said.
How the match is calculated
Each question uses a five-point scale, from strongly disagree to strongly agree. Your score on a question comes from how close your answer is to the candidate’s: the same answer scores 100%, one step apart scores 75%, and opposite ends score 0%.
Not every question counts equally. You can mark each one as not important, somewhat important, important or very important, and the ones you care about most carry more weight in your match — if you don’t touch these settings, every question is weighted as important by default. AI estimates carry a little over half the weight of a sourced position, so they move your score less.
Your match is the weighted average across every question you and the candidate both answered. Questions you skip, or mark as undecided or prefer not to answer, don’t count.
What this can’t tell you
- Whether a candidate will actually do what they say, or whether a mayor even has the power to do it.
- Anything about a candidate’s track record, skills or conduct.
- How a candidate would handle an issue that isn’t asked about here.
- Whether a candidate has changed their mind since the source was published.
Corrections
If we got a position wrong, it’s out of date, or it’s credited to the wrong person, let us know and we’ll fix it. Email [email protected] with the candidate, the question and a source, and we’ll make the correction and keep a record of the change. Candidates and campaigns are welcome to send us their positions directly. A candidate’s own answer is the best source we can get.
About the data
The data was last updated on October 2, 2026, and sources were last checked on October 2, 2026. The research, the tools that build the data and this site are all open source, so you can trace any number here back to where it came from.
This project was inspired by CBC’s Vote Compass. It isn’t affiliated with CBC or Vote Compass.