Start with the reference period.
Election statistics are snapshots tied to a specific publication or election cycle. Elector counts, polling stations, turnout and constituency boundaries can change, so a figure should be read together with its date and source.
Compare like with like.
Before comparing two elections, check whether the constituency boundaries, category definitions and reporting units are consistent. A numerical difference can be misleading when the underlying geography or definitions changed.
Separate results from interpretation.
Official results describe recorded votes and outcomes. Explanations about why an outcome occurred require additional evidence and should be presented as analysis rather than as a fact contained in the result itself.
Use aggregate information responsibly.
Public research can examine constituency and booth-level patterns where lawful data is available, but aggregate patterns should not be converted into unsupported claims about an individual voter’s political preference.