Ask who was measured and when
Polls are tied to a fieldwork period, target population, question wording and method. A survey of registered voters and a survey of likely voters are not automatically equivalent. A national finding is not a district-level result. Begin with the original pollster’s release and methodology rather than comparing two headline percentages in isolation.
Read the market’s actual contract
Prediction-market prices concern specific contracts with resolution rules. A contract may ask about nomination, control of a chamber or another event that differs from a candidate’s vote share. Prices can change, and a displayed snapshot has a timestamp. A market quote should remain labeled as market information, rather than being rewritten as a polling percentage.
Keep a snapshot separate from a live feed
An embedded provider widget and an imported dataset can update on different schedules. Electionpedia preserves source labels and dates so readers can distinguish them. A recently refreshed page does not prove every underlying source has new observations. Check the actual provider or fieldwork date before calling a number current.
Use uncertainty as a research question
Differences between polls, markets and eventual results can reflect different questions, timing and methods. They should lead to a closer examination of the source, not a guarantee about an outcome. The available numbers can inform research while leaving the reader’s judgment—and the official result—separate.
Continue with the sources
These publications explain the institutions and record systems discussed in this guide. Individual entries link their own specific documents and dates.