Mistaking a projection for an official result
A projection is an assessment that a candidate is likely to win, made before the election authority has completed its formal process. News organizations and other analysts may use reported vote totals, information about ballots still to be counted, and statistical methods to make that assessment. A projected winner can be a useful piece of information, but the projection does not itself count ballots or determine who takes office.
Official results come from election administrators through procedures that can include checking returns, resolving eligible ballots, and certifying totals. The details vary by jurisdiction. When following election night, look for the label attached to a number or announcement: a reported vote total, an estimated vote share, a projected winner, and a certified result are different things.
Assuming the first lead will last
Early results reflect the votes that have been reported so far, not necessarily a representative sample of all votes cast. One area may finish reporting before another. In-person, early, and mail ballots may also be processed or released in different orders, depending on local rules. A candidate can therefore lead in the first returns and later fall behind without anything unusual having happened.
Instead of asking only who is ahead, ask where the reported votes came from and what remains. Are populous areas still outstanding? Are particular kinds of ballots still being processed? Even a large-looking lead may be difficult to interpret without that context. The share of precincts reporting can help, but precincts differ in size, and that figure may not describe how many votes are still uncounted.
Treating percentages as the whole story
A vote percentage describes a candidate’s share of votes in a stated set of returns. If that set grows, the percentage can change. Likewise, a margin of several percentage points among reported votes is not automatically a prediction of the final margin. The remaining ballots may come from places or voting methods with different patterns from those already reported.
Pay attention to the denominator. “Half the votes counted” and “half the reporting units counted” do not mean the same thing. An estimate of outstanding votes is useful but may be revised as more information becomes available. When a graphic displays a precise percentage, read its accompanying label before treating that precision as certainty about the final outcome.
Confusing polls, models, and race calls
Pre-election polls measure the preferences of the people they reach, subject to sampling and other uncertainties; they are not vote counts. Surveys of voters conducted around election day also have limitations. A projection model may combine survey information with actual returns and assumptions about votes yet to be reported. A race call is the analyst’s conclusion from the available evidence, not a separate batch of votes.
Different outlets can make calls at different times because they use different methods or thresholds for confidence. An outlet that has not called a race is not necessarily saying it expects a different winner. It may simply judge that too much uncertainty remains. Compare what each source is claiming rather than treating the timing of a call as a result in itself.
Reading normal updates as evidence of wrongdoing
Election-night totals are provisional. They can rise in batches, and posted figures may occasionally be corrected as election offices reconcile records or fix reporting errors. A sudden change on a chart does not, by itself, explain why it happened. Before drawing a conclusion, check whether the change corresponds to newly reported areas, a batch of ballots, or a correction acknowledged by the reporting authority.
A useful habit is to separate an observation from an explanation. “The lead changed after a new batch was reported” describes what the displayed totals show. A claim about why that batch favored a candidate requires additional evidence. If the outcome is close, patience matters: outstanding ballots and routine verification steps may be consequential. Election-night coverage is a developing account of the count, not a substitute for the completed result.
