Election results and political polls answer different questions, and confusing them can make an accurate update sound misleading. This guide explains how to read vote totals, margins, turnout, projections, certified results, sample sizes, margins of error, and likely-voter models so you can follow US election updates and policy debates with greater precision.
Overview
An election result describes votes that were cast and counted. A poll describes opinions measured among a selected group of respondents at a particular time. Neither should automatically be treated as a forecast, and neither tells the whole story by itself.
Results can change as election offices count additional ballots, resolve questions about eligibility, or complete required canvassing procedures. News organizations may report a projected winner before the result is formally certified, while election officials follow their own legal process. A projection is an editorial assessment based on available information; certification is an official administrative step.
Polls require a different kind of caution. A survey may show a candidate leading among registered voters, likely voters, or a narrower subgroup. It may measure a race for a specific office, approval of a policy, or public opinion on an issue. The wording, timing, sample, weighting, and response method all affect what the poll can reasonably show.
The most reliable reading begins with a simple question: Is this information describing what has happened, estimating what may happen, or measuring what people currently say?
What to track
Use the following reference table when reviewing election results or a new political poll.
| Term | What it means | What it does not mean |
|---|---|---|
| Vote total | The number of reported votes attributed to a candidate or choice at a given point. | It may not be the final total if counting or canvassing is still underway. |
| Margin | The difference between the leading choice and the next choice, usually measured in votes or percentage points. | A narrow margin does not necessarily indicate an error; it indicates a close result. |
| Turnout | The number or share of eligible or registered voters who participated, depending on the denominator used. | A turnout percentage is not meaningful without knowing which population was used. |
| Projection | A news organization’s assessment that one candidate is highly likely to win based on available returns and other information. | It is not the same as an official certification. |
| Certified result | An official result completed through the election jurisdiction’s required canvassing and certification process. | Certification does not mean every earlier news report was a final result. |
| Sample size | The number of people included in a poll’s reported sample or a particular subgroup. | A larger sample does not remove wording problems, selection bias, or nonresponse concerns. |
| Margin of error | A model-based estimate of the range of sampling variation for a poll under stated assumptions. | It does not measure every possible source of error and should not be used as a guarantee. |
| Likely-voter model | A pollster’s method for estimating which respondents are most likely to vote. | It is not a direct count of who will cast a ballot. |
Results: follow the status, not only the headline
When reading election results, record the jurisdiction, office, election type, reporting status, vote totals, and margin. A headline may refer to an early count, a projected winner, or a certified outcome. Look for labels such as “unofficial,” “partial,” “projected,” or “certified,” and check whether the report explains what remains outstanding.
Also note that vote totals and percentages can move at different speeds. If additional ballots are added, the total number of votes rises and the percentages may change. A small shift in percentage points does not necessarily represent a reversal in voter preference; it may reflect which batches were counted next.
Polls: inspect the methodology before the result
For a poll, identify the field dates, sponsor, pollster, sample size, population surveyed, question wording, response method, and weighting approach. A poll conducted before a major event may not reflect opinion afterward. A poll of registered voters is not directly interchangeable with one using a likely-voter screen.
Compare polls only when their questions, populations, dates, and methods are reasonably similar. Treat one survey as a snapshot rather than proof of a lasting trend. Several independent polls pointing in the same direction can provide broader context, but aggregation does not eliminate uncertainty.
Cadence and checkpoints
Different information deserves different monitoring schedules.
- During election night or a major primary: Check the election office and reputable live updates for the reporting status, then return later for updated totals. Avoid treating early returns as representative of the final electorate.
- During the canvass period: Track whether the result is unofficial, projected, certified, or subject to a required process such as a recount or challenge. The applicable process depends on the jurisdiction and election.
- During a campaign: Review new polls by field date and methodology instead of reacting to every isolated percentage change. A weekly or periodic comparison is often more informative than minute-by-minute monitoring.
- During a policy debate: Check whether a poll measures general support, support for a specific proposal, or approval of an official’s handling of an issue. Small wording differences can change what respondents are being asked to evaluate.
- After rules change: Recheck official election-administration information, deadlines, ballot procedures, and definitions before publishing an explainer or updating a tracker.
For publishers, maintain a simple update log. Record the date checked, the source category, the status of the result or poll, and any terminology that changed. This makes recurring election coverage easier to audit and reduces the risk of carrying an outdated label into a new article.
How to interpret changes
Separate a change in the underlying event from a change in the information available about it. A candidate’s vote share may change because more ballots were counted, not because votes already reported were altered. A poll’s estimate may change because a new sample was surveyed, not because every voter changed position.
Be careful with the margin of error. If two candidates are separated by less than, or by a similar amount to, a poll’s stated margin of error, the poll should generally be described as showing a close contest rather than a definitive lead. The margin of error is not a license to add and subtract percentages mechanically, especially when comparing different polls or subgroups.
Look for the difference between statistical uncertainty and practical significance. A small movement may be real but politically unimportant, while a modest change in turnout or support in a closely contested jurisdiction could matter greatly. Explain both the size of the change and the context in which it occurred.
Finally, avoid converting correlation into cause. If support changes after a debate, court decision, economic report, or news event, that timing may be relevant, but it does not by itself prove why opinions moved. Use cautious language such as “followed,” “coincided with,” or “may reflect” unless stronger evidence is available. For broader news-literacy practices, see How to Verify Breaking News Before You Share It.
When to revisit
Revisit this guide whenever a new election cycle begins, a primary or general election is underway, a ballot measure is being debated, or a major policy poll attracts attention. Editors should also review it on a monthly or quarterly schedule when maintaining an election tracker, and immediately after election-administration rules or official terminology change.
Before sharing a new result or poll, use this five-step checklist:
- Identify whether the item is a reported result, projection, certified result, or opinion poll.
- Check the date, jurisdiction, office, population surveyed, and reporting status.
- Read the methodology, including sample size, question wording, margin of error, and likely-voter model when available.
- Compare the claim with an official election source or the poll’s full release rather than relying only on a social post or headline.
- Describe uncertainty plainly and update the wording when the underlying status changes.
This approach keeps election updates useful after the first headline fades. It also helps readers distinguish what voters have done, what officials have confirmed, and what surveys suggest people may do next.
