Polling & Election Surveys

A survey conducted before an election to gauge voter preferences and predict likely outcomes. Pre-election polls are the most widely covered form of political polling and the source of most horse-race election coverage. Their accuracy depends heavily on how well the pollster identifies likely voters, how the questions are worded, and how representative the sample is of the actual electorate. Pre-election polls have faced increasing scrutiny in recent cycles following high-profile misses in 2016 and 2020.

A survey conducted outside polling places on Election Day in which voters are asked how they voted and why, as well as demographic questions about their age, income, education, and other characteristics. Exit polls serve two purposes: they are used by news organizations to project winners before all votes are counted, and they provide detailed data about the composition of the electorate and the voting patterns of different demographic groups. Exit poll projections are released only after polls close in a given state to avoid influencing voters who have not yet cast their ballots.

A survey designed to measure the views, preferences, and attitudes of a defined population on a given topic. In the context of elections and governance, public opinion polls track voter preferences for candidates, approval ratings for elected officials, and public attitudes toward policy issues. The quality of a public opinion poll depends on the size and representativeness of the sample, the neutrality of the question wording, and the rigor of the methodology used to conduct and analyze it.

A poll conducted repeatedly over a period of time using a consistent methodology, allowing changes in public opinion or candidate support to be measured from one period to the next. Tracking polls are commonly used during campaigns to monitor momentum and measure the impact of events, debates, advertisements, and news coverage on voter preferences. Because they are conducted frequently and often use smaller daily samples, individual days in a tracking poll can show significant fluctuation that may reflect statistical noise rather than genuine shifts in opinion.

A poll conducted by or for a political campaign or party for the purpose of informing campaign strategy rather than for public release. Internal polls often use larger samples and more detailed questions than public polls, and campaigns consider them among their most sensitive strategic assets. When campaigns do release internal poll results publicly, it is typically because the results are favorable and they want to shape the media narrative, which means publicly released internal polls should be evaluated with additional skepticism.

A form of political telemarketing disguised as a survey, in which large numbers of voters are called and asked leading questions designed to plant negative information about a candidate rather than to genuinely measure opinion. Push polls are not legitimate research tools but rather a form of political persuasion and opposition research delivery. A key distinguishing feature is scale: a genuine poll samples a few hundred to a few thousand people, while a push poll may contact tens or hundreds of thousands. The term is sometimes mistakenly applied to legitimate polls that simply ask tough or negative questions about candidates.

A methodology used by pollsters to estimate which registered voters are actually likely to cast ballots in a given election, and then to weight or filter the poll sample accordingly. Because turnout varies significantly across elections and demographic groups, a poll of all registered voters will often produce different results than a poll of likely voters. Different pollsters use different criteria to define likely voters, including past voting history, stated intent to vote, and level of interest in the election, which is one reason polls from different organizations can show different results even when conducted at the same time.

A polling methodology that surveys all registered voters rather than filtering down to likely voters. Registered voter polls tend to show somewhat different results than likely voter polls, often with greater support for Democratic candidates, because the pool of registered voters is broader and includes many people who may not ultimately cast ballots. Registered voter polls are sometimes used early in an election cycle before turnout patterns become clearer.

A statistical measure of the uncertainty in a poll's results, reflecting the fact that a poll surveys a sample of the population rather than the entire population. A margin of error of plus or minus three percentage points means that the true value in the full population is likely within three points of the reported figure. Crucially, the margin of error applies to each candidate's number individually, meaning the actual margin between two candidates could be as much as double the stated margin of error. Results within the margin of error are considered too close to call.

The set of procedures and techniques a pollster uses to design, conduct, and analyze a survey. Key elements of polling methodology include how the sample is selected, how respondents are contacted (phone, online, mail, in-person), how the sample is weighted to match the target population, how likely voters are identified, and how the questions are worded and ordered. Transparent disclosure of methodology is a hallmark of reputable polling organizations and is essential for evaluating the reliability of results.

A flaw in a poll's methodology in which the sample of people surveyed does not accurately represent the population being studied, leading to results that are systematically skewed in a particular direction. Sampling bias can arise from many sources, including the method used to contact respondents, differential response rates among demographic groups, and errors in the weighting process. Polls with significant sampling bias can produce results that are consistently wrong in predictable ways, which is one explanation offered for polling errors in recent presidential elections.

A composite estimate of candidate support or other survey measures calculated by averaging results across multiple polls, often weighted by recency, sample size, and pollster quality. Polling averages are used by election forecasters and analysts to smooth out the noise in individual polls and provide a more stable picture of where a race stands. Organizations like FiveThirtyEight and RealClearPolitics publish polling averages that are widely followed during election seasons.

A measure of how positively or negatively the public views a particular person, organization, or policy, typically expressed as the percentage of respondents who have a favorable opinion versus the percentage who have an unfavorable opinion. Favorability ratings are tracked for candidates, elected officials, political parties, and major institutions. A candidate's favorability rating is considered one of the most reliable predictors of electoral performance and is closely watched throughout a campaign.

A measure of how many people approve of the job an elected official is doing, typically expressed as a percentage of survey respondents. Approval ratings are most commonly tracked for the president but are also measured for governors, members of Congress, and other officials. A president's approval rating is closely watched as an indicator of their political strength, their party's prospects in upcoming elections, and the public's assessment of how major events and policies are being handled.

A poll question that asks respondents whether they would vote for a generic Democratic or Republican candidate for Congress, without naming specific candidates. The generic ballot is used as a broad indicator of the national political environment and which party has the advantage heading into a midterm or congressional election. A party that leads on the generic ballot by a significant margin is generally favored to perform well in competitive congressional races, though the relationship between the generic ballot and actual seat gains is not perfectly linear.

A poll that measures support for two specific candidates matched against each other in a hypothetical or actual electoral contest. Head-to-head polls are the most direct measure of where a specific race stands and are used to track the relative standing of candidates throughout a campaign. In primary seasons, multiple head-to-head matchups may be tested to assess how different potential nominees would perform against their likely general election opponent.

A poll conducted specifically within a competitive state that is considered likely to be decisive in a presidential or other statewide election. Because presidential campaigns are won through the Electoral College rather than the national popular vote, state-level polls in battleground states are often more predictive of election outcomes than national polls. Battleground state polls are heavily scrutinized during presidential campaigns and are a primary driver of campaign resource allocation decisions.

Short for cross-tabulations, the breakdowns of poll results by demographic subgroups such as age, race, gender, education level, party affiliation, and geographic region. Crosstabs allow analysts to see not just overall topline results but how different groups within the sample responded. They are an essential tool for understanding which voters a candidate is winning or losing and why, and they are typically included in the full data release that accompanies a publicly released poll.

A service or organization that collects results from multiple polls and combines them into a composite estimate or forecast. Poll aggregators weight individual polls by factors such as recency, sample size, and the historical accuracy of the polling organization. By combining multiple data points, aggregators aim to provide a more reliable picture of public opinion than any single poll can offer. Well-known aggregators include FiveThirtyEight, RealClearPolitics, and the polling averages published by major news organizations.