The primary difference between registered voters and likely voters in political polling is that registered voters are simply everyone signed up to vote, while likely voters are a filtered subset deemed highly probable to actually cast a ballot. Understanding this distinction is crucial because it explains why two polls conducted at the exact same time can display noticeably different results.
When you look at political data, you are not just looking at a snapshot of public opinion; you are looking at a specific slice of the population. Whether a pollster decides to survey “all adults,” “registered voters,” or “likely voters” dramatically influences the final numbers. For anyone trying to make sense of campaign season, decoding these classifications is the first step to reading the data like a professional.
What is a Registered Voter (RV)? #
A registered voter (RV) is any individual who is legally eligible to vote and has successfully completed the registration process in their state. This group represents a broad cross-section of the politically active and semi-active population.
When pollsters conduct surveys of registered voters, they are capturing the opinions of people who have the potential to vote. However, registration does not guarantee participation. In a typical US presidential election, only about 60% to 66% of voting-eligible citizens actually cast a ballot. In midterm or primary elections, that number drops even lower—often between 30% and 50%.
Because millions of registered voters choose to stay home on Election Day, RV polls tend to paint a broader, more generalized picture of public sentiment. They tell us what the registered public thinks, but not necessarily what the actual electorate will look like when the voting booths close.
For this reason, early in an election cycle, pollsters rely heavily on registered voters because it is too early to accurately predict who will show up. If you are tracking the latest presidential polls a year out from an election, you will almost exclusively see registered voter samples.
What is a Likely Voter (LV)? #
A likely voter (LV) is a registered voter whom a pollster believes, based on a variety of modeling techniques, has a high probability of actually voting in the upcoming election.
There is no official government list of “likely voters.” Instead, every polling organization must design its own proprietary formula—known as a “likely voter screen”—to determine who fits this description. Because these formulas are proprietary and vary from one pollster to another, one poll’s definition of a “likely voter” might be slightly more or less restrictive than another’s.
Likely voter polls are the gold standard for predicting election outcomes. By filtering out the individuals who are registered but unlikely to make the effort to vote, LVs offer a more precise projection of the actual electorate. As Election Day approaches, reputable polling organizations shift their methodology from registered voters to likely voters to increase their accuracy.
How Pollsters Determine Who is “Likely” to Vote #
Because “likeliness” is a projection, pollsters must act as social scientists and data analysts to build their screens. They generally use a combination of direct questions, historical data, and demographic modeling.
Here are the primary metrics pollsters use to calculate a respondent’s likelihood of voting:
1. Self-Reported Intent and Interest #
The simplest way to find out if someone will vote is to ask them. Pollsters often ask questions such as:
- “On a scale of 1 to 10, how likely are you to vote in the upcoming election?”
- “How much thought have you given to the election?”
- “How closely are you following the news about the campaigns?”
While useful, self-reported intent has a major flaw: social desirability bias. People want to appear to be good citizens, so they frequently tell pollsters they are “definitely voting” even when they have no real intention of doing so.
2. Verified Past Voting History #
To bypass the bias of self-reporting, many high-quality pollsters match their survey respondents to official state voter files. These files are public records showing whether a person voted in previous elections (though they do not show who the person voted for).
If a respondent voted in the 2020 and 2024 presidential elections, as well as the 2022 midterms, they are highly likely to vote again. If they have never voted in a midterm or primary, they are flagged as less likely to participate in those specific cycles.
3. Logistical Knowledge #
Pollsters sometimes ask practical questions to test a respondent’s true preparation. Questions like “Do you know where your local polling place is?” or “Do you know what identification is required to vote in your state?” help filter out casual observers from active participants.
4. Demographic and Turnout Weighting #
Finally, pollsters look at historical turnout trends. Historically, older, wealthier, and highly educated citizens vote at much higher rates than younger, lower-income, or less-educated citizens. A pollster’s likely voter model will adjust the weight of these demographic groups based on what they expect the final turnout composition to look like on Election Day.
Using an interactive election tracker allows you to see how different pollsters apply these screens, as some methodologies lean more conservative or progressive based on how tightly they restrict their likely voter pools.
Why the Difference Shifts the Poll Numbers #
The transition from a registered voter poll to a likely voter poll almost always changes the margin between candidates. This shift occurs because the demographics of those who can vote (registered voters) differ substantially from the demographics of those who do vote (likely voters).
Traditionally, likely voter screens tended to favor the Republican Party. Historically, the Republican base consisted of older, wealthier, and more reliable voters, whereas the Democratic base relied heavily on younger voters, lower-income individuals, and minority groups—demographics that historically turn out at lower rates. Therefore, when a pollster applied a likely voter screen, the Republican candidate would often see a 1- to 3-point bump compared to the registered voter poll.
However, in recent election cycles, this traditional dynamic has shifted due to educational and demographic realignment. Highly educated voters (who vote at very high rates) have increasingly shifted toward the Democratic Party, while working-class and non-college-educated voters (who can have more volatile turnout rates) have shifted toward the Republican Party.
Consequently, the “likely voter premium” is no longer a guaranteed advantage for one side. Depending on the state, the specific election, and the turnout model used, the likely voter screen can shift the margin in either direction. By monitoring fluctuations in voter sentiment across both RV and LV models, you can observe these demographic tug-of-wars playing out in real-time.
How to Read Polls with This Knowledge #
When analyzing political data, always look for the sample label, which is typically listed next to the poll’s sample size (e.g., “N=1,000 LV” or “N=1,200 RV”).
Here are three rules of thumb for reading the polls:
- Compare apples to apples: Never compare a registered voter poll from June to a likely voter poll from October and assume a candidate has gained or lost momentum. The change in numbers might simply be the result of a more restrictive voter screen being applied.
- Expect tighter margins in LV polls: Likely voter polls generally reflect a more partisan, politically engaged audience. Because of this, undecided percentages are usually lower in LV polls than in RV polls.
- Look at the date: Be skeptical of likely voter polls conducted very early in the cycle (such as the spring of an election year). At that stage, turnout models are highly speculative. Trust RV polls early on, and transition your focus to LV polls after Labor Day.
Frequently Asked Questions #
Which poll is more accurate: RV or LV? #
Likely voter (LV) polls are generally more accurate for predicting the final outcome of an election, especially as Election Day draws near. Registered voter (RV) polls are better for understanding general public opinion or the maximum potential coalition a candidate could assemble if they successfully maximized turnout.
What is the difference between “All Adults” and “Registered Voters”? #
An “All Adults” (sometimes labeled as “A”) sample includes everyone over the age of 18, regardless of their citizenship or registration status. These polls are useful for broad policy issues but are the least accurate for predicting elections because they include non-citizens and millions of people who are not registered to vote.
Do likely voter models ever get it wrong? #
Yes. If an election experiences historically high turnout (such as the 2020 presidential election) or unexpected demographic surges, a pollster’s likely voter model may fail to capture these new voters. If a model assumes youth turnout will be low, but young voters turn out in record numbers, the likely voter poll will underrepresent their preferred candidate.
Why do some pollsters stick to registered voters? #
Some media organizations and academic institutions prefer registered voter polls because they are more transparent and do not rely on the pollster’s subjective assumptions about turnout. This avoids “herding”—the practice of pollsters adjusting their likely voter models to match the consensus of other polls.