- The likely voter (LV) screen is the single most consequential methodological choice in election polling — it can shift a result by 5+ points compared to a registered voter (RV) sample, typically in the Republican direction.
- LV polls show a Republican advantage of 2-4 points compared to RV polls because Democratic-leaning groups (young voters, occasional voters) are more often screened out as unlikely to vote — a systematic partisan skew baked into the methodology.
- The seven-question Gallup LV model is the gold standard, but pollsters use dozens of variants with meaningfully different selectivity, making LV polls from different organizations non-comparable without understanding each model.
- In 2022, LV polls significantly underestimated Democratic turnout: the predicted "red wave" didn't materialize because post-Dobbs abortion mobilization drove higher-than-expected participation among young and irregular voters that LV screens had filtered out.
- For 2026, the same risk applies: if Medicaid cuts, DOGE, and abortion access generate Democratic enthusiasm among irregular voters, LV models calibrated on 2020-2022 baselines may again systematically underestimate the Democratic share of the actual electorate.
LV vs. RV Polling Comparison: Key 2026 Pollsters
| Pollster | Methodology | LV or RV | Generic Ballot (R vs D) | LV-RV Gap Estimate |
|---|---|---|---|---|
| Marquette Law | Live phone | LV | D+5.8 | ~R+2 |
| Quinnipiac | Live phone | LV | D+6.1 | ~R+2 |
| Monmouth | Live phone | LV/RV both | D+5.9 / D+7.4 | R+1.5 |
| Fox News | Live phone | LV | D+4.8 | ~R+2 |
| Emerson | Online/IVR | LV | D+3.9 | ~R+3 |
| Reuters/Ipsos | Online panel | RV | D+7.2 | N/A |
| Marist | Live phone | LV | D+5.6 | ~R+2 |
How Likely Voter Screens Work: The Seven-Question Model and Its Variants
The likely voter screen is the methodological decision that most affects the partisan direction of any political poll. The classical Gallup likely voter model, developed in the 1950s, uses a seven-question battery to assess each respondent’s probability of voting: whether they know where their polling place is, whether they voted in past elections, how much thought they have given to the upcoming election, whether they plan to vote, how interested they are in politics generally, how often they vote, and whether they voted in the previous comparable election. Respondents who score above a threshold are classified as likely voters and retained in the sample; those below are dropped. The practical effect of this screen is almost always to shift the sample slightly in the Republican direction relative to registered voters, because higher-propensity voters tend to be somewhat older, somewhat more educated, and somewhat more Republican in composition. In a typical midterm, the shift from registered voter to likely voter polling is approximately R+2, meaning a D+7 registered voter poll would translate to approximately D+5 among likely voters. However, this relationship is not constant: in a high-enthusiasm Democratic environment — such as 2018, when Democratic voter enthusiasm was extremely high — the LV/RV gap narrowed significantly because Democratic voters were showing up at high rates despite traditional models suggesting they wouldn’t. This is exactly why the 2018 cycle was well-predicted by polling: the surge in Democratic enthusiasm translated into actual higher Democratic turnout that the LV screens captured reasonably well.
Why 2022 LV Polls Were Wrong and What Pollsters Learned
The 2022 midterm cycle produced one of the most significant polling failures in recent history, specifically around likely voter screens. Most live-interview polls showed Republicans with a significant generic ballot advantage of R+2 to R+4, which translated to forecasts of a substantial Republican “red wave” capturing 20-40 House seats. Democrats instead gained Senate seats and lost only a handful of House seats, a dramatically better-than-expected performance. Post-election analysis identified several contributing factors, but likely voter screens were central: polls were systematically undercounting younger voters, suburban college-educated women, and first-time 2020 voters who did not appear in 2018 validated voter files. These groups — particularly motivated by abortion rights after the Dobbs decision in June 2022 — showed up at unexpectedly high rates, and the LV screens that relied on past voting history had no mechanism to capture their intentions. The lesson pollsters have drawn is that validated past-vote screens need to be supplemented with current enthusiasm measures, and that in environments where the abortion issue or a comparable high-salience mobilization issue is present, simple LV screens will undercount the motivated non-habitual voter. For 2026, where Medicaid cuts, democratic norm concerns, and economic grievances are potential mobilization issues on the Democratic side, the LV/RV gap may be narrower than the historical R+2 average.
What This Means for 2026
Likely voter screens in 2026 will likely show Republican candidates performing R+1 to R+2 better than registered voter polls, based on historical midterm patterns. However, if Democratic enthusiasm remains elevated around Medicaid, healthcare, and democracy issues, the actual LV-RV gap may be smaller than expected — as it was in 2022. Readers should look for pollsters who report both LV and RV numbers and compare the gap.
