ActiVote

§1About

✎ Drafted by AI from the public sources cited. Each sentence links to its source.

ActiVote runs a free voter-information app and publishes as polls the candidate preferences its users enter in the app, according to its website and a 2024 article by 538.^2,3,4 Its team page lists three co-founders, Sara Gifford as chief executive, Paul-Erik Raué as chief technology officer and Victor Allis as chief pollster, and says ActiVote, Inc. was incorporated on January 3, 2019.⁵ Its methodology page says respondents are app users who choose to state a preference and that results are weighted by age, gender, party, ethnicity, income, education and region; 538 reported in 2024 that the firm checks respondents against the L2 voter file and also uses L2 data for weighting.^3,4

Founded

2019-01-03⁵

§2Who it works with

Why this matters: Owners, news partners and the firms behind a poll's interviews shape what gets asked, how, and what gets published.

§3Whom it works for

Why this matters: A firm that also works for one party's campaigns can still publish accurate polls, but readers deserve to know.

Says it does no work for parties or candidates. The ActiVote website describes its app as a free, nonpartisan tool; the pages read do not say whether the company does paid work for candidates or parties.^2,3,6

Of its 32 polls listed since the 2024 election, VoteHub flags 0 as paid for by a Democratic-aligned sponsor and 0 by a Republican-aligned sponsor⁷.

§4How its numbers compare with other polls

Why this matters: A firm whose polls run steadily to one side of everyone else's is worth reading alongside others.

Each of its polls since the 2024 election is compared with other pollsters' polls of the same question whose field period ended within 14 days, each other pollster counted once⁷. The figure is the average gap, with a 95% margin. This compares it with other polls, not with the truth: the others can be the ones that are off.

Party margin

R+1.0 (±1.4) across 12 polls of the generic ballot and 2026 races, in 8 contests.

Net approval: Donald Trump

+2.1 points (±1.2) across 16 polls, against other polls of the same question.

D+ means its polls showed Democrats doing better than other polls did; R+, Republicans. A figure from fewer than 5 polls isn't shown.

§52026 record against the count

Why this matters: How close this pollster's last 2026 polls came to how people voted, measured the way FiveThirtyEight measured earlier elections.

13 polls of 9 races between a Democrat and a Republican, with field periods ending in the 21 days before November 3, 2026, compared with each race's count. Some of those races are still being counted, so these figures use the unofficial count and will change until states certify.

Average miss

R+7.8

Relative to other polls of the same races

R+0.7

Typical size of miss

7.9 pts

Error is the poll's D−R margin minus the count's, as shares of all votes in the race; D+ means it showed Democrats doing better than they did. From 13 polls in all.

§6Documented findings

Why this matters: Only findings by a professional body, a court, a regulator, an aggregator's own published decision, or the pollster itself are listed, each cited to that decision.

• Silver Bulletin, February 17, 2025. Silver Bulletin's pollster ratings published February 17, 2025, kept unchanged in an archive dated January 14, 2026, name ActiVote as an example of polls that do not meet its standards for scientific surveys, which it says never enter its database and receive no rating. It gives the reasons for that group as a whole, polls run by nonprofessionals on low-cost online survey platforms or polls built on non-random convenience samples without adequate correction, without saying which applies to ActiVote, and says such polls are not banned as such.⁸

§7Its polls

§8What we looked for and didn't find

• No headquarters location was found; the website says only that the app is made in the USA.
• Its pages name ActiVote, Inc. but do not say whether it is a for-profit company or a nonprofit, or how it is funded, so kind is left empty.
• No standing news partners or non-news clients were found; 538 reported that it works with voter-engagement nonprofits to recruit app users but did not name them.
• No outside fieldwork or panel vendor was found beyond its own app users; the L2 voter file use rests on 538's 2024 report.
• The 2024 538 article describes a rule for unrated firms that release many polls quickly; whether later averages still apply it to ActiVote was not checked.
• Beyond the finding recorded, searches found no AAPOR statement, court ruling or filing, regulator action, other aggregator statement, or retraction by the firm.