Technical Documentation
A full account of how Pollfinity Research designs, fields, weights, and reports public opinion surveys — and why we believe methodological transparency is non-negotiable.
Pollfinity Research is a new polling operation. We launched our first independent surveys in mid-2026, using an SMS-to-web panel methodology built on a registered-voter sampling frame. This page describes what we actually do today, not an idealized end state — as our sample sizes, weighting practice, and infrastructure mature, we will update this page to match and will flag material changes by date.
Every design decision we make — from sample selection to weighting approach to margin-of-error calculation — is documented and published alongside our topline results, survey by survey. Because our surveys currently vary in sample size and weighting method as we build out our process, the specific methodology for each survey is authoritative for that survey — see the "Survey Methodology" box on each release page. This overview describes our general approach and current limitations.
Selected respondents receive an SMS message from Pollfinity Research containing a unique survey link. The link directs respondents to a mobile-optimized web survey. Outreach is conducted in compliance with the Telephone Consumer Protection Act (TCPA); respondents may opt out at any time by replying STOP.
SMS-to-web delivers results faster and at lower cost per completed interview than live-caller telephone surveys, and the web survey format allows for richer questionnaire design including visual ballot-style question presentation. Like all survey modes, SMS-to-web produces a low response rate; we disclose the actual response rate for each survey in that survey's methodology box.
A registration screener is applied to every survey: respondents who report they are not registered to vote in the surveyed state are kept in the published dataset and flagged, so any user can exclude them. The number flagged is disclosed in each survey's methodology box. Every respondent is also asked a self-reported turnout question ("How likely are you to vote in the upcoming election?"), and that distribution is published in the topline. A turnout/likely-voter screen beyond registration status is not applied by default; where a survey does apply one, that survey's methodology box states so explicitly, along with the resulting base size.
As of September 2026, most Pollfinity surveys are published unweighted or with a single-variable weight (most commonly recalled 2024 presidential vote, compared against each state's certified result). We are not yet running a full demographic rake across party registration, geography, age, and gender — building that requires frame targets we are still assembling on a state-by-state basis.
A full multi-variable weighting scheme (party registration, geographic region, age cohort, and gender, with race/ethnicity and education added where cell counts allow) is our target end state and something we are actively building toward. Each release page's methodology box states plainly whether that survey was weighted, and on what variable(s).
Our surveys to date have ranged from roughly 200 to 450 completed interviews per state, smaller than the samples run by longer-established national pollsters. Smaller samples carry a wider margin of error and less precision in subgroup analysis; we disclose the exact sample size and resulting margin of error on every release page and do not describe any survey as more precise than its sample supports.
As a reference point, the margin of error (MoE) for a simple random sample of n = 1,500 at the 95% confidence level is approximately ±2.5 percentage points for a 50%/50% proportion. Our actual per-survey sample sizes and margins are smaller; see each release's methodology box for the figure that applies to that survey. For proportions further from 50% (e.g., 20% or 80%), the margin narrows.
The formula for a proportion's 95% confidence interval margin is:
Reported margins of error apply to the full sample. Subgroup analyses (e.g., results among Republicans only, or among voters in New York City) carry larger margins owing to smaller cell sizes. Pollfinity Research reports subgroup margins in its full technical crosstabs.
Important caveat: Sampling error is one source of total survey error. Other sources include non-response bias, coverage error (voters not reachable through our sampling method), question wording effects, and weighting model assumptions. We endeavor to minimize each of these and to document them honestly.
Pollfinity's initial contact list is drawn from a registered-voter sampling frame with a known selection probability. However, because participation requires a respondent to voluntarily complete a web survey after receiving an SMS invitation, and our current data pipeline does not calculate response-rate-adjusted design effects, Pollfinity's realized survey samples should be treated as non-probability samples in practice.
The reported margin of error describes sampling variability under a simple-random-sample assumption; it does not account for self-selection or non-response bias, both of which are real and currently undisclosed sources of error beyond the stated margin. We are publishing this distinction plainly rather than letting the presence of a margin-of-error figure imply a level of rigor the underlying sampling process does not yet support.
Pollfinity Research LLC does not accept funding from candidates, campaigns, political action committees, political parties, or advocacy organizations. All survey work is self-funded. This independence is structural: it is enforced by our operating policy, not merely stated as an aspiration. Pollfinity is owned by Lars Mapstead, a 2024 Libertarian presidential candidate. We disclose that on every release.
We publish results regardless of which candidate or position leads. We do not offer private or proprietary polling services and have no commercial relationship with any electoral actor.
For every published survey, Pollfinity Research provides:
We would rather disclose a limitation than omit it. If a survey's sample skews on a measurable dimension — for example, a 2024 presidential vote recall that diverges meaningfully from the state's certified result — we say so on that survey's release page rather than only in an internal note.
Full crosstabs are published as a public download alongside every release at pollfinity.com. Raw, anonymized respondent-level data and additional technical documentation are available on request to journalists and academic researchers. Contact polls@pollfinity.com.
Prior to publication, each dataset undergoes the following review steps. As with the rest of this page, we describe our current practice rather than an aspirational checklist, and we are expanding this list as our data pipeline supports more checks.
Three checks are not yet part of our process and we are not claiming them. Our current survey platform (Prompt.io) exports a single submission timestamp per respondent and no duration field, so we cannot screen for implausibly fast completions until we obtain timing data from the platform or add our own. Our instruments do not yet include a dedicated trap or attention-check item, so there is no embedded consistency check for a respondent to fail. We also do not yet cross-validate topline figures against subgroup breakdowns as a standing automated check. Each is a straightforward addition we intend to make; we will update this page and add them to the list once they are actually running, rather than describing them here as current practice.
Because our current samples are not weighted on a full demographic rake, a survey's composition can diverge from a state's actual electorate in ways that affect head-to-head ballot test results more than they affect attitudinal questions. Where we have a benchmark to check against (most commonly recalled 2024 presidential vote compared to the state's certified result), we disclose the size of that gap on the release page.
Pollfinity does not apply a fixed numerical cutoff for when a ballot test may lead a release. That judgment is made survey by survey, and the reasoning, including the recalled-vote gap against the certified result, is disclosed on each release. Ballot-test results are always published in full in the topline and crosstabs.
In general elections, every ballot-qualified party nominee is included, major or minor. When a race has more than 10 candidates, we include party nominees plus the top fundraisers, ranked by total receipts (including candidate self-funding) in the most recent state or FEC filing before fielding, up to 10 names. The filing used is cited on each release. Every ballot test offers "Someone else" with a write-in; write-in names are published, and any candidate drawing 2% or more is added to our next survey of that race. Candidate order is randomized.
Journalists, academics, and researchers may request full crosstabs, raw (anonymized) data, or technical documentation for any published Pollfinity Research survey. Contact: polls@pollfinity.com.