Poll Types We Use
Multiple Choice
Voters select one option. Best for head-to-head matchups and clear either/or questions. Results shown as percentage of total votes.
Preference Ranking
Respondents drag-and-drop to rank candidates or options by preference. Results display first-choice percentages and cumulative preferences of the respondents.
Scale (1–10)
Voters place themselves on a scale. Ideal for measuring intensity of opinion, not just direction. We report average score and distribution.
Time-Limited Polls
Some polls have defined open and close dates. This allows us to compare opinion at specific moments during a campaign or legislative session.
How Votes Are Counted
Each respondent may cast one vote per poll. We use a combination of IP address and browser fingerprinting to prevent duplicate votes from the same device. This is not a perfect system — a determined person can vote more than once — but it provides meaningful friction for casual duplication.
All votes are stored in a secure database. We do not collect or store personally identifiable information about voters.
Sample & Representativeness
Important disclosure: PredictOregon uses different methodologies depending on the poll. Our methodology for each poll is documented in full on the relevant results page.
Poll #3 (2026 Republican Gubernatorial Primary, May 2026) used live telephone interviews conducted by trained volunteer interviewers working from L2 voter file contact data, supplemented by a professional CATI (computer-assisted telephone interviewing) vendor. Respondents were drawn from a stratified random sample of registered Republican likely voters. Fewer than 20 respondents who could not complete a live interview were offered a secure, one-time web link to complete the survey. Results are weighted by gender. This poll was a scientific probability-based sample with a margin of error of ±3.83% at 95% confidence.
Polls #1 and #2 used a self-selected online sample. Respondents chose to participate and were not randomly selected. Those results cannot be treated as statistically representative and should be interpreted as directional indicators from engaged Oregon voters, not a scientific probability sample. We disclosed this limitation prominently in all published outputs for those polls.
We believe transparency about methodology is essential. Many widely-cited polls make no such disclosure. We do, because informed readers make better decisions with honest data.
We believe transparency about this limitation is essential. Many widely-cited online polls make no such disclosure. We do, because we believe informed readers make better decisions with honest data than with numbers that falsely claim precision they don't have.
Question Design
We take question wording seriously. Subtle changes in how a question is asked can dramatically shift results. Our principles are to use plain language that any Oregon voter would understand, avoid leading or loaded language, present options in a balanced order, and include "Undecided" or "Other" options where appropriate so voters aren't forced into a false choice.
Methodology Evolution
Predict Oregon's methodology has evolved as the organization has grown. Our first two polls used self-selected online samples — appropriate for the exploratory, directional research they were designed to produce. Poll #3 used a stratified random sample drawn from Oregon's statewide voter file, used telephone interviewing, demographic weighting, and built our own CATI dashboard. You can expect more of these polls in the near future.