By the Wavefield Research team · Published Aug 27, 2026
Survey screening questions are the short set at the start of a questionnaire that decides who belongs in the study: qualified respondents continue, everyone else is politely ended. They define who your results represent, they are the questions professional respondents try hardest to game, and they set your fieldwork cost through the incidence rate.
(Looking for job screening questions — the kind employers ask candidates? That is a different topic; this page is about research surveys.)
Every bad screening question fails the same way: it shows the respondent what answer qualifies. Panel respondents complete dozens of surveys a week and read screeners the way poker players read tells. The fixes are mechanical:
Weak: “Do you own a dog? We're surveying dog owners about pet food.”
Better: “Which of these pets, if any, live in your household?” (cat, dog, bird, fish, reptile, none)
A good screening question hides what qualifies: professional survey-takers pass anything when the wanted answer is visible. A neutral list with a none option conceals the target, and the extra options cost nothing.
Weak: “Have you purchased skincare products in the past 3 months?” Yes / No
Better: “When did you last purchase skincare products?” (past month, 1–3 months, 3–12 months, over a year, never)
A yes/no screener is a coin flip for an inattentive respondent. A time distribution collects the same fact, screens on the range you need, and doubles as a usable question.
Weak: “Are you a decision-maker for software purchases at your company?”
Better: “Which of these were you personally involved in during the past year?” (evaluated vendors, approved budget, signed contracts, none of these)
Identity-based screener questions invite flattering self-reports — everyone is a decision-maker. Screening on concrete recent behaviors is harder to fake and defines the segment more precisely anyway.
Weak: (No industry-exclusion question at all)
Better: “Do you or anyone in your household work in any of these industries?” (advertising, market research, journalism, the category being studied)
Category insiders answer like insiders, and their responses contaminate concept tests and message batteries. The classic security screener runs first, before any topic is revealed.
Wording craft beyond screeners — balanced scales, one idea per question — is covered in survey question types.
Most screeners fall into five families, and a typical study uses two or three of them:
Age, gender, region, income — “Which of the following age groups do you belong to?” Usually paired with quotas, so the same question both screens the ineligible and balances the cells that qualify.
Purchase recency, category usage, brand awareness — the recency-distribution pattern from the examples above. These carry the most gaming risk, so they get the disguising techniques.
The B2B family: involvement in decisions, budget authority, team size. Screen on concrete recent behaviors, never on self-declared titles.
The industry-exclusion question — category insiders, market researchers, journalists. Runs first, before the topic is revealed, so insiders cannot adapt their answers.
Residence in a district, registration to vote, likely-voter screens — the backbone of political polling. The strictest form: a wrong answer is a hard fact, not a soft mismatch.
Every screener is an ordinary question plus terminate logic: answers that fail route to a polite ending instead of the next question. That skip logic is part of questionnaire programming — see what survey scripting involves.
The terminate wiring itself — routing failed screeners to an ending — is covered in survey scripting, explained.
Every screening question you add lowers your incidence rate — the percentage of people who pass the screeners and qualify for the study. Incidence is the number that panel pricing runs on, because the panel must send you enough respondents to screen, not just the ones who complete. The multiplication is simple and worth staring at:
| Incidence rate | People screened per complete | Typical situation |
|---|---|---|
| 80% | ≈1.3 | General population with light screening |
| 50% | ≈2 | One meaningful filter (category buyers) |
| 20% | ≈5 | A real segment — recent purchasers of a specific type |
| 5% | ≈20 | Niche B2B roles, rare conditions, single districts |
There is no universal “good” incidence rate for a survey: 50–80% is typical for lightly screened general-population studies, 15–30% for a defined category segment, and under 10% usually means niche B2B roles, rare conditions, or a single district — with panel pricing rising steeply at each step down.
A study needing 400 completes at 20% incidence means roughly 2,000 people enter the survey. Panels price accordingly — quoted cost per complete rises steeply as claimed incidence falls, and if fieldwork reveals your real incidence is lower than you told the panel, the price gets renegotiated mid-field.
This is why a soft launch matters: field the first 5–10% of sample, read the actual incidence off the early screen-out data, and correct the estimate before committing the full budget. It is also why each screening question must earn its place: a nice-to-have filter that halves your incidence doubles your fieldwork cost.
Order matters too: put the most exclusionary screener first. Unqualified respondents exit after one question instead of four, which respects their time and, on panels that bill partial interviews, your budget.
The respondent never qualified — wrong region, wrong behavior, works in the industry. Screen-outs measure your incidence rate. They should exit early, politely, and cheaply.
The respondent qualified, but their bucket is already full — the 30th woman 18–34 when the quota wanted 25. Quota-fulls measure your quota design, not your incidence. Panels credit and bill the two differently.
Conflating the two corrupts both numbers: your incidence looks worse than it is, and your quota pressure becomes invisible. The platform should track them as separate outcomes with separate panel redirects — ours does, and neither counts as a complete: quotas fill on genuine completes only. How quota cells are set and enforced is covered in quota sampling vs. random sampling.
If respondents can tell which answer ends the survey, some will answer to stay in — and your screeners stop screening. The button should say Next on every page, never Submit, so a screener page is indistinguishable from any other. (Our runner enforces this.)
“Thanks — this study is now closed for your group” beats “You did not qualify.” Naming the disqualifying answer teaches the respondent what to say next time, and panel respondents remember.
Panel respondents must land back at the panel with the right status — complete, screen-out, or quota-full — or they don't get credited and the panel's numbers drift from yours. Separate redirect URLs per outcome are a platform feature, not a nice-to-have.
One analysis rule follows from all this: keep the screen-out records. They are not failed interviews; they are your incidence data, your screener-performance data, and your early warning when a quota cell is starving. Field monitoring should show where people exit and why — completes, screen-outs, and quota terminations as separate counts (our field monitor reports exactly this).
The reliable patterns: a category list with a none option instead of a yes/no (“Which of these pets live in your household?”), a recency distribution instead of a binary (“When did you last purchase…”), concrete behaviors instead of identity claims (“Which of these were you involved in?”), and a security screener excluding people who work in the category, market research, or media. Each hides what qualifies, resists guessing, and produces usable data on its own.
A screener survey is the short block of screener questions at the start of a questionnaire that decides whether a respondent belongs in the study at all. Qualified respondents continue to the main questionnaire; others are politely ended (screened out) — and with panel sample, redirected back to the panel so they get credited. Screeners define who the results represent, which makes them the highest-stakes questions in the instrument.
Three to five, almost always. Each additional screener lowers your incidence rate — the share of people who qualify — and cost rises in proportion: at 20% incidence you screen five people per complete, at 5% you screen twenty. Combine criteria into single well-built questions where possible, and put the most exclusionary question first so unqualified respondents exit before spending effort.
No — that is the other meaning of this search. Job screening questions are what employers ask candidates in applications and phone screens. This page covers research screeners: the questions at the start of a survey that select which respondents belong in a study. If you landed here preparing for a job application screen, the survey-methods advice below will not help you.
Related: quota sampling vs. random sampling · sample size calculator · political polling
The agent builds the screener block, wires the terminates and quota checks, and the field monitor reports incidence live. From $99 per study.
Start a study