Last updated: 17 September 2026
Quick Answer: Survey screening questions decide who qualifies for a study before the main questionnaire starts. A working screener tests category behavior, quota fit, security exclusions and respondent honesty, usually in five to eight items. The qualifying answer hides among plausible alternatives, because a respondent who can see what qualifies will often select it.
In November 2024, Pew Research Center asked 11,114 online opt-in respondents about a social media platform called Fizzypress and a trip to the International Space Station. Trap questions flagged 18 percent of the sample, all of whom had cleared their supplier's qualification.
A screener is the only instrument in a study that decides who is in it. It is also the biggest driver of fieldwork cost, because incidence rate is a screener output before it is a budget line.
What Does a Screener Actually Have to Rule Out?
A screener has four jobs: confirm the category behavior the study is about, capture the variables the sample is balanced on, exclude respondents whose profession compromises the data, and catch people answering to qualify.
- Category behavior, not category attitude. Write the item so the target says yes even if others also do, then narrow. The National Academies review of finding small populations calls that high sensitivity first, verify afterward.
- Quota variables, including ones you will not screen on. Settle which matter first, so the cells are segments that hold up.
- Security and industry exclusions. The standard battery removes anyone working in market research, advertising, journalism or the category under test, plus their households.
- Answering to qualify. Pew found respondents flagged as bogus disproportionately claimed to be young or Hispanic, overclaiming on exactly the demographic items where a screener has least defense.
What Are Examples of Screening Questions?
Four families, one per heading, usable as written.
Behavioral Screening Questions
Which of the following have you personally bought in the past three months? Select all that apply. [target category] · [four adjacent categories] · None of these
Qualifies on recency plus personal purchase.
Category Usage Questions
How many times in the past 30 days have you used [category]? Numeric entry.
Banded options tell the respondent where the cut-off probably sits. A numeric box does not.
Firmographic and Role Questions
Which of the following best describes your role in choosing [product or service] for your organization? I make the final decision · I recommend, someone else approves · One of several who must agree · No involvement
The failure version is "Are you a decision maker for [product]?" Everyone says yes.
Life-Stage and Household Questions
Which of these describe your household? Select all that apply. A child under 5 · A child aged 5 to 12 · A teenager · An adult over 70 · A pet · None of these
Qualifies on life stage, the target hidden among equally plausible alternatives.
How Do You Stop Respondents From Gaming a Screener?
Hide the qualifying answer. Every item should offer several answers that look equally likely to qualify, ask about behavior rather than intention, and avoid an introduction naming the category. Put the screener at the beginning of the survey, before any stimulus or substantive question, and hold it to five to eight items.
Eckman and colleagues ran 1,200 telephone interviews in Germany across 21,539 filter questions, then checked the answers against administrative records, which beats a survey validated against itself. Where follow-ups came straight after each filter, respondents gave qualifying answers 37.9 percent of the time against 42.4 percent when filters were grouped, and the records show the difference was false negatives. They had learned that saying yes bought more work. A paid screener inverts that incentive and the same behavior produces false positives.
Allow a longer screener where verification earns its length: for a rare clinical population or a narrow B2B role the verification items are the study, though the P-FROST framework on fraudulent participation in online studies warns that anti-fraud measures deter genuine participants too.
Plant one item that cannot be true, a brand that does not exist or a service unavailable in the market. A flag is a signal to review, not a verdict: in Pew's three-method test, trap questions flagged Hispanic respondents 22 percentage points above non-Hispanic White respondents. Wording is one layer only, since whether the person is who they claim to be is platform-side verification.
How Does Incidence Rate Change Cost and Fieldwork Time?
Incidence rate is the share of screened people who qualify, and it sets the sample bill and the field calendar.
The table assumes 200 interviews and 500 screener completes a day, sorted by incidence rate, highest first.
| Incidence rate | Screener completes per interview | Screener completes for 200 interviews | Field days at 500 a day |
|---|---|---|---|
| 25 percent | 4 | 800 | 1.6 |
| 10 percent | 10 | 2,000 | 4 |
| 5 percent | 20 | 4,000 | 8 |
| 2 percent | 50 | 10,000 | 20 |
The binding cell sets the close date, not the average. Split 200 interviews across four quota cells, three at 15 percent incidence and one at 3 percent. The easy three need 1,000 screener completes between them; the hard cell needs 1,667 alone, 63 percent of the effort for a quarter of the sample. Planning against the 12 percent average buys 1,667 screeners for a study needing 2,667. The AAPOR Standard Definitions, tenth edition since 2023, is the framework for reporting eligibility consistently.
Live adaptive screening answers the binding-cell problem, because quotas that close as they fill stop a study buying into a full cell. Alchemic's managed fieldwork runs screening and quotas live, recruiting from its own panel network, from client lists, or as a hybrid top-up: managed fieldwork or bring your own.
Who Does a Screener Exclude That You Did Not Intend to Exclude?
Before a screener rejects anyone, the mode has already rejected people, and none of that reaches the incidence report.
ITU's Facts and Figures 2025 puts 2.2 billion people outside the internet entirely, with mobile broadband coverage nearly universal while affordability gaps persist. Literacy filters developed markets too: the 2023 PIAAC results for the United States put 28 percent of adults aged 16 to 65 at Level 1 literacy or below, against 19 percent in 2017.
The mode decision fixes this rather than measuring it. Alchemic screens inside the channel the respondent picks: a web link, a WhatsApp thread with no link and no app, or an outbound AI phone call. It publishes 57+ languages including Spanish, Arabic and Mandarin, and fields across 14 markets including the USA and the UK. Where a study needs people a link cannot reach, modes that work without a smartphone settle it before wording does.
Where a Screener Cannot Fix the Sample
A screener qualifies individuals. It does not build a sampling frame, and no wording compensates for a frame that never held the population.
Pew's conclusion on its three-method test is that there is no way to know what share of bogus respondents a method catches, nor how many genuine ones it misidentifies.
Finally, the case against elaborate screening. For a high-incidence consumer category, a five-item screener on a self-serve panel platform is the right tool and managed fieldwork is overhead. The apparatus here earns its cost when incidence is low, cells are narrow, the population is hard to reach, or the incentive draws people who want in.