Screening Questions

Where it is

Screening in the sidebar (/screening/).

Settings

• Enable participant screening — the master switch. Turning it off keeps the questions for later.
• Disqualification Message — what a screened-out participant is meant to see.
• Redirect URL (optional) — where to send them, e.g. a panel provider's screen-out URL.

Click Save Changes after editing settings.

Known issue (September 2026). The participant is currently shown a generic "Thank you for your time. Based on your responses, you don't meet the criteria for this study." — your custom Disqualification Message is not used.

Adding a question

Add Question opens a dialog:

• Question Text
• Question Type — Yes/No or Multiple Choice
• Options — for multiple choice, two or more
• Qualification Logic — Equals or Does not equal
• Qualification Value — the answer(s) that decide. For multiple choice this is a multi-select list (hold Ctrl/Cmd).
• Translations — pick a language and click Translate

Save with Save Question. Questions can be reordered by dragging, edited with the pencil, and deleted with the bin.

Note (September 2026). Translate works on a saved question only — on a new one you are asked to save it first, because a translation is stored against the question.

What participants see

Screening questions are asked as single-choice prompts with clickable chips, plus an automatic "Prefer not to answer". The sidebar shows progress (for example "Q2 / 2").

If an answer cannot be matched to an option the interviewer re-asks once; a second unmatched answer ends the session as disqualified. Keep options short and distinct.

How qualification is stored

Each question stores options and acceptable_options. acceptable_options lists the answers that qualify — everything else screens out. This is worth knowing because it is exactly what the API exposes, and it is the reason the "Does not equal" bug above matters.

With MCP

get_screening             {"study_id": "…"}
set_screening_settings    {"study_id": "…", "enabled": true,
                           "disqualification_message": "Thanks for your interest…",
                           "redirect_url": "https://app.prolific.com/submissions/complete?cc=…"}
create_screening_question {"study_id": "…",
                           "question": "Which service did you use most recently?",
                           "options": ["HelloFresh", "Blue Apron", "I have never used one"],
                           "acceptable_options": ["HelloFresh", "Blue Apron"]}
update_screening_question {"study_id": "…", "question_id": "…",
                           "acceptable_options": ["HelloFresh"]}
reorder_screening_questions {"study_id": "…", "question_ids": ["…", "…"]}
delete_screening_question   {"study_id": "…", "question_id": "…"}

The screen-out redirect_url takes the same placeholder as the completion one: {{external_participant_id}} is replaced with the ID this participant arrived with, so a panel is told who bounced as well as who finished — see tagging a link.

The API takes acceptable_options directly, so it is not affected by the dashboard's logic bug — and update_screening_question is a reliable way to repair a question saved with the wrong rule. acceptable_options must be a subset of options. Deleting a question renumbers the rest and deletes recorded answers to it.

Best practices

1. Three to five questions. Every extra one loses participants.
2. Screen on behaviour, not intention — "have you cancelled in the last 3 months" beats "would you ever cancel".
3. Avoid giveaway phrasing. If the qualifying answer is obvious, people will select it.
4. Test both paths with Try Interview — one run that qualifies, one that does not.
5. Set a redirect when using a paid panel, so screen-outs are returned correctly — with {{external_participant_id}} in it when the panel needs to know which person bounced.

Last updated: September 2026