> For the complete documentation index, see [llms.txt](https://docs.askyazi.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.askyazi.com/build-a-study/screening-and-incidence.md).

# Screening and incidence

Design screeners that let the right people in, estimate incidence before launch, and know what to do when too few people qualify.

A screener is the set of questions at the start of a study that decides who qualifies. Incidence is the share of people who start the screener and qualify. Getting both right matters more than any platform setting: the worst fieldwork outcomes we see come from screener design, not from technical faults. This page is guidance on designing screeners in Yazi.

{% hint style="info" %}
**In short:** put screening questions first and screen out with logic, ask only what you need, check that each criterion measures what you actually mean, and estimate incidence before launch by multiplying the share who pass each screen. Watch the funnel on the first day and be ready to loosen a screen that is collapsing incidence.
{% endhint %}

## How screening works in Yazi

* Screening questions are ordinary questions, usually Single Select, placed at the start of the activity.
* A routing rule on each one sends people who do not qualify to **Screen out**. They get your screen-out message and leave the study. See [Logic and routing](https://docs.askyazi.com/build-a-study/logic-and-routing.md).
* **Quotas** can screen out people in groups that are already full. See [Quotas](https://docs.askyazi.com/build-a-study/quotas.md).
* Screening works the same in in-chat surveys and WhatsApp Flows.
* People who are screened out are not counted as completes and do not receive reminders.

Set the screen-out message in **Study configuration > Privacy & screening > Screen-Out Message**. It is translated into each participant's language, and you can add a link button (for example back to a panel) or a quick-reply button that starts another study.

## Profile fields versus screening questions

Some details can come from the people you invite rather than from questions:

| Detail                                             | Where it can come from                                                       |
| -------------------------------------------------- | ---------------------------------------------------------------------------- |
| Age, gender, region, city, language                | Your uploaded contact list (People fields) or the panel provider's targeting |
| Income, purchase behaviour, product usage, recency | Must be asked as a question in the study                                     |

If you rely on a contact list or panel targeting for a detail, you do not need to spend a question on it. If you want to route on it or set a quota on it inside the study, ask it as a question. When in doubt, ask: it costs one tap and gives you an answer from the participant themselves.

## Design a screener

### Order

1. **Consent first**, if your study needs it. Route a "No" to **Screen out**, not to Complete.
2. **Hard qualifiers next**: the criteria that exclude most people, so they leave quickly.
3. **Quota questions** (gender, age band, region) once you know the person qualifies.
4. **Then the survey.**

Tell participants up front what the study is about and roughly how long it takes. A first question with no context can look like a broken link.

### Ask what you mean, not a proxy

Read your screens in order and ask of each one: "Is this measuring what I actually care about, or something that stands in for it?"

A worked example: a study wanted people living in townships. The screener asked for province, and excluded several provinces entirely. Province was a proxy for area type, and a poor one: it excluded many qualifying people and let in many who did not qualify. A direct question about the type of area the person lives in would have screened correctly.

### Watch the stacking effect

Every screen multiplies. If 40% of people pass a recency screen, 30% pass an income screen and 50% pass a region screen, only 0.4 × 0.3 × 0.5 = 6% qualify, roughly 1 in 17 starters. Three reasonable-looking screens can produce very low incidence together.

### Keep it short

A screener sits in front of everything else, and people have not yet committed to your study. Keep it to the questions that decide eligibility. Move "nice to know" profiling questions to the end of the survey.

## Estimate incidence before launch

1. Write down each screening criterion.
2. Estimate the share of your invited audience that passes each one, from published data, previous studies or the panel provider's feasibility figures.
3. Multiply them together. That is your expected incidence.
4. Divide your target completes by the incidence, then by the expected completion rate among qualifiers. That is how many people need to start.

For example, 300 completes at 6% incidence and 80% completion needs about 300 ÷ 0.06 ÷ 0.8 = 6,250 starters. If that is more than your list or budget allows, loosen a screen before launch rather than during it.

For paid panels, the provider's audience targeting does part of the screening before people reach you, which raises incidence inside the study. See [Panel recruitment](https://docs.askyazi.com/reach-participants/panel-recruitment.md).

## Monitor incidence during fieldwork

On the first day, open **Share > Participants** and compare how many people started with how many were screened out and how many are still in the survey. Yazi does not alert you automatically when incidence drops, so check it yourself.

<figure><img src="https://358667285-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FzMRQUXiGQlBwWdg6WEmG%2Fuploads%2F6dDLI6GOSKUKNqYv7EAu%2Fshare-tab-participants.webp?alt=media&amp;token=c703715c-b6aa-4ad3-844d-5561db895290" alt="The Share tab Participants view: the funnel from sent to completed, status filters and the participant table"><figcaption></figcaption></figure>

If almost nobody qualifies:

1. **Find the screen doing the damage.** Look at answers to each screening question in **Reporting > Graph Data** (with test data hidden) to see where most people are screened out.
2. **Ask whether that screen is a proxy** for something else, or stricter than the research needs.
3. **Pause the study**, change the screen (or widen the quota), test, and resume. People already screened out are not brought back automatically.
4. **Tell whoever is recruiting**, so invite volumes can be adjusted.

## Limits and things to know

* **No automatic incidence alert.** Monitor the funnel yourself, especially on day one.
* **Screen-outs cannot be undone for a participant.** After you loosen a screen, earlier screen-outs stay screened out. Re-invite them if appropriate.
* **Answers do not carry between activities or studies.** A screener in one activity cannot route a later activity. Split groups at invite stage instead.
* **Editing a live screener** affects only people who reach it afterwards. Pause first.
* **Incentive and length affect who stays.** A long survey with no incentive will not hold people through a screener.

## Related

* [Logic and routing](https://docs.askyazi.com/build-a-study/logic-and-routing.md)
* [Quotas](https://docs.askyazi.com/build-a-study/quotas.md)
* [People and contacts](https://docs.askyazi.com/reach-participants/people-and-contacts.md)
* [Panel recruitment](https://docs.askyazi.com/reach-participants/panel-recruitment.md)
* [Participants tab](https://docs.askyazi.com/monitor-fieldwork/participants-tab.md)
