Incidence rate (IR) in survey research

In survey research, the incidence rate (IR) is the share of people who pass your screening questions and qualify for the survey: qualified respondents divided by everyone screened. It tells you how many people you need to reach to get your target number of completes, so it drives fieldwork time and panel cost.

Panel companies also call it the qualification rate or simply "incidence". In epidemiology the same words mean something else: the number of new cases of a condition per population at risk over a period. This page covers the market research meaning.

How to calculate the incidence rate

Incidence rate = qualified respondents / respondents screened × 100%

"Screened" means everyone who answered the screening questions, both those who qualified and those who were screened out. It does not include people who were invited and never opened the survey; that is the response rate, a separate number.

Worked example

You need 400 completed interviews with people who bought a robot vacuum in the last 12 months. In a soft launch, 1,800 people answer the screener and 486 qualify.

• IR = 486 / 1,800 = 27%.
• Screenings needed for 400 completes = 400 / 0.27 ≈ 1,482.
• If 10% of qualified people drop out before the end, you need 400 / (0.27 × 0.90) ≈ 1,647 screenings.

The incidence rate calculator does the first two steps: enter how many people took the screener and how many qualified, and it returns the IR and the screenings needed for your target.

How to read the result

A lower IR means more people have to be reached and screened for each complete, so fieldwork takes longer and costs more. That is why panels price low-incidence studies higher per complete. Treat the IR as a planning number with uncertainty: an estimate from 100 screenings can move a lot once you have 2,000.

The IR also checks your assumptions about the market. If you expected about 30% of online adults to own a product and only 8% qualify, either the criteria are stricter than you meant, the screener wording is unclear, or the market is smaller than you thought.

Where an IR estimate comes from

• A soft launch. Send the survey to a small part of the sample first and measure the real IR before the full launch. This is the most reliable source.
• Earlier studies with the same criteria and the same source of respondents.
• Secondary data such as official statistics or industry reports, multiplied for each criterion. Two independent criteria of 40% and 50% give about 20%, but criteria are often correlated, so check the result in a soft launch.

Common mistakes

• Counting over-quota people as not qualified. Someone who qualifies but lands in a full quota cell is "over quota", not a screenout. Mixing the two makes the IR look lower than it is.
• Dividing by invitations. Use the people who answered the screener as the base, otherwise the IR absorbs the response rate.
• Screeners that reveal the target. "Do you own a robot vacuum? (yes/no)" as the first question invites guessing. Ask about several categories in a multiple-choice list so the qualifying answer is not obvious. A leaky screener inflates the IR and lets the wrong people in.
• Criteria nobody asked for. Every extra filter multiplies down the IR. Keep only criteria that change the decision.
• Not adjusting during fieldwork. If the real IR is half the estimate, the timeline and budget double. Recalculate after the soft launch.

Measuring incidence in your own survey

1. Put the screening questions at the very start.
2. Use logic to end the survey for people who do not qualify. In SurveyNinja, logic (jumps, show and hide, ending the survey, screenout) and quotas work on every plan, including Free.
3. Set quotas for the groups you need so that full cells close on their own.
4. After a soft launch, count how many people answered the screener and how many passed, and put both numbers into the calculator.
5. Use the result to plan the rest: the panel cost calculator estimates the budget, and the invites needed calculator covers response rate on top.

To size the final sample itself, see sample size and quota sampling. For sources of respondents, read where to find survey respondents.