An Incident of Incidence Part 1: Definitions
Incidence is one of the most critical pricing factors in the online market research world. In this article, we focus on understanding the concept of incidence and what the difference is between conversion, incidence, and qualification rate.
Let us first look at what incidence rate is and how it is being understood by different individuals:
According to the dictionary, in general, incidence is defined as the rate at which events happen.
In market research, incidence is a measure for the rate of occurrence, or the percentage of persons eligible to participate in a study. For example, in a study of US automobile owners, if the eligibility for research purposes is defined as owning an automobile, then the incidence rate (IR) would be the percentage of the US who own an automobile, out of the total number of the population (meaning out of all Americans, whether they own an automobile or not). The formula would be the total number of qualified respondents divided by the total number of respondents who were screened for the study (qualified plus non-qualified).
For a researcher, incidence is given by market penetration, brand awareness, and so on, representing the percentage of a population that has or does something in common or behaves in a certain way.
For a sample provider, the definition is similar to the researcher’s but focused on the operational side: how many respondents do you need to interview in order to get the required number of respondents to complete the survey? (i.e. percent of respondents that have or do something in common out of total respondents taking the survey).
Ideally, the conversion, incidence, and qualification rates would all be the same. However, in practice, many surveys use multiple sample suppliers with various targeting capabilities and entry points to surveys, making quotas necessary. Hence, overlap of respondents between sources is almost guaranteed.
As such, a sample supplier looks at incidence more along the lines of conversion where the supplier needs to account for deduplication between his panel and other sources, needs to account for quotas that fill in various demographics faster than others, plus account for respondent fatigue that causes respondents drop mid survey. As such, the supplier’s incidence (a.k.a. conversion) will always be lower than the survey incidence which would only account for finished interviews that either qualified or disqualified (i.e. completes and terminates).
A researcher looking to understand market penetration would need to allow all respondents to go through the entire screener and count quota full respondents as qualified respondents to get an accurate penetration (because those respondents would have completed the survey if quotas weren’t closed). In this case, the incidence takes on the meaning of qualification rate. A researcher would not want to consider deduplication and would not want to account for errors in targeting, as these do not fall into the definition.
On top of it all, variations based on sampling universe (the incidence rate of Carlsberg beer drinkers is different when targeting beer drinkers versus alcohol drinkers versus general population) add another layer of complexity.
In any study, you would have
- – Completes
- – Targetable & Non-Targetable Terminates
- – Quota Full
- – Partials (respondents who abandon)
- – Security Terminates
Targetable terminates are those terms that occur on variables that information is outdated (i.e. on income, we might know the respondent has a specific salary, but that respondent may have changed jobs/lost their job and until the information is updated, the respondent can be invited to surveys incorrectly).
Security terminates include terminates based on any checks the main survey has (i.e. GeoIP, RelevantID checks, cookie duplication checks, trap questions fails, etc.)
Non-targetable quota full respondents are those respondents that would terminate if the study were fielding strictly to the specific quota where they disqualified (ex: purchase intenders vs non-intenders of a specific product)
- – Conversion is Completes / All clicks.
- – Incidence is Completes / (Completes + Non-targetable Terminates + Non-targetable Quota full)
- – Qualification Rate is (Completes + Quota Full) / (Completes + Quota Full + non-Targetable Terminates)
Clearly understanding incidence and how all parties interpret it is key in understanding how feasibility & pricing are impacted. In Part 2 of our Incidence series, we will talk about pricing in relation to incidence.