Why am I not seeing enough data in my segment or filter?

Displaying results from small subgroups

In order to maintain a representative sample, Zencity-Managed survey results need to have large enough numbers of respondents to support making defensible decisions. In some cases, we suppress results from small demographic groups because there’s not enough data for a subgroup. This is called cell suppression.

Cell suppression applies when results are filtered down to a subgroup, such as a demographic group or a geographic area. Filtered results remain representative (we never exclude groups from the results because of small sample sizes) but sometimes we don’t display small groups that have too few responses to support reliable conclusions. 

Call suppression

When a filtered view contains fewer than 30 responses, we suppress the result rather than display a number that could mislead. These responses remain in the survey results, but we prevent users from filtering down to the results of just that subgroup. If this happens, you’ll see a message like this:

When a filtered view contains fewer than 50 responses, but has 30 or more, we add a warning to note that the small number of responses makes the results about this subgroup less credible. You will see this message:

This threshold follows common statistical practice. For example, the CDC’s National Center for Health Statistics also requires a minimum sample size of 30 before it reports a proportion.

Why do we do this?

Representativeness and precision are two separate characteristics of the survey data. Representativeness comes from how we build the sample: our sampling is designed to reach the whole community, so results reflect the perceptions of the community as a whole. Precision comes from how many people respond. 

A large sample drawn from only one part of the community is still not representative, and a representative sample with too few responses is still not precise.

Zencity displays data when the data are both precise enough to provide defensible conclusions and when the results are representative of the communities we work with.

Learn more about representativeness here.