Unlike PoS data which often has near-census level coverage, survey data is based on a smaller subset (sample) of all buyers and sampling error can occur. As such, users should interpret data with this in mind.
How to spot insights which may be more indicative?
- View the number of ‘Interviews’ shown to the top left of charts (and by hovering over chart elements). Sample error is higher for insights based on fewer interviews
- View the level of fluctuation when in ‘trend view’, controlled by the line chart icon on the top right of most charts. More wave-on-wave fluctuation typically points to more sample error.
- Consult confidence intervals – see below
Confidence intervals
Where a more exact assessment is needed, refer to the 90% confidence interval table below. This shows, the max level of error on either side of the observed percentage for a given sample size:
As an example of how to read this, if the percentage observed is 75% for a sample of N=800 interviews, we can be 90% certain that the true number is +/- 3% of the observed 75%, i.e. within a range of 72% to 78%.