How many completes are required: a beginner guide to estimating the number of completes needed for a statistically significant survey

Although some surveys are based on “hey, let’s get 1000 completes”, there are sound statistical rules to calculate the number of completes that a study needs. These are called the “Confidence Interval” and the “Margin of Error”. These need to be valid for all the segments you plan to make from your survey data.

The number of completes you need is derived from a formula:

N = (0.25/s) x (z/MOE)^2

where N is the number of completes, s is the smallest share of the panelists that you want to segment, z is a z-score which is found from a lookup table of the required confidence interval, and the MOE is the margin of error that you are OK with.

This is detailed in the white paper, with examples, lookup tables and reference charts. There is also a downloadable Excel Calculator to do the math for you if you just need the answer.

Download white paper: Calculating Completes for a Survey (pdf)

Download Excel calculator: Completes and MOE Calculator (xlsx)