How many surveys should I collect?

We recommend that organisations should aim to collect 380 surveys. This number will give results at around the 5% margin of error at the 95% confidence level.

We understand that there is a trade-off between level of accuracy you want to reach and feasibility of obtaining a large sample size within the budget and resources available. However, collecting at least 380 responses will give results which, if sampled well, can be relied upon to represent your audiences. This is generally true of most organisations.

If you are collecting survey data for Arts Council England, please refer to Illuminate’s survey sampling guidance (opens as a Word .doc file).

What does margin of error and confidence level mean?

The margin of error and confidence level are two measures that affect the accuracy of your data.

Margin of error (or confidence intervals)

In short, this is the deviation between the opinions of your respondents and the opinion your entire audience population. For example, you set your margin of error on 5%. If 70% of your survey respondents rate your new exhibitions as ‘Very good’, a 5% margin of error means that you can be ‘sure’ that between 65% (70%-5) and 75% (70%+5) of your audience population thinks your new exhibition is very good.

Confidence level

This tells you how often the percentage of the population that thinks your new exhibition is very good, lies within the boundaries of the margin of error. Or, following on our previous example, it tells you how sure you can be that between 70% and 75% of the population thinks your new exhibition is very good. Suppose you chose the 95% confidence level – which is pretty much the standard in quantitative research – then 95% of the time between 70% and 75% of the population thinks your new exhibition is very good.

Sample size required by audience size to give a confidence level of 95% and a margin of error of 5%

Audience Size	Sample Required
100	80
500	217
1,000	278
10,000	370
100,000	383
500,000	384
1,000,000	384