Help & Guidance
Help & Guidance
Step 1: What do I want to measure?
Step 2: Selecting a ‘Geographical Area of Interest’
Step 3: Creating questionnaire(s)
Step 4: Collecting questionnaires
Step 5: Cleaning the data
Step 6: Analysing the data
Step 7: Monitoring information
Step 8: Completing the calculators
Interpreting the report
Frequently asked questions
Impact estimates for the cultural sector
Click the button below to get started
How do I achieve a representative sample?
How do I achieve a representative sample?
Ideally, all participants will complete questionnaires. However, this is unrealistic in most cases, especially where there are a large number of participants. The good news is that by speaking to a large enough proportion of participants, we can make statistically sound assumptions about the audience as a whole. The quantity section below outlines how many respondents you will need to complete your questionnaire in order to ensure reliability in your results.
However, It's not just about the number of people that complete the questionnaire - it's also extremely important to ensure that you make every effort to collect questionnaire responses from the full range of people that have taken part in a project. In other words, you need to achieve a 'representative sample'. The best way to do this is to select questionnaire respondents at random, which is outlined below in the quality section.
If you can’t get everyone to complete questionnaires, how many people need to complete questionnaires for you to be able to say with confidence that your questionnaire results are reliable? The table below should help answer this question. The ‘Sample size’ column on the left gives a range of different sample sizes (i.e. how many people you speak to) and the ‘Margin of error’ column shows how reliable your questionnaire results will be if you achieve the sample size.
The smaller the margin of error, the more accurate your results can be said to be. So for example, if you manage to get 96 people to complete questionnaires and 50% of questionnaire respondents picked a particular answer, you can be ‘sure’ that between 40% and 60% of all participants would have picked that answer because there is a margin for error of plus or minus 10%. If 43 people completed questionnaires and 50% of respondents picked a particular answer then you can only be ‘sure’ that between 35% and 65% of all participants would have picked that answer: a wider margin for error.
As a general rule, researchers tend to aim for no more than a 10% margin of error (which means no less than 96 questionnaire respondents). Ideally, we would recommend aiming for a margin of error of ±5%.
Sample size (No. of respondents)
Margin of error (%)
 The margin for error figures have been calculated assuming a 50/50 split in the results, an undefined expectation of number of participants and are given at the 95% confidence level
As you might expect, events with fewer participants will have a smaller target sample size. However, this is not proportional – speaking to 5% of participants at a large event will give you much more accurate results than speaking to 5% of people at a small event. For example, if you are expecting 3,500 participants and you are aiming for a margin of error of ±5%, you would need to speak to 346 participants (or around one in ten participants); if you are expecting 100 participants you would need to speak to 80 of them (four out of five participants) to achieve the same margin of error.
There are many
freely available sample size calculators
on the internet which can help you set a target sample size.
Please note: The Economic Impact Calculator and Social Return on Investment (SROI) Calculator will tell you the eventual margin of error in your results so there’s no need to worry too much about achieving any of the exact sample sizes shown above: they are just intended to provide an indication of the sample size you should to try and aim for. If you receive more responses, the accuracy of your findings will only improve.
The accuracy of your questionnaire research is also determined by who is included in your survey. It’s important to get a full range of people to complete questionnaires so that your research is representative of the full range of people that took part in your project. There are various techniques that researchers use in order to achieve a ‘representative sample’.
For example, instead of standing in the same place at an event and handing out questionnaires, you could vary where you stand so that you capture the different sets of people who have chosen to visit specific areas of the event. Another technique (when surveying groups of people) is to select randomly from the group (to avoid only surveying the most assertive/outgoing people in any group). You could do this, for example, by always handing your questionnaire to the person in the group whose birthday is next.
At activities (that go on over a period of time), you could select people to complete the questionnaires at random from a list. So, for example, if you're aiming to get 89 activity participants to complete the questionnaire (out of 200 participants) then you could generate 89 numbers (between 1-200) using a random number generator and then pick your sample of 89 accordingly from the list or 'register' of all participants. For this to work, your list of participants should be numbered (in this example - from 1-200) so that every person has a unique number. There are a number of free random number generators on the web such as
This may all sound like hard work, but time spent planning your sample is certainly time well spent. Achieving a good, representative sample will mean that you will be able to show your results really are representative of the audience as a whole, and not just based on the experiences of a certain type of visitor. In a worst case scenario where you weren’t able to achieve a large sample, at least the data you do have will be honest and representative of the wider audience (albeit with a large margin of error).
Back to step 4
Go to step 5