In preparing for a session of the Power of Market Research class that I teach at UCSC Extension Silicon Valley, I was brushing up my ‘spiel’ about how to determine sample size for research studies (surveys, focus groups, etc.)…and realize that this is a mysterious concept that could use some plain talk.
Sample size is a very important element of any primary research. Very rarely can you get input from everyone within your target group, so you need to figure out a way to ensure that you have a ‘sample’ of the group that is big enough and representative enough of the entire group to provide ‘reliable’ results—in other words, how many people do you need in order to predict how the entire group you are researching thinks/feels?
When determining sample size, you want to think about the reliability of your consolidated results (the input of all respondents/participants) as well as the reliability of the data you receive for any groups/segments you are analyzing. For example, if you are going to segment your results by household income or annual revenue, you need to make sure that you get enough respondents in each category of income or revenue to provide reliable results. This will mean that you need many more respondents than required to provide reliable results for the consolidated group as a whole.
Click here for a tool to help determine the sample size you need to get reliable results, based on the size of your target group and the confidence interval you are aiming for. Tip: I suggest using a 95% +/- 10% confidence level as a good goal for how reliable you want your data to be. Using that approach, for any target group consisting of 2000 or more people, you would need about 90 respondents.
Hopefully this info will help you increase the validity of your data and, more importantly, your own confidence in the data you gather.