The simple answer is – it depends on how reliable your data needs to be. But you rarely need more than 400 interviews.
Sample Size Calculator
|How many interviews?||How robust the data?|
You will need to know what the total population of your survey is – how many customers, how many businesses or how many residents in an area.
The confidence interval or margin of error, is the plus-or-minus figure usually reported in newspapers or on television. Suppose 41% percent of your sample gives a particular answer. If you have a confidence interval of 5, you can be certain that between 36% (41-5) and 46% (41+5) of your total population would have given the same answer.
The confidence level is a percentage measure of how sure you can be that your findings are accurate. If your sample size conforms to the 95% confidence level, then 95 people in every 100 in your total population would answer the same way. Most of our clients use 95%. 97.5% gives greater accuracy but can require a disproportionately larger sample.
When people are virtually unanimous, you can be sure of the answer, whatever the sample size. When people are split more or less 50/50, the consequences of error may be much greater. This is why market researchers assume the worst case, a 50/50 split, when calculating the sample size needed.