How many surveys do you need to have to be statistically significant?

Question

We are doing a project and we need to know how many surveys we need to have filled out in order to be statistically significant. Please help!

:]

Answer 1

Statistical significance isn't really the term you should be using (see below) a survey can be done very well but not return any 'statistically significant' results simply because there are none to return. The question is really, how many surveys do you need to do to be confident of getting meaningful results. Well, that depends on what total population size you're sampling (10 year olds in your neighbourhood or the entire human race) and how variable you expect whatever you're measuring to be (number of siblings or number of books in their house). You may not know that second thing until you've done your survey and if for example you only surveyed 20 people but got an answer of anything from 0 to 100 then you'd probably not feel too confident of your results. If the same sample size gave results ranging from 10 to 14 then you would feel much more confident.

Sorry I couldn't give a more helpful answer, but as a project I think you do what you can and then discuss those sorts of issues when writing it up.