Here is a question about stats, which is a bit out of my league, can you help?
I live in a town called Ballarat, with a population of about 88,000 people in south eastern Australia.
Recently a group I am involved with did a telephone survey in the Ballarat area. From my (limited) stats knowledge I am comfortable that the sample was sufficient in size and that the recipients were selected at random.
Here’s the problem though, of the 500 or so people called about 150 agreed to complete the survey. So while the telephone numbers were selected randomly, I am not sure that we ended up with a random sample, because the people who agreed to complete the survey might have a different attitude/motivation to those who didn’t.
I thought about comparing the demographic info (age, gender, education level, suburb etc.) from the survey to the data for the same area that is available from the Australia government census, to see if the survey was “representative”.
So here’s my questions:
1. Am I right in questioning the methodology?
2. Is there some way to find out if the survey is representative of the population of Ballarat, given that it’s already been completed?
3. Are there techniques that could be used in designing such a survey to ensure it is representative? Any pointers to further info would be welcome!
Please leave a comment if you can help.
{ 0 comments }
