If I use a random sample, it will mean a 15% response rate for around 400 responses.

If I use quota sampling using an existing online panel that is representative of the UK population by age, gender and region (so quota sampling and not a random sample), I can get 1,000 responses.

Which one would be better?

  • $\begingroup$ Please provide more context. Better is subjective. What are you trying to show? Higher n means you have more power (is 400 how much you'll get or would you get $.15(400)=60$). Random sampling means you are, in theory, controlling for unobserved variables. However, you have missing data. Can you assume the data is missing at random? The answer will balance these and other concerns. $\endgroup$ – le_andrew Apr 16 '15 at 14:14
  • $\begingroup$ Thanks @le_andrew It would be 400 responses returned. I'm not sure if I can assume the data is missing at random. Its a general survey with all of the public in the UK to gauge their views about a particular topic. Based on other published studies, I am estimating the response rate to be around 15%. However, what I'm unsure is about whether I will receive enough responses from each age group, as I expect there to be a correlation between age and some of the responses. $\endgroup$ – CakeNRun Apr 16 '15 at 14:25
  • $\begingroup$ If your response rate is $< 1.0$ (i.e., if there are any non-responders) you can pretty much rest assumed that the sample is not random. A response proportion of 0.15 typically means the sample is not reliable for most purposes. $\endgroup$ – Frank Harrell Apr 16 '15 at 15:02
  • $\begingroup$ Hi @frankHarrell, thanks for your response. So does this mean the other quota sample would be better in this instance? $\endgroup$ – CakeNRun Apr 16 '15 at 15:10
  • $\begingroup$ If I understand the setup, yes. $\endgroup$ – Frank Harrell Apr 16 '15 at 15:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.