Quote:
Originally Posted by HarryT
My point is that by preferentially not interviewing people who are unlikely to have a reader, you're artificially inflating the proportion of people that your survey finds do have one.
It would be like conducting a survey on how many people use email - by sending out email questionnaires.
|
Possibly, but as I said, I believe this is regarded as a credible group, and I've been told there are ways to correct for that.
(For example, that 40% may have been a raw 80 percent but they knew to adjust it based on the population not sampled, or based on data from another door-to-door survey...don't know, just speculating. Is there a survey expert in the house?)
ApK