 |
Non response is a nuisance not only because
the missing responses but because the sample size gets smaller.
You can not say " I surveyed 500 subjects and of the 400 who responded
this question 220 say yes and 180 say no, so the majority say yes". because
of the 100 non responders. If they all would have said no, then the result
would have been 220 for yes and 280 for no. Also in computing the standard
error of the percentage the n is 400 and not 500, it increased by 12 %
because of the non response.
|
