Components of "Total Errors" in traditional surveys
Since the total error in a survey is the result of two different types of error , sampling errors and non sampling errors, when a web survey is carried out on a non probabilistic sample the estimation error cannot be calculated within traditional estimation theory. Furthermore it should be noted that in web surveys there are coverage (especially under coverage) and no response errors which have a certain weight as not all the target population have web access and so cannot be part of the frame. Another point to consider is that the people who have web access could be significantly different from other units. This fact could further bias the results. Coverage errors cannot be estimated using statistical inference theory but further in-depth study is extremely important as they are directly linked to the quality of the survey.
More:
This article has been read 837 times
