Web Survey Bibliography
There are at least two reasons why we are discussing inference from non-probability samples:
1. The continued predominance of non-probability samples in certain fields, certain countries, and certain domains;
2. Declining response rates having two effects: a) calling into question the extent to which we can use classical randomisation-based inference methods, even after standard non-response adjustments, and b) increasing the cost differential between probability and non-probability samples and thereby causing us to reassess the justification for probability sampling.....