Web Survey Bibliography
There has been much debate over the growing use of internet panels in survey research. Most panels are not probability based and may produce a sample that is not representative of the population. Weighting by demographic characteristics can account for some diff erences between the panel and the population, but bias may remain. Th e propensity scoring method tries to deal with bias by using the likelihood of response to a telephone survey as compared to an internet survey to produce representative results. A recent internet survey on drug use and abuse included questions similar to those of the National Household Survey on Drug Use and Health (NSDUH). Using these questions and the propensity scores in the weighting process, we hoped to produce estimates that would compare to this dwelling unit survey. We found that the weighted internet survey produced very similar results to NSDUH.
Conference homepage (abstract)