Web Survey Bibliography

Title Adjusting Responses in a Nonprobability Web Panel Survey by the Propensity Score-weighting
Year 2004
Access date 25.08.2004
Abstract As is often the case with nonprobability web surveys, responses differ from other traditional surveys. Difference in response to the same question in two different surveys may caused by differnces in survey modes, survey administrators, survey periods, design and/or layout of questionnaires, etc., and difference in sample. If two surveys differ in only sample, a survey result could be adjusted to another survey by weighting. The propensity score weighting is a useful technique when the number of covariates to be considered is large. To examine the availability and effectiveness of the propensity score weighting in practice, an experimental web survey was conducted on two different panels, and then tried to adjust responses in nonprobability panel to that probability one. The response variables were owner rates of 14 kinds of information equipments, and the covariates adopted in estimating the propensity weights were 10 demographic variables. The possibility of a general weight that could adjust responses in a nonprobability web survey and the effective usage of the propensity score weighting in web surverys were discussed.
Access/Direct link Homepage - conference (abstract)
Year of publication2004
Bibliographic typeConferences, workshops, tutorials, presentations
Print