Notice: the WebSM website has not been updated since the beginning of 2018.

Web Survey Bibliography

Title Adjustment of Web Panel Survey Estimates by Regression Imputation
Year 2005
Access date 27.10.2005
Abstract When using a panel of volunteer respondents in a Web survey one can suspect people in the observed sample to have different characteristics and attitudes than the general public. Thus some adjustment method is needed to make inference about the public. We have developed an adjustment method based on the idea of regression imputation, a method often used to adjust for nonresponse. To make regression imputation possible, some auxiliary variable correlated to the study variable has to be observed in the Web sample and in a parallel survey with a probability sample. The Web sample is then used to estimate the relationship between the study variable and auxiliary variables. The predicted value on the study variable is imputed into the probability sample, which is used to estimate the population mean.
 
In an ideal situation where all the model assumptions hold, our adjustment method gives an unbiased estimator of the mean in the population. Its variance is derived along with an estimator of the variance. The suggested estimator might be sensitive to model violations and therefore simulations were carried out to study the behavior of the estimator under different model violations. The results were compared with an estimator based on post stratification. The estimator based on regression imputation turned out to be sensitive to non linear relationships and missing relevant auxiliary variables, while the estimator based on post stratification was more robust to such model violations.
Year of publication2005
Bibliographic typeConferences, workshops, tutorials, presentations
Print

Web survey bibliography - 2005 (76)

Page:
  • 1
  • 2
Page:
  • 1
  • 2