Contributions to Adatpive Web Sampling Designs

Investigation of the characteristics and estimation of quantities of hidden and hard-to-access population are of interest to scientists. Such populations are difficult to target because of their elusive nature or other prohibitive characteristics. So crafting designs of a representative sample and creating estimation methods to obtain efficient information from the sampled data are core challenges for people who investigatesampling theories and applications. Thompson (2006a) proposed an adaptive web sampling ( AWS ) scheme which takes into account the social networks between subjects to get a more efficient sample, and the procedures are more flexible than adaptive sampling. Three papers which contribute to AWS are included in this thesis. They focus on new resampling mythologies to improve the inferential estimation, produce designs with practical restrictions to minimize the cost and maximize the sampling simultaneously andmodel based estimation for non-responses. Simulated and real data sets are used to demonstrate implementation.