Simple estimators for network sampling

04/03/2018

∙

by   Steve Thompson, et al.

∙

0

∙

share

Some conceptually simple estimators for network sampling are introduced. The new estimators are based on using fast sampling designs or processes on the network sample to estimate relative approximate inclusion probabilities of the actual sampling. We evaluate the effectiveness of the new estimators with simulations using a well-known empirical network data set as the exper- imental population. We find that the new estimators can bring substantial reductions in bias and mean square error compared to estimators in common use.