ABSTRACT
The strategy of bipartite incidence graph sampling (BIGS) and incidence weighting estimator (IWE) is applicable to many unconventional sampling methods, which require some rules of observation in addition to an initial sample. A theorem is established for the applicability of the strategy to arbitrary graph sampling based on incident observation procedures. Discussions of network sampling, line-intercept sampling and sampling from relational databases clarify the common structure underlying these seemingly unrelated problems, which consists of the distinction between sampling and study units and the many-to-many observation links between them. Moreover, general approaches are introduced for dealing with situations where the many-to-many observation links cannot be fully observed in a given sample graph.
