Sampling errors for the six degrees of separation (Stats NZ WP 12-03)

Sampling errors for the six degrees of separation arose from a study in Statistics New Zealand of a large network, similar to the acquaintance graph. This network is of interest because of insight it may give us into knowledge diffusion between firms. The paper explains some statistical problems which came up in the course of the study, and their resolution. The solution to the statistical problems can be obtained from a standard result in sampling theory, but applied in a rather different context from the usual one.