Abstract
Abstract :
The confidence intervals (CT’s) conventionally constructed in large‐scale sample surveys assuming asymptotic normality often leads to unsatisfactory results when the population under study is rare or clustered. Adaptive cluster sampling is a promising sampling technique to effectively catch rare, geographically clustered or localized population elements. Christman and Pontius (2000) applied several bootstrap techniques to construct confidence intervals when simple random samples are selected without replacement (SRSWOR) and adaptive cluster sampling is used to sample localized population units. Here we extend Sitter's (1992a, b) ‘mirror‐match’ (MM) bootstrap to a practical survey set‐up using varying selection probability. We also demonstrate using real data from the Indian Economic Census how the extended procedure can be applied to adaptive cluster sampling adopted for estimating simultaneously the numbers of carners engaged in a number of localized unorganized rural industries.
Get full access to this article
View all access options for this article.