ENBIS-17 in Naples

9 – 14 September 2017; Naples (Italy) Abstract submission: 21 November 2016 – 10 May 2017

Sensitivity in Sample Size Reduction with Respect to Prior Information

11 September 2017, 16:00 – 16:20

Abstract

Submitted by
Jens Bischoff
Authors
Jens Bischoff (University of Würzburg), Rainer Göb (University of Würzburg)
Abstract
Small sample size are indispensable in industrial sampling. We investigate minimum sample sizes relative to prescribed levels of confidence intervals for a proportion nonconforming p, e. g., the proportion of erroneous financial account entries. We consider the shortest two-sided confidence intervals under prior information suggested by Göb and Lurz (2014). The confidence intervals are used for a testing for a prescribed tolerable upper limit for p. We compare different ways of expressing prior information in a sensitivity analysis. It is shown that the required sample size can be reduced considerably below the sample sizes required under the customary Clopper & Pearson intervals. As a rule of thumb, the use of prior information leads to reductions of 20 % in sample sizes.
View paper

Return to programme