Sampling with Decomposed Variance

by Gregory B. Baecher, (M.ASCE), Assoc. Prof. of Civ. Engrg.; Massachusetts Inst. of Tech., Cambridge, Mass.,

Serial Information: Journal of the Engineering Mechanics Division, 1982, Vol. 108, Issue 5, Pg. 1013-1017

Document Type: Journal Paper


In many problems of statistical sampling in civil engineering a prior distribution reflecting some global population is available and a small sample of observations is made on a particular subset of that global population. From this combined sample one estimates the behavior of the particular subset. Despite the familiarity of the problem, it presents statistical difficulties because the subset specific data are masked by the more numerous global data. These difficulties may be overcome by decomposing the data variance into and among subset parts and within a subset part. Common Bayesian sampling techniques are modified to achieve this goal.

Subject Headings: Statistics | Bayesian analysis

Services: Buy this book/Buy this article


Return to search