NCJ Number
72001
Journal
Operations Research Volume: 24 Issue: 2 Dated: (March/April 1976) Pages: 207-219
Date Published
1976
Length
13 pages
Annotation
Sampling methods that will allow jury commissioners to use multiple overlapping lists to produce a random sample from those lists for juror selection are presented.
Abstract
A population of potential jurors is composed of 'L' district individuals. Each individual is listed in at least one and possibly as many as 'k' different but overlapping frames. Regardless of any overlaps among the frames, the objective is to obtain an equal probability sample of size 'n' which is defined as any sample in which each individual has a probability of 'n/L' selection. Five plans for sampling at random from overlapping lists of names are analyzed for their probabilistic and cost properties. Under plan 1, the 'k' frames are merged into a single master frame with one entry for each of the 'L' individuals. A simple random element sample of size 'n' is then chosen, Under plan 2, a frame is selected at random where frame 'i' is selected with a specific probability. Under plan 3, names are selected at random from frame 'i' and are checked until 'nli/L' good names have been chosen. Under plan 4, names are selected at random from unflagged names on frame 'i' until 'pni' different names have been checked. Under plan 5, a name is chosen at random from the remaining entries on all frames. The remaining 'k-l' frames are searched for the presence of the chosen name. In terms of cost plan 2 appears the best choice if 'li' elements are not known; plan 3 is best if these elements are known. Results apply to sampling from overlapping frames in any context. Thirteen references are included.