A Test of Cluster Sampling in Forest Inventories

Publisher Website
Google Scholar
Abstract
During a recent pilot survey in Newfoundland, forest data were collected using a stratified, one-stage cluster-sampling design. The data were analyzed to determine if, within the stratified framework, cluster sampling would be more efficient than simple random sampling. Each cluster consisted of five point-samples located in a straight line. For the analysis, volume and variance estimates were determined from clusters of 1,2,3,4 and 5 points. During the survey, records were kept of the time required to complete various field-sampling tasks. These were used in lieu of cost data in the analysis. Results indicated that, for the given conditions, simple random sampling was more efficient than cluster sampling. However, relatively small changes in the conditions would make cluster sampling more efficient.

This publication has 0 references indexed in Scilit: