Open Access Open Access  Restricted Access Subscription Access
Open Access Open Access Open Access  Restricted Access Restricted Access Subscription Access

Statistical Methods in Forest Products Research


     

   Subscribe/Renew Journal


Statistical methods can help in (1) sampling of test material (2) experimental design and (3) interpretation of results. The paper cites examples of the application of statistical methods in experimental design and analysis of data in research conducted in the Forest Products Laboratories at Madison, Princes Risborough, Melbourne and Dehra Dun. It has been found that to obtain greater precision for species averages of strength properties of timber, at a given cost, the test material should come from as large a number of trees as possible with consequent reduction in the number of specimens from each tree. The range is almost as good a measure of variation as the standard deviation in samples of 6 or less number of observations. Its use is therefore recommended for these sizes of samples and the method of indirectly estimating the standard deviation from the range is explained. Finally, the paper recommends the use, wherever possible, of factorial designs, confounding, fractional replications, incomplete blocks, etc., in designing experiments on laboratory and pilot plant scale.
Font Size

User
About The Author

K. R. Nair


Subscription Login to verify subscription
Notifications

Abstract Views: 263

PDF Views: 0




  • Statistical Methods in Forest Products Research

Abstract Views: 263  |  PDF Views: 0

Authors

Abstract


Statistical methods can help in (1) sampling of test material (2) experimental design and (3) interpretation of results. The paper cites examples of the application of statistical methods in experimental design and analysis of data in research conducted in the Forest Products Laboratories at Madison, Princes Risborough, Melbourne and Dehra Dun. It has been found that to obtain greater precision for species averages of strength properties of timber, at a given cost, the test material should come from as large a number of trees as possible with consequent reduction in the number of specimens from each tree. The range is almost as good a measure of variation as the standard deviation in samples of 6 or less number of observations. Its use is therefore recommended for these sizes of samples and the method of indirectly estimating the standard deviation from the range is explained. Finally, the paper recommends the use, wherever possible, of factorial designs, confounding, fractional replications, incomplete blocks, etc., in designing experiments on laboratory and pilot plant scale.