Weighted Regression in Successive Forest Inventories
The main objective of this study was to conduct a rigorous analysis of available data to test the validity of updating timber volume estimates by means of least-squares regression. The data were tested to determine if the updating model developed by Ware (1960) violates any of the assumptions of least-squares regression; namely, independence, normality, and homogeneous variance of regression residuals, as well as freedom from error in the independant variables. Analysis indicated that all assumptions were justified except homogeneity of the variance. It was found that weighted regression--employing a function of the expected current volume as a weight--stabilized the variance. In general, unweighted regression underestimated the variance of a volume predicted for a date later than the year of remeasurement. The results also suggest widespread applicability of the weighting procedure for estimating variances efficiently.
No Supplementary Data
No Article Media
Document Type: Journal Article
Affiliations: Mathematical Statistician at the Northeastern Forest Expt. Sta., Forest Service, U.S. Dept Agric., Upper Darby, Pa.
Publication date: 01 December 1966