-37- 



SECTION 4. STATISTICAL ANALYSIS 



Overview 



The purpose of INVENT 's statistical analysis is to indicate the reli- 

 ability of the volume estimates. This is accomplished by computing a 

 confidence interval for all volume estimates that are summed across the 

 diameter classes. This confidence interval is specified as a percent of 

 the respective volume estimate. The user specifies the confidence level 

 as a decimal in the range .65 to .99. All statistical analysis assumes a 

 stratified random sample where each sample unit is a cluster. There is an 

 option to double sample for height within the clusters. 



INVENT calculates confidence intervals for sawlog cubic foot volume, 

 sawlog board foot volume, and pulpwood cubic foot volume. The analysis 

 for each volume is carried out independently of the other volumes. Since 

 the method is exactly the same for each volume, the calculation of error 

 for an arbitrary volume V is presented. Throughout the discussion we will 

 assume the reader is familiar with the techniques for calculating the mean 

 volume per acre when the heights of all volume trees are measured. (Cf. 

 Barrett and Nutt, 1975; Avery, 1975; Husch, Miller, and Beers, 1972; 

 Dillworth and Bell, 1976.) 



The statistical analysis is presented in the same sequence as the 

 program performs it. For each individual stratum the standard error of the 

 volume must be calculated. If the height subsampling option is used, the 

 program first calculates the standard error due to height estimation. Next 

 the percent cruise of the stratum is calculated. This is used to refine 

 the standard error due to the variation between clusters which is calculated 

 next. This standard error is then combined with the height estimation 

 standard error to give the volume standard error of the stratum. Finally, 

 the volume standard errors of all the strata are combined to yield the 

 standard error of the stratified volume estimate along with the effective 

 degrees of freedom of that estimate. 



