-62- 



Stratified Random Sampling 



Often a forester divides the area into compartments or stands before 

 beginning the inventory. In other words he stratifies the area. 



If the forester's primary interest is in the estimates within each 

 stratum, then he can specify the allowable error for each stratum and apply 

 Equation 1 to estimate sample size. 



If his primary interest is in the mean volume per acre or total volume 

 over all strata, then he can estimate sample size by one of two techniques. 

 Each technique depends on how sample units are allocated to strata. The 

 two techniques will be illustrated with an example. 



First consider proportional allocation . By proportional allocation, 



the forester determines sample size and allocates sample units to strata in 



proportion to the acreage of each stratum. The equation for sample size is 



t 2 L 

 n = [-^] I W.CV7 Equation 2 



1=1 



where t is the t-value, E is the allowable error, W. is the proportion of 

 acreage in stratum i, CV. is the coefficient of variation in stratum i, 

 and L is the total number of strata. 



Assume a forester has the following information on an 80-acre tract he 

 wishes to sample with a 10-B.A.F. prism. 



Suppose he wants to estimate the mean volume per acre within 10 percent 



with 95 percent confidence. Then, by applying equation 2, he estimates the 



number of points required. 



2 L 2 



n = [-^] Z W^CV^ = [j|-] (3047.50) = 122 points 



He would allocate points to strata as follows: 



