Use of Figures 2 and 3 to determine sample sizes 



The use of the chart in Figure 2 can be illustrated by the following 

 example : 



From the junction of the 0,03 accuracy line and the 0=6 coefficient 

 of variation line in the lower right-hand graph, follow up to the 

 intersection of the 2000-tree curve, and thence across (in a left- 

 hand direction) to the right-hand axis of the alignment chart and 

 establish a point <, Now, starting on the left-hand graph, find the 

 intersection of the 0o8 correlation coefficient line and the four 

 cost ratio curve, and move across (in a right-hand direction) to 

 the left axis of the alignment chart and establish a second points 

 Connect the two points with a straight edge, and read 500 trees for 

 the sample size on the central axis of the alignment chart. The 

 formula in Appendix VIII shows the exact answer to be 527 trees. 

 The sampling rate would be one in four. 



The amount of check scaling needed in the example vidiich has been 

 described can be determined from Figure 3, Connect with a straight 

 edge the point for a correlation coefficient of 0,8 on the left- 

 hand axis of the alignment chart with the point for a cost ratio 

 of four on the right-hand axis, and read about a 0o4 ratio or factor 

 on the central axis. This factor, OA^ multiplied by 500 gives 200, 

 the number of check scaled trees required. The operation can be 

 checked by the formula in Appendix IX. 



As the correlation is improved by better tree measuring, the number 

 of trees needed in the first sample actually increases. But a look 

 at the check scale size required when correlation is high clearly 

 shovjs the advantage. As few as one-fifth of the measured trees need 

 1)6 check scaled vjhen the correlation coefficient is 0,9 or better. 

 Thus the big saving is made on the expensive part of the job, and 

 increased emphasis is placed on the more easily obtained first sample. 



Given : 



accuracy standard 

 coefficient of variation 

 sale size 



correlation coefficient 

 cost ratio 



0,03 

 0,6 

 2000 

 0,8 



4,0 



-24- 



