FOREST TAXATION IN THE UNITED STATES 543 
The process of assessment roughly outlined above is, no doubt, 
much the same process as that through which many conscientious 
assessors go at the present time, although usually with inadequate 
information. Competent and conscientious employment of such a 
method would generally result in a reasonably accurate assessment 
of real property. 
ASSESSMENT BY MULTIPLE CORRELATION AND GRAPHICAL CORRELATION 
The process of assessment could be further refined by the use of 
multiple correlation, provided there were sufficient sales and not too 
many variable factors affecting value to warrant the use of this 
statistical tool. The use of multiple correlation in valuing property 
may be described as the process of averaging values added by different 
factors to properties in a certain sample to determine the most 
probable values which would be added by these factors to any prop- 
erty in the district from which the sample was chosen. This method 
has been used to a limited extent by investigators, and it was applied 
in this study in an extensive appraisal of properties in the town of 
Loudon, N. H., as described in part 4, page 112. It would seldom, 
if ever, be practicable in view of actual assessment conditions. 
A more useful device is found in the graphical method of correlation, 
which has the advantage of not requiring any knowledge of higher 
mathematics. This graphical method is merely a method of checking 
and improving tentative estimates. Assume, for instance, that an 
assessor estimates from his general knowledge that crop land in his 
town is worth, on the average, $40 per acre. He applies this unit 
value, together with estimated unit values for other factors, to those 
properties for which there are bona fide sales. He subtracts from his 
estimated total value of each property the actual sale value of that 
property to obtain a “‘residual.’”’ He then plots this residual against 
the percentage of area in crop land in that property. After all 
residuals are thus plotted, he draws that line which his eye judges 
to be an average of the various points established on the graph. If 
this line has an upward slope as the percentage of area in crop land 
increases, he then knows that his $40 per acre estimate for crop land 
is too high, or, in other words, that the nearer the property is to being 
all crop land, the greater is the difference (the residual) between the 
$40 unit value and the actual unit value as determined by sales. He 
then lowers his crop land unit value, calculates new total values for 
the different properties, and plots the residuals between these new 
values and sale values against some other value factor, such as distance 
from market, for instance. As a result of the new graph he may find 
he has to change the unit value for distance from market. He then 
checks other unit values, and when through, starts all over again with 
crop land. He proceeds in this way until the residuals plotted against 
his various value factors average approximately zero for all combina- 
tions of these value factors, and the average line drawn through the 
points is a horizontal straight line. It should not require more than 
two or three sets of graphs to come to this approximate result. 
Irrespective of the use of formal surveys, the rural assessor needs 
to handle his pencil much more than he now does in building up his 
valuations. Even those who are able to guess values with a fair 
degree of accuracy would be able to make their guesses still more 
accurate if they would set down on paper the various factors involved 
