WHAT MAKES THE PRICE OF OATS 
25 
Table 15 shows the coefficients of correlation resulting from the 
use of different methods of treating the variables, as in the study of 
prices of oats. It will be seen that with wheat the results were ap- 
proximately the same in the three cases. 
NOTES ON METHOD 
As a study of wheat and oat prices has expanded, so also has the 
statistical technique employed been adapted to new uses. At the 
beginning, for example, variables were expressed as ratios to trends. 
Later, variables expressed as percentage changes over the preceding 
year were used for comparisons extending through and beyond the 
period of the World War, because straight-line trends could not be 
satisfactorily extended through that period. 
Table 15. — Table of coefficients of multiple correlation relative to wheat prices, 
period 1895 to 191 4 
Dependent variable 
Independent variables 
Coeffi- 
cient 
(a) Ratio of price divided by farm 
products relative to its straight- 
line trend. 
(a) Ratio of price divided by farm 
products relative to the same of 
the preceding year. 
(a) Ratio of price, uncorrected, to 
that of preceding year. 
(b) Ratio of world production plus carry over of wheat to 
its straight-line trend. 
(c) Ratio of world production of barley to its straight-line 
trend. 
(d) Ratio of world production of rye to its straight-line 
trend. 
(b) Ratio of world production plus cany over of wheat to 
the same of preceding year. 
(c) Ratio of world production of barley to that of preceding 
year. 
(d) Ratio of world production of rye to that of the preced- 
ing year. 
(b) Ratio of farm products index number to that of preced- 
ing year. 
(c) Ratio of world production plus carry over of wheat to 
the same of the preceding year. 
(d) Ratio of world production of barley to that of the pre- 
ceding year. 
(e) Ratio of world production of rye to that of the preced- 
. ing year. 
0.83 
.82 
.82 
On page 12 it was suggested that variables be expressed as per- 
centage changes over the average of the figures for the preceding and 
current years. Finally variables expressed as simple changes in 
cents, tens of millions of bushels, and points of an index have come 
to be used. Table 10 shows that for the purpose at hand variables 
expressed in this way were relatively accurate as well as simple, 
direct, and convenient. 
No generalizations are made to show the superiority of any method. 
In one case it may be desirable to sacrifice exactness for the sake of 
simplicity; in another inaccuracy may be too high a price to pay for 
simplicity; whereas in some instances simplicity may accord with 
the greatest accuracy. In the first part of the bulletin straight-line 
trends were used because they best described the data. Linear com- 
parisons were used in relating oat prices to production because, for 
the data at hand, curves suggested but slightly closer agreement and 
involved considerably more work. For expressing seasonal variation 
the methods of link relatives and averages gave closely comparable 
results. 
