WHAT MAKES THE PEICE OF OATS 11 
Using estimates for the period 1881 to 1913 and Chicago prices 
gave similar results. Of course, these findings are not conclusive, 
because at best they are estimates and because considerations such 
as local prices, differences due to grades, and cost of storage have 
not been taken into account. However, they do suggest that there 
is an economic basis for efforts to distribute the supply in a more 
orderly manner. 
MULTIPLE CORRELATION OF PRICE FACTORS 
Since a correlation high enough to furnish an accurate basis for 
forecasting prices can not be obtained by using production alone, 
it will be necessary to look for other factors that have an influence 
upon the price of oats and to measure their influence. Those factors 
wnich may be expected to have some effect include: (1) Changes 
in the general price level; (2) the year-to-year carry over of oats; 
(3) substitute crops, such as corn; and (4) production of oats in 
Canada. 
The effects of long-time changes in the general price level were 
taken account of to some extent in the first correlation by eliminating 
the long-time price trends. This is not an accurate method for the 
trend corresponds only roughly with the actual year-to-year changes 
in the price level. By the use of multiple correlation the general 
price level may be treated as a separate factor or variable, along with 
production and price. The annual carry over of oats, which is, in 
effect, merely an addition to the annual production, may be taken 
care of by adding it to the production for each year. 
A multiple correlation u using (1) the percentage change in the 
price per bushel over the price of the previous year, (2) the percentage 
change in the index number 12 and (3) the percentage change in the 
United States production of oats plus carry over 13 gave a coefficient 
of 0.86. The inclusion of two additional factors has thus raised the 
coefficient by four points. The significance of a coefficient of multi- 
ple correlation differs somewhat from that of a coefficient of simple 
correlation. It indicates the closeness of the relation between the 
dependent variable and the independent variables, but not the nature 
of the relationship, as does the coefficient of simple correlation. 
Accordingly, it is not accompanied by a positive or negative sign. 
Additional correlations, using the other factors mentioned, show 
that corn added as a fourth variable does not increase the coefficient 
of correlation. This may be partly owing to the fact that the corn 
crop tends to be small when the oat crop is small and large when the 
oat crop is large. Using the index number of all commodities of the 
Bureau of Labor Statistics in place of the index number of farm 
products gives a correlation that is still 0.86. Adding Canadian 
production to the above, as a fourth variable, does not raise the 
coefficient. The period covered in each of the correlations was 
1896 to 1922, with the omission of the crop years 1916, 1917, 1918, 
1919, and 1920. 
11 The method of working out a correlation of three variables is explained in Table IV, Appendix A, p. 31. 
12 The index of farm products of the United States Bureau of Labor Statistics converted to the crop year, 
July to June, by averaging the monthly index numbers back to 1913. Previous to 1913 the annual index 
numbers for two calendar years were averaged to give an index number for the crop year. 
13 Carry over includes old stocks of oats on farms August 1, 1895 to 1923, obtained from the publication 
Weather Crops and Markets, United States Department of Agriculture, August 11 and 18, 1923, and the 
visible supply of oats on August 1, as reported by the Yearbooks of the United States Department of Agri- 
culture. 
