FACTORS AFFECTING THE PRICE OF HOGS 43 
The following section shows that 65.8 per cent of the variation in 
the index of total demand can be accounted for by five independent 
factors. If a demand index computed from these factors were 
substituted for the demand index based partly on observed prices, at 
least this 65.8 per cent of the variation accounted for by the original 
demand index would probably be covered. Since factors other than 
demand accounted for all but 18.3 per cent (1—0.904 2 ), of the original 
price variation, using this second demand index would leave but 
1 — (0.658 X 0.843) times 18.3 per cent or 8.15 per cent of the original 
price variation unaccounted for. That is, introducing the individual 
demand factors in the equation in place of the computed index of 
demand woul d probably result in a multiple correlation of at least 
R = 0.958 (=^1—0.0815). This coefficient would be significant, since 
all of the independent factors concerned would be based on other 
things than the price itself. 
The correlation of the five factors mentioned with price gave a basis 
for measuring the relative importance of each. The coefficients of 
determination were as follows: 
Per cent 
Xi 5 Trend of market receipts • 21. 7 
Xi6 Upward trend of demand. 60. 2 
X 20 Index of total demand —6. 
X 21 Effect of price level ._ 18. 
X 22 Seasonal variation in price 3. 2 
Total determination by all factors 97. 1 
The statements as to the relative importance of the several major 
forces were based upon this solution. This rests upon the assumption 
that the index of demand as computed is a true measure of demand, 
and that hog prices have a 1 to 1 relationship with price level, a change 
of 25 per cent in price level meaning a change of 25 per cent in hog 
prices. Obviously neither of these assumptions is necessarily entirely 
true; but they do serve in this particular case to give some idea of the 
relative significance of the effects of the different components. Were 
all the different influences individually represented in a single equa- 
tion it would be possible to make a more accurate measurement of the 
importance of each within the limits of the mathematical assumptions 
implicit in the multiple correlation analysis (6). u 
FACTORS AFFECTING DEMAND 
A correlation study was made to determine how closely the index of 
demand, derived from the price by the method described previously, 
could be estimated from other factors. The factors taken into ac- 
count were as follows : 
X4 Storage of pork provisions, adjusted to eliminate normal seasonal variation. 
X 5 Index of business cycle. 
X 7 Price of steers -^Bureau of Labor Wholesale Price Index. 
X 8 Index of European demand. 
X 9 Time. 
X 20 Index of total demand. 
Correlating the five factors with X 20 for the period January, 1904, to 
December, 1915, a multiple correlation of 11 = 0.57 was obtained. 
Further analysis by curvilinear multiple correlation revealed that 
several of the regressions were very decidedly curvilinear, and raised 
13 For a noninathematical discussion of the assumptions which the statistical analysis itself involves, 
see the first part of the following: Ezeeiel, M., a statistical examination of factors eilytepto 
lamb prices. Jour. Pol. Econ. [now in press]. 
