42 
BULLETIN 1: 
U. S. DEPARTMENT OF AGRICULTURE 
The use of multiple correlation does eliminate the variations in 
other factors and obtain results from which the effects of all other 
(measurable) factors have been eliminated. Such results reveal the 
Corn Fed and Pork Produced; Lighter Feeding 
250 
200 
150 
100 
5C 
^ 
• " 
^^ 
Regression Line 
■ — 'One bushe/ of corn was 
accompanied bu 9.4- pounds of pork. 
/fusjj 
5 10 15 20 
BUSHELS OF CORN FED PER PIG AFTER WEANING 
25 
Fig. 14. — These pigs received less than 12 bushels of corn on an average. The regres- 
sion line, obtained by gross correlation, shows an increase of 9.4 pounds of pork for 
each bushel of corn 
Corn Fed and Pork Produced ; Heavier Feeding 
200 
150 
I 00 
? 50 
^ 
' 
' 
<^^ 
. 
"^^ ' 1 
Regression Line 
<^>^^^ ' 
' — ^^One bu 
accom 
she/ of corn wo 
panied by 7.2 p 
ounds of pork. 
* 
(fus| 
5 10 15 20 
BUSHELS OF CORN FED PER PIG AFTER WEANING 
25 
Fig. 3 5. — These pigs were fed about 13£ bushels of corn apiece. They sliov only about 
three-fourths as much gain per bushel as do the pigs on the lighter feeding 
net effect in actual farm practice of variations in an individual input 
practically as accurately as those obtained in the artificial setting of 
controlled conditions in laboratory or lost-plat. 20 
20 References on multiple correlation: 
Tolley, H. K.. and M. J. E. Ezekiel. A Method of Handling Multiple Correlation 
Problems, Journal of the American Statistical Assoc, Vol. XVI 1 1, New Series, No. 144, 
Dec., L923. 
Kelley. Truman L. Statistical Method, pp. 27f> to 29a. 1023. 
Yule, '.. I .. Id., ]>. 229-253. 
