17 



States, tenant farms are usually somewhat larger, on the average, 

 than owner farms, while share tenant farms are larger than cash 

 tenant farms. In order to get at the real influence of type of tenure 

 we must therefore eliminate the effect of size of farm. This may be 

 done by using the adjusted labor income. If the number of farms 

 in the survey were large enough the same thing could be accom- 

 plished by dealing only with farms of approximately the same mag- 

 nitude of business. Again, the number of productive work units 

 per man varies with size of farm, at least on small farms. We can 

 eliminate the influence of size of farm and study directly the effect 

 of number of work units per man on labor income by using the 

 adjusted labor income. The proportion of feed bought also depends 

 on size of farm; but if we use the adjusted labor income instead 

 of the actual labor income of each farm, we are able to study 

 the effect on labor income of the percentage of feed bought. 

 Many other illustrations might be given of the usefulness of this 

 adjustment. 



Even when the factor the influence of which is to be studied has 

 no relation to size of farm, such, for instance, as the effect ot per- 

 centage of receipts from dairy products on labor income, we must 

 depend on the law of averages to eliminate the fluctuations in 

 labor income due to differences in magnitude of business. If the 

 number of farms in each group to be averaged is quite large tl..is 

 will occur, but if the number is not large the elimination may not 

 occur. 



The use of the adjusted labor income in such cases removes this 

 difficulty and permits! the use of a smaller number of farms in each 

 group than would otherwise be permissible. That is, it renders 

 the work more accurate. 



It is evident that unless the survey includes several hundred 

 farms it is not advisable to adjust labor income in this manner, 

 because there would be too few farmd in each size group to give a 

 reliable average labor income for each group. But when the ad- 

 justed labor income has once been obtained, reliable averages can 

 then be obtained from smaller groups of farms than before, for the 

 most important fluctuation in labor income has then been elimi- 

 nated. 



1095216 3 



