RELATION OF LAND [NCOME TO LAND VALUE. 
17 
rents are based. (See Table 2 in the Appendix.) Furthermore, these 
correlations would be considerably higher than they are if average 
rents were correlated with all the factors which affect productivity. 
When percentage of woodland and percentage of land needing drain- 
age hv and d) were added to yield and percentage of improved land, 
trie correlation coefficient in group 7 was raised from .65 to .74, in 
group 10 from .81 to .86, in groups 37, 42, and 44 from .85 to .90. 
These adjusted rents, however, are less accurate in the other 
groups than the land values. In considering the reasons for their 
lower degree of accuracy, it should be remembered how they were 
derived. The ratios of rent to value on the rented farms were multi- 
plied by the county average value of land to get the county average 
rent. (See p. 6.) The average rent, therefore, can not be more 
accurate than the county average value. But since the values are 
highly accurate, this source of error is not great. Another source of 
inaccuracy in the adjusted rents is to be found in the inaccuracy of 
the ratios of rent to value on the rented farms, and it is the inaccuracy 
in this figure that causes the correlation of rent and the productivity 
factors to fall below that of value and these factors. 
This forces one back to the question of the accuracy of the average 
rents on the rented farms and the accuracy of the estimated values 
on these farms, Are the variations in these ratios from county to 
county due to inaccuracies in the average rents or to inaccuracies in 
the estimated values on the rented farms ? An answer to this ques- 
tion was sought, but it has been necessary to take this indirect 
method because the data are not available to determine the relative 
accuracy of the average rents or the estimated values on the rented 
farms directly. However, the following correlation coefficients in- 
dicate where the error lies. 
Table o. — Correlation between adjusted and unadjusted average rents with county 
average land values. 
Group. 
Correlation coefficient. 
Adjusted 
rent and 
value. 
Unadjusted 
rent and 
value. 
7 
.s2 
.89 
.75 
.91 
.So 
.95 
si 
in 
90 
11 
13 
32,33,37, 41, 42, 44. and 4."i 
.go 
37, 42, and 44 
.94 
Table 5 shows that the correlation between unadjusted rent and 
value was about the same, in every group except 11, as the correla- 
tion between adjusted rent and value. In group 11 the correlation 
is actually higher in the case of unadjusted rent. This indicates one 
of two things: (1) The rented farms are generally of the same aver- 
age grade and of the same average value as the average of all the 
farms in the county; (2) there is considerable error in the estimated 
values on the rented farms. 
It is not likely that the first of these propositions is true. There 
is only one area in which the number of one-year tenant farms per 
77.402°— 24 2 
