FARMERS' EARNINGS IN SOUTHEASTERN PENNSYLVANIA 49 
could be accounted for on the basis of this study, 21 but it did make 
possible a much closer measurement of the relations on the non- 
mushroom farms than the first results gave, this final result being 
about 50 per cent more accurate for these farms than the first solution. 
The relative importance of the several factors in determining farm 
value per acre, as shown by their coefficients of net determination, 
was as follows: 
Factor: Percent 
A. Dwelling value 11. 95 
B. Dairv buildings value 12. 45 
C. Other buildings value 19. 21 
D. Crop index 4. 55 
E. Percentage of area tillable 2. 81 
F. Percentage of area level 6. 16 
G. Tvpe of road 0. 47 
H. Distance to town 2. 08 
This solution is in much better agreement with the known facts of 
the case. This is an area of very uniform soil, practically all of the 
same type, and as a result differences in soil fertility are not so 
important as a cause of differences in land value as they might be 
in other regions where soils are much more variable. Likewise, the 
topography of this area is irregular, and as a result the proportion 
of the land area which is level is much more important than it might 
be in other regions of more uniform topography. 
Of the casual factors included, buildings as a whole are responsible 
for 44 per cent of the variations in farm value; the land, including 
fertility, topography, and proportion usable, 13 ]/o per cent; and 
general farm factors of location and road, 2J^ per cent. These 
conclusions, however, are reached in a study where 40 per cent of 
the variation in value is still left to be accounted for. Inclusion 
in the analysis of some of the other casual factors might materially 
change the apparent relative importance of some of these factors. 
As would be expected in a dairy region like this, the dairy buildings 
proved to be even more important than the dwelling house in deter- 
mining farm value. Kather surprisingly, though, the value of other 
buildings is more important than either. This may be caused by the 
fact that the coefficient of determination measures not only the 
variation due to the influence of the individual factor considered, 
but also variations due to the influence of any other factors not 
included in the study which are highly correlated with the individual 
factor. Thus the farms in this group with high values for ''other 
buildings'' are mainly those with large poultry plants. It may be 
that these farms tend to be located near the larger towns, and that 
therefore their values are higher, out of proportion to their distance 
from town. The higher variability of the value of '' other buildings " 
also tends to exaggerate their significance. Much more detailed 
analysis would be required to answer all questions of this sort that 
might be raised. In general, however, these conclusions are in 
substantial agreement with the expectations based on conditions 
within the area. 
2i The significance of a coefficient of correlation is shown by the square of its value. Thus a correlation 
0.772 shows that (0.772 )*, or 0.596, of the original variation has been accounted for, and that 1-1.0.772)2 
or 0.404, of the original variation has not been accounted for. 
