6 
[5 
BANKING STANDARDS 
amounts in each line so as to get the net average percentage 
deviation for the factor in the caption treated as the dependent 
variable; and (8) reversing processes (6) and (7) so as to get the 
average percentage dispersion of the factor in the caption, when 
used as the independent variable, and the net average per- 
centage dispersion for the factor in the stub, when treated as 
the dependent variable. 
The results secured from these processes for two series are 
contained in Table 96. The total columns at the right show, for 
districts with ratios of loans and discounts to earning assets 
which deviate different percentage amounts above or below their 
respective seven-year averages, the net percentage amount of 
deviation in their ratios of gross earnings to earning assets—the 
loans and discounts ratios being the independent and the gross 
earnings ratios the dependent variable. On the other hand, the 
totals at the bottom of the table show, for districts with ratios 
of gross earnings to earning assets which deviate different per- 
centage amounts above or below their respective seven-year aver- 
ages, the net average percentage amount of deviation in their 
ratios of loans and discounts to earning assets—the gross earnings 
ratios being the independent, and loans and discounts the depend- 
ent variable. 
By carrying through the successive steps enumerated above, 
the net average percentage deviation of ratios of gross earnings 
ior each percentage group-deviation of ratios of loans and dis- 
counts, as shown in Table 97, was secured. In a similar manner, 
the net average percentage deviation of ratios of loans and dis- 
counts for each percentage group-deviation of gross earnings, 
as given in Table 131, was obtained. 
The procedure followed in correlating district differences in 
paired series is the same as that used in correlating year-to-year 
changes and deviations from the levels for the country as a 
whole, the only exception being, of course, that the deviations 
and changes are measured in ways suitable to the comparisons 
involved.