EXPENSES IN DISTRICTS I AND II 293 
pair of years the arithmetic mean ratios for all of the banks in 
the district. (2) Determine the percentage amount and direc- 
tion by which the ratio for each bank each year deviates from 
the yearly average for all banks for that year. (3) Select suitable 
groups for the deviations above and below the averages in the 
first and in the second of each pair of years. (4) Distribute the 
deviations for the several banks by signs and percentage amounts 
into the groups provided for in (3) above. (5) For each per- 
centage group, determine for the first year the arithmetic mean 
deviation, and for the second year the arithmetic mean net devia- 
tion. To secure the net deviation in the second year, add the 
amounts of the deviations algebraically and divide by the num- 
ber of deviations. (6) For each of the groups and for the totals 
above and below the averages for the respective years, compare 
the amounts of deviation as secured in (5). The differences 
between them represent the amounts of regression to type as 
measured by differences in mean dispersion—minus changes for 
ratios larger than the average, and plus changes for those 
smaller than the averages indicating regression toward the 
average. 
By carrying through the steps in the process just indicated, 
:hree sets of results are obtained for the pairs of years between 
1922 and 1925. These may then be combined and averaged to 
secure regression measures for the entire experience. The dif- 
ferences for the three pairs of years, separately and combined, 
provided for in (6) above, are indicated in Table 168. 
What is the meaning of the details in this table? It will be 
observed, first, that for ratios larger than the average in the first 
year of each pair of years (See the stub class marked “Above”) 
the signs in the columns headed “Second year less than first 
year,” are invariably negative, thus indicating that ratios of the 
banks involved deviated less in the second year from the average 
in that year than did the ratios of the same banks in the first 
year deviate from the average for that year. Second, that for 
ratios smaller than the average in the first year of each pair 
of years, (See the stub class marked “Below”), the signs in the 
columns headed “Second year less than first year” are invariably 
positive, thus indicating that ratios of the banks involved devi- 
ated less in the second year from the average in that year than