Final Comparison of Results. 
271 
paring the Milwaukee 1 with the Chicago tables, we may see that 
the same laws hold good in both cities as regards the drop¬ 
ping out of pupils. 
Maximum Limit of Error. Let us now see whether we have in 
the inductive method as applied to Chicago the maximum limit 
of error. We have reduced the grade enrollments too much if 
anything; for the population-corrections, seven to thirteen per 
cent., included the increase from births as well as from immi¬ 
gration. But if the enrollments in Table XIY are made larger, 
the percentages in Table XV will be smaller and the per cents 
in Table XVI smaller. Hence so far as all except the first grade 
are concerned, the per cents are now as large as possible. But 
the first grade should be made smaller, if anything, because of 
double enrollment. This also would make the percentages in 
Tables XV and XVI smaller. So from both points of view the 
per cents that drop out are much too large, are the largest we 
can suppose possible, and, therefore, indicate the maximum limit, 
within which the truth lies. 
<* 
Finally, if our enrollment in Table XIV be corrected not only 
for population but for double enrollment also, it is gratifying 
to find that we reach practically the same results as by our de¬ 
ductive method. 2 We thus find that the slow process of follow¬ 
ing a real class through grade after grade and making allow¬ 
ances for all possible errors only proves the close accuracy of 
the shorter deductive method. 
FINAL COMPARISON OF RESULTS. 
The next table 3 brings together for final comparison all the 
results thus far obtained, and shows that the conclusions of the 
deductive method occupy a middle position between the maxi- 
1 Table XIX. Enrollment by Classes (Milwaukee), p. 300. 
Table XX. Number Dropped out by End of Each Grade (Mil¬ 
waukee), p. 301. 
Table XXI. Per Cent, of First Grade Enrollment Dropped out 
by End of Each Grade (Milwaukee), p. 302. 
2 Table XXII. Per Cent, that Dropped out after Corrections for 
Double Enrollment , p. 303. Compare with Table IV. 
3 Table XXIII. Final Conclusions as to Per Cents that Drop out, 
p. 304. 
