Septembek 27, 1912] 



SCIENCE 



405 



school — reasons which have nothing to do with 

 the intellectual ability of the student. The 

 percentage of successful men and women who 

 have never completed a high school course is 

 still too large to warrant the conclusion that 

 the elimination which takes place can be ac- 

 curately expressed in terms of scholarship. 

 That some selection with respect to scholar- 

 ship takes place may very well be admitted 

 but at present we are not in a position to iso- 

 late this from the numerous other factors 

 which make college enrollment less than that 

 of the high schools. 



Further, to assume that 75 per cent, of its 

 students are above the general average, as is 

 done in the Reed College system, is a verdict 

 which should come from some other source 

 than Eeed College. If it is found, for in- 

 stance, that Eeed College graduates invariably 

 do better work than the graduate students 

 from other schools, it would be possible to 

 calculate the superiority of Eeed College stu- 

 dents. This is the only sense in which the 

 term selection would have any significance. 

 It is the product, not the raw material, which 

 should characterize a school. 



Eeed College could also calculate the rela- 

 tive standing of the high schools from which 

 it draws its students. This would be a com- 

 paratively simple task if all the high schools 

 graded their students in strict compliance 

 with the normal curve, but if the several 

 schools adopted curves which deviated from 

 the normal, each school deviating to the ex- 

 tent which most appealed to it, an attempt at 

 fixing a definite value for a particular grade 

 would be almost as hopeless a task as it is 

 now. 



If, however, all educational institutions 

 awarded grades strictly in compliance with 

 the normal curve, these grades would at least 

 express the same relative scholarship. If then 

 the graduate schools found that the students 

 from one institution did better work than 

 students having the same grade from other 

 institutions, the graduate schools could easily 

 calculate a selection coefficient which would 

 express the degree of selection for the differ- 



ent schools. Of course, this can not be done 

 at the present time, but the writer merely 

 wishes to point out that the factor of selec- 

 tion can not be scientifically evaluated with 

 the data now at our disposal. To encourage 

 a faculty to express this ambiguity in its 

 grades is only transferring the grading idio- 

 syncrasies from the individual instructor to 

 the faculty. 



With respect to (2), it is questionable 

 whether the selection of scholarship which ex- 

 ists is of such a nature as to change the form 

 of the distribution in a measurable degree. It 

 is of course to be understood that the average 

 accomplishment of a poor class may be less 

 than the average accomplishment of a better 

 class, but it does not follow that the grades 

 are distributed differently on either side of 

 the respective class averages. 



An elementary class in experimental psy- 

 chology of about 150 students at the Univer- 

 sity of Missouri were graded for a whole 

 semester according to the average accom- 

 plishment of the whole class. The Ebbing- 

 haus conjectural method of examination was 

 used so that the personal equation of the in- 

 structor might be eliminated as much as pos- 

 sible, and also to approach more closely to the 

 absolute accomplishment of the students.^ 

 The correlation between the actual grades and 

 those expected from a normal type of distri- 

 bution was then calculated according to the 

 following scheme. 



Number o£ Extent of Correlation 



Examinations witli Normal 



1 730 



2 970 



9 996 



We have here a gradual approach toward 

 the normal type of distribution. If only a 

 single examination is given the distribution 

 may be decidedly skewed. This does not, how- 

 ever, show conclusively that the scholarship 

 is not normally distributed. If the examina- 

 tion does not fairly test the scholarship of the 



'A more detailed report of this investigation is 

 to be found in The Jcmrnal of Experimental Peda- 

 gogy and Training College Record (Sheffield), 

 Vol. 1, No. i, June, 1912. 



