416 AXXAL^s X-h-\V YORK ACADEMY OF ^CIEXCES 



tion seemed to eliminate many of the objections wliicli have been brought 

 against the use of completion tests with public-school children. 



Since it was desired to test the child's abilit}' to think about and to use 

 intelligently the ordinary words of the English language, the subject- 

 matter of the sentences was taken from general experience and human 

 relations, avoiding in so far as possible the more specialized fields of 

 knowledge. 



An elaborate scheme of evaluated completions was at first followed in 

 scoring each sentence, giving five points score to each perfectly completed 

 sentence, four points to each sentence only slightly imperfect, three points 

 to each sentence containing a more serious error, two points to a very im- 

 perfectly completed sentence and one point if the sentence showed any 

 evidence whatever that the child had understood the printed words. The 

 present scheme of scoring gives two points score where five were originally 

 given, one point where four or three were given and no score at all where 

 two or one were at first assigned. 



That the present method of scoring is practically as reliable as the 

 older, more elaborate method is indicated by the fact that the fifty-six 

 sentences tend to hold their relative rank regardless of which method is 

 used. The rank obtained by testing fifty-seven pupils in the last half of 

 the eighth grade showed a correlation (Spearman's method of squared 

 differences) of .965 with the rank obtained from testing thirty-four pupils 

 in the first half of the sixth grade, when the older method of scoring was 

 .employed, while with the newer method of scoring, r = .9623. The ranks 

 obtained by the two methods from the fifty-seven pupils in the eighth 

 grade gave a correlation of .9823. and with the thirty-four pupils in the 

 sixth orade. r= .9768. Such small differences between the two methods 

 of scoring the tests were not thought sufficient to warrant the enormous 

 amount of additional labor required to score the sentences by the old, 

 more elaborate scheme. 



Since each of the twenty-four sentences of Language Scale A may re- 

 ceive two points credit, the maximum score is forty-eight points. The 

 average scores in Language Scale A of over five thousand children in three 

 different school systems are given below by grades, the P.E. of any grade 

 average being 3.5 points. 



School grade. . . TI TIT IV Y VI VII VIII IX X XI XII 

 Av. score 3.4 7.7 12.1 18.2 19.5 2-2.5 26.4 28.9 32.9 84.8 37.3 



Mr. May said in abstract: The work here reported represents a pre- 

 liminary attempt to apply the order of merit method to the study of 

 "religious values.'* The materials used for the experiment were twenty- 



