60 BASES FOR CURRICULUM MAKING 



answered frankly stating their skepticism of the proposed meth- 

 od. On the other hand no institution was 



Many instructors found in which the ratings were taken by 

 and professors on the writer in person where the professors 

 understanding the and instructors in charge were not keenly 

 problems and the interested and evidently sympathetic with 

 method became the method and the possible value of the 

 keenly interested. results. Many of these men voluntarily 



took part in the exercise, submitting their 

 reports with those of the cless. 



It is probable that most of the errors are of a kind that 

 would tend to balance each other and, therefore, have little ef- 

 fect on the final value of the coefficients. 



Most of the errors There are, however, undoubtedly two 

 were probably of a kinds of constant errors that should be 

 kind to offset each noted. One is what may be known as the 

 other thereby caus- "halo." 38 This in a few words is a tend- 

 ing little effect on ency on the part of judges to ascribe 

 the finally accepted higher values than should be in all quali- 

 values. ties to certain men because of their gen- 



eral standing and success. For example a 



judge placing a man high in the scale for intelligence, would tend 

 to place him high also for information or skill or managerial 

 ability regardless of the facts in the case. Opposed to this is 

 another error for which correction often should be made. This 

 is known as "attenuation" 39 and in general is an error that tends 

 to reduce the value of the coefficient. Since the error due to the 

 "halo" and the error due to "attenuation" operate in opposite di- 

 rections and since with the present data corrections for neither 

 can be made to advantage, each will be assumed to equal the 

 other and therefore to have little or no effect upon the finally 

 accepted coefficient values. 



38. See article by Dr. E. L. Thorndike "A Constant 

 Error in Psychological Ratings," pages 25-29. The Journal of 

 Applied Psychology Vol. 4, No. 1. 



39. Mental and Social Measurements by E. L. Thorndike, 

 pages 177-180. 



