Deviation of Random Samples from Average 

 Conditions and Significance to Traffic Men 



By E. C. MOLINA and R. P. CROWELL 



THK traffic executive deals with questions which lead him into 

 the consideration of problems of widely differing natures. At 

 almost every turn he is confronted by the fact that his decisions and 

 programs in relation to these different phases of the work must be 

 based on records which are seldom continuous and in most cases are 

 merely "samples." These sample records are assumed to measure 

 the characteristics of the entire volume of facts or data of which they 

 are taken to be representative. In the use and analysis of these 

 records there are a number of perplexing questions which come to 

 his mind if he allows himself the luxury of a little theoretical speculation. 



Practically all of his information regarding the efficiency with 

 which his office is run and on which he must base his plans for con- 

 tinued efficiency is obtained from the peg counts. These peg counts 

 are records of the number of calls handled and are taken on two or 

 three days out of each month. At the same time that the calls are 

 counted, the number of employee hours used in the handling of the 

 traffic is counted. The results of these peg counts are used to repre- 

 sent the performance of that office for the month. When the inquiring 

 traffic man meditates a little on the subject of these peg counts he 

 soon begins to wonder how nearly representative they are of his every 

 day performance. He can — and sometimes does — think up a number 

 of things which will explain any poor results which show up. 



One of the means taken to insure the accuracy of the peg count is 

 to observe the counting of 25 to 50 calls each by as many of the oper- 

 ators as possible, with the idea of determining how accurately the 

 operators count. In this way from 1,500 to 3,000 observations are 

 made on the accuracy of the operators' counting, in a period of two 

 or three days. The traffic man occasionally questions whether he 

 can rely on the results of this comparati\ely small number of checking 

 observations to give him an indication of the accuracy of the count 

 as a whole. 



In order that comparisons may be made of the performance of 

 different offices and the cost of handling different kinds of calls, it 

 is the practice to translate all the work done into terms of traffic units 

 (representing the relation of the labor value of the different opera- 

 tions to a fixed value arbitrarih- selected). In order to do this, at 

 longer intervals than the regular peg counts, the traffic is counted in 



