A Method of Graphical Analysis 



By HELENE C. BATEMAN 



Introduction 



IN connection with many telephone problems of an economic 

 character, it is necessary to develop methods for making estimates 

 and forecasts of the effects of changes in conditions. When the changes 

 in conditions are such that direct experimentation is impracticable 

 the development of logical methods and bases for estimates involves 

 analyses of past experience in specific situations and, in so far as is 

 feasible, the generalization of such experience. It is the purpose of 

 this paper to describe briefly a graphical method by which complex 

 economic data may be generalized for use in forecasting probable 

 future conditions. 



In some problems, it is necessary to determine the effects of changes 

 in a specific situation, the results being applicable particularly to the 

 given situation, and only very generally to other situations. The 

 effect of a change in population upon station growth in a given ex- 

 change is an example of such a problem. In other problems, it is 

 practicable to generalize experience so that the results of analyses 

 may be applied, under proper conditions and limitations, to various 

 specific situations. Moreover, it is often necessary to apply a general 

 conclusion to a specific situation because no specific experience is 

 available. An example of this type of analysis is the generalization 

 of results of various rate treatments in different exchanges. In meet- 

 ing this type of problem graphical methods are utilized to compare 

 experience of a similar nature in various situations. The factors 

 which may be indices of differences in conditions among various 

 situations are studied to determine their relation to the differences 

 encountered. Finally an attempt is made to derive quantitative re- 

 lationships from the experience analyzed. 



The assumption made in utilizing such methods is that the experience 

 in different situations, from which generalizations are to be made, is 

 essentially similar in certain respects, and that the variation in the 

 quantitative unit to be estimated is due to varying conditions, as 

 between the different situations, which may be measured in part 

 by quantitative factors. There are, of course, certain types of prob- 

 lems where essential similarity between different situations does not 

 exist or where it is difficult, if not impossible, to isolate quantitative 

 factors sufficiently reliable to form a basis for estimates. On the 



77 



