100 BELL SYSTEM TECHNICAL JOURNAL 



siderable error. In other cases it may happen that some part of the 

 data necessary for making a complete estimate is not available. It 

 may be practicable, however, to approximate the required informa- 

 tion and make a rough estimate which may be more accurate than the 

 alternative of basing the estimate upon less complete analysis. 



In applying the results of such analyses, satisfactory conclusions 

 can be reached only if due consideration is given to the following 

 points: 



1. The quantitative readings from the normal curves. 



2. All the qualitative relationships developed in the course of the 

 analysis. 



3. Any additional data available for the particular case or cases in 

 question. 



4. Any peculiar special conditions known to exist for that case or 

 which probably exist because of comparison with similar cases. 



5. Changes affecting general levels since the date of the study. 



It follows that the making of such estimates cannot be left in inex- 

 perienced hands any more than can the progress of the original study. 

 Good judgment and a complete knowledge of the problem are of 

 paramount importance both in making the general analysis and in 

 the application of results to specific problems. 



To those accustomed to working in the more exact fields of physics, 

 chemistry, etc., it will undoubtedly appear that the methods de- 

 scribed above may be inexact and unsatisfactory. Undoubtedly, the 

 average errors of estimate are considerably greater than would be 

 allowed in fields where more exact data are obtainable. Yet the 

 reason for this lies rather in the material itself than in the methods of 

 dealing with it. An economic quantity is extremely complex and 

 difficult to estimate because it is usually dependent upon the action 

 of hundreds or thousands of individuals each one of whom is in- 

 fluenced by individual needs and desires which at best can only be 

 partially measured by such quantitative factors as reflect variations 

 in these needs or desires. Estimates of such quantities are neces- 

 sarily subject to a relatively high degree of error if comparisons are 

 made with the fields of physical science. Yet such estimates must be 

 made and the problem is to make them as accurately as practicable. 

 Judged from this standpoint, experience indicates that such analyses 

 are an important aid in connection with certain phases of many 

 economic problems. 



