TESTING TO SPECIFICATION 
I0 5 
° n e to be derived from the simplification of the method 
°f making requisitions for clerks. Because of the many 
kinds of work, the requests from various offices and de 
partments for new clerks came into the employment 
office in terms that were extremely vague. The term 
clerk, for example, might mean a clerk of almost any kind. 
The term stenographer, specific in some respects, might 
still be vague in essential details, for some stenographers 
tftight be required to have filing or statistical ability and 
others not. The term payroll clerk was found to apply 
^dually well to six or seven kinds of clerks, each engaged 
ln work of a different kind. This vagueness made it very 
difficult to examine clerks with reference to the specific 
a ffilities which were required of them by the specific tasks 
ffif which they were intended. In order to overcome this 
difficulty, it was proposed to base the requisitions for 
fffirks upon the fundamental factors which were involved 
ln e ach kind of work. That is, instead of asking simply 
°. r a P a yroll clerk, let us say, without specifying what 
ln d of payroll clerk, the request would have to state 
ffie exact specifications according to which this clerk was 
to be furnished. If the payroll clerk required was to 
P°ssess ability in division, that fact would have to be 
sta ted. If ability in sorting, that fact would be noted, and 
So forth. With the specifications given in this way, it 
Would become possible to give the appropriate tests and 
to select the candidates who most closely approached the 
s P e cifications outlined. 
The survey upon which this change of procedure was to 
c based was conducted by going directly to each office 
0r section head and asking him to explain in detail the 
ta sks of those in his group. At this point a serious obstacle 
Was immediately encountered. When the head of a group