THE  MEASURE  OF  COMPARATIVE  PRODUCTIVENESS  3OI
the  employment  manager  or  the  committee  which  is  supposed ­
  to  assist  in  the  decision  is  still  largely  dependent
upon  a  set  of  extremely  variable  personal  opinions.  It
must  rely  either  upon  the  word  of  the  employee  in  question, ­
  or  upon  general  impressions  of  its  own,  or  upon  the
word  of  the  foreman.  In  the  midst  of  this  conglomeration
of  personal  opinions,  that  of  the  foreman  is  bound  to  have
the  greatest  weight,  because  it  has  the  authority  of  one
who  is  most  closely  in  touch  with  the  employee  and  who,
even  if  he  does  not  know  the  employee  well,  is  at  least
supposed  to  know  him  better  than  anybody  else.
Most  companies  endeavor  to  cast  light  upon  this  problem ­
  by  computing  their  percentage  of  labor  turnover,
and  by  trying  to  analyze  and  interpret  this  percentage.
This  is  a  valuable  step  in  determining  the  degree  of  agreement ­
  between  selection  and  retention.  However,  up  to  the
present  time  labor-turnover  figures  have  remained  very
largely  an  unfathomable  mystery.  Although  it  has  been
possible  to  give  roughly  some  of  the  principal  reasons  for
labor  turnover,  it  is  generally  acknowledged  that  an
extremely  large  proportion  of  reasons  for  leaving  hinge
upon  personal  factors  which,  in  the  present  scheme  of
things,  can  never  be  adequately  determined.  As  long  as
the  process  of  retention  is  based  principally  upon  a  large
variety  of  constantly  changing  personal  equations  instead
of  upon  some  standard  and  impersonal  basis,  turnover  figures ­
  will  retain  their  sphinx-like  inscrutability.  Many  companies ­
  have  an  attendance  record  and  a  few  are  now  keeping ­
  a  record  of  the  work  done  by  each  employee;  but  on
the  whole,  these  records  are  still  very  haphazard  and  their
full  value  is  not  yet  grasped.  Nevertheless,  such  records
are  the  greatest  single  advance  in  the  right  direction.  In
order  to  minimize  the  errors  due  to  the  personal  equation,