A  FIRST  EXPERIMENT

37

Correlations
Kind  of  Work  No.  of  Card  Sort.  Cane.  No.  Group
Girls  Test  Test  Checking
Shell  inspecting  51  -55  -63  -7*
Bullet  inspecting  10  .52  -4&  62
Cartridge  inspecting  28  .49  .26  .58
S.  S.  inspecting  30  .13  .11  .02

laid  off.  By  this  process  both  extremes  had  been  eliminated; ­
  that  is,  the  slow  girls  had  been  laid  off  and  the  fast
girls  had  modified  their  pace  to  suit  the  retarded  pace
°f  production.  Thus,  all  the  girls  who  were  left  tended
t0  accomplish  about  the  same  amount  of  work  per  day.
These  conditions  made  the  production  of  these  girls  an
unfair  indication  of  their  relative  ability  under  normal
conditions,  and  consequently  made  it  impossible  to  obtain ­
  a  fair  estimate  of  the  tests  on  the  basis  of  production. ­
  This  was  especially  true  in  the  case  of  the  paper
shot  shell  inspectors  where  the  lowest  correlations  were
found.  In  fact,  the  average  difference  in  the  rate  of  production ­
  between  these  girls  was  only  two  and  two-tenths
Per  cent  of  the  average  day’s  work,  as  contrasted  with  an
average  of  twenty-seven  per  cent  in  the  case  of  shell  inspectors. ­
  The  average  difference  for  the  four  groups  was
as  follows:
Per  Cent.

1.  Shell  inspectors  27.7
2.  Bullet  inspectors  18.7
3.  Cartridge  inspectors  6.9
4.  P.  S.  S.  inspectors  2.2

t  can  readily  be  seen  from  this  that  the  smaller  the  averse ­
  difference  between  the  quantity  of  work  done  by  the
w °rkers  of  each  group,  the  lower  the  correlations  in  the