Tuesday, 15 May 2012

Stratified Sampling

  Manoj       Tuesday, 15 May 2012

There  may  often  be  factors  which divide  up  the  population  into  sub populations  (groups  /  strata)  and  we may expect the measurement of interest to  vary  among  the  different  sub-populations. This  has to  be  accounted for  when  we select  a  sample from  the population in order that  we  obtain a sample that  is  representative  of  the population. This is achieved by stratified sampling.
A stratified  sample is obtained  by taking  samples  from  each  stratum  or sub-group of a population. When  we  sample  a  population  with several strata, we generally  require  that the  proportion  of  each  stratum  in  the sample  should  be  the  same  as  in  the population.
Stratified  sampling  techniques  are generally  used  when  the  population  is heterogeneous, or dissimilar,  where certain  homogeneous, or similar, sub populations can  be isolated  (strata). Simple  random  sampling  is  most appropriate  when  the  entire  population from  which  the  sample  is  taken  is homogeneous. Some  reasons  for using stratified  sampling  over  simple  random sampling are:
  • the  cost  per  observation  in  the survey may be reduced;
  • estimates  of  the  population parameters  may  be  wanted  for each sub-population;
  • increased accuracy at given cost.

Example 

   Suppose a farmer wishes to work out  the  average milk  yield  of  each  cow type  in  his  herd  which  consists  of Ayrshire, Friesian, Galloway and Jersey cows. He  could  divide  up  his  herd  into the  four  sub-groups  and  take  samples from these.
logoblog

Thanks for reading Stratified Sampling

Previous
« Prev Post

No comments:

Post a Comment