Sampling Methods at a glance

#sampling_methods #simple_random_sampling #random_number_tables #stratified_sampling

saiprasad padhy Sept 20 2020 · 1 min read
Share this

Simple Random Sampling:

A Sampling procedure in which each possible member has an equal chance of being selected.
There are two types of Simple Random sampling:

  • Simple random sampling with replacement (SRSWR) - A member of the population can be selected more then once.
  • Simple random sampling without replacement - A member of the population can be selected at most once. 
  • Eg: We have population of size 5 (A, B, C, D, E, F) and we have to select a samples of size 2 from this population without replacement. then the possible samples are:
    Possible samples: AB, AC, AD, AE, AF, BC, BD, BE, BF,  CD, CE, CF, DE, DF, EF and the probability of choosing each sample is 1 in 15. 

    Random-Number tables:

    Obtaining samples by picking slips of paper every time is impractical for large data sets. in these cases we should use some practical methods to choose samples. one common method involved israndom number tables - a table of randomnly choosen digits.
    Eg: We have to pick a sample of size 15 from a population (Size 1000). then we will assign numbers between 1 and 1000 for each member and then we will select one random number from a table, from that random number we will choose our requaired sample by eliminating duplicates. 

    Systematic random sampling:

     Simple random sample will fail if the population is widely scattered. one of the common method to overcome this problem issystematic random sampling. comparatively it will take less efforts then simple random sampling.
    Implementation steps : 

  • Divide the population size by sample size and down the results to a nearest whole number (m).
  • Use a random number table to select a value (n) between 1 and m.
  • Select the requaired number of samples - m, m+n, m+2n, ....etc.,
  • Cluster Sampling :

    Cluster sampling is widely used when our data is widely scattered.
    Implementation steps : 

  • Divide the population into clusters.
  • Select random clusters from the group of clusters. 
  • Consider all members present in the clusters (selected in Setp 2) as sample.
  • Stratified sampling :

    Stratified sampling is more reliable then cluster sampling, In this sampling the population is divided into sub-populations (strata) and then sampling is done from each strata.
    Implementation steps : 

  • Divide the population into sub-population.
  • From each strata obtain a simple random sample of size proportional to the size of the stratum.
  • Use all members obtained as a sample.
  • Comments
    Read next