Page 64 - SREENARAYANAGURU OPEN UNIVERSITY

The table usually contains 5-digit numbers, arranged in rows
and columns, for ease of reading.

◆ Assume you have the test scores for a population of 200
students. Each student has been assigned a number from 1 to
200. We want to randomly sample only 5 of the students for
this demo.

◆ Since the population size is a three-digit number, we will use
the first three digits of the numbers listed in the table.

◆ Without looking, point to a starting spot in the table. Assume
we land on 75635 (3rd column, 2nd entry).

◆ Series of digits ◆ This location gives the first three digits to be 756. This choice
arranged in rows is too large (> 200), so we choose the next number in that
and columns column. Keep in mind that we are looking for numbers whose
first three digits are from 001 to 200 (representing students).

◆ The second choice gives the first three digits to be 407, also
too large. Continue down the column until you find 5 of the
numbers whose first three digits are less than or equal to 200.

◆ From this table, we arrive at 070 (07015), 038 (03811), 045
(04594), 055 (05542), and 194 (19428).
◆ RESULT: Students 38, 45, 55, 70, and 194 will be used for
our random sample.

B. Stratified random sampling
This is similar to simple random sampling but the population
is first divided into groups which possess similar characteristics
◆ Heterogeneous strata known as ‘strata’. Sample is drawn from each stratum to constitute
with homogeneous a representative sample. For example, a class of 50 students can
elements be divided based on gender as boys (20) and girls (30) then this
sample would be 6 and 9 respectively if we are to take 30 percent
as sample.
All the sub groups have a chance of being selected thus making
the sample more representative. Higher statistical efficiency is
ensured and it is easy to carry out. Whereas to adopt stratified
sampling the researcher has to know the composition of the
population so as to effectively form strata. This is comparatively
expensive and time consuming than simple random sampling.
Also, there are chances of classification errors while building up
strata.
◆ Proportionate and
disproportionate Stratified samples can be taken in two ways: one is to take
stratified sampling proportionate samples from each of the strata (like in earlier
example 6 and 9). This is called proportionate stratified sampling.

58 SGOU - SLM -MCom Research Methodology