textbook: Business Research Methods (Rev: 9th)

1

What are the essential differences among nominal, ordinal, interval, and ratio scales?
Ans:
Measurement Scales are important in research for many reasons.
Depending upon the type of research, different objectives and hypotheses are formulated.
Variables are identified to achieve these objectives and test the hypotheses. The variables which
are being studied, are measured to generate data. This data is further analysed to draw
conclusions of the research using different statistical techniques. The statistical techniques to be
used depend upon the type of data generated by the instruments of data collection. The data of
different types like demographic, psychographic and financial need a variety of statistical
techniques.
Four different types of data scales viz Nominal, Ordinal, Interval and Ratio; and their
characteristics along with the applicable statistical tests are given below.
Nominal:
Characteristics:
1. The data is collected in the form of frequencies and later categorized.
2. This scale is used when the data is to be categorized or labelled under different heads.
Hence the name ‘Nominal’ indicating naming. For example data related to Gender,
Nationality or Political orientation etc. This data is categorical. The categories are codes
which may be numerical but they are just the labels eg. Gender – Male = 1, Female = 2
and Other = 3.
3. These numbers (codes) do not have any numerical value. So any arithmetical operations
like addition, subtraction, multiplication and division cannot be performed using them.
Also as they do not have any numerical value, order relations like greater than, equal to
or lesser than cannot be applied to these numbers.
Statistical Techniques:
1. As the data is in the discrete form, and does not possess numerical values arithmetic
operations cannot be performed. Hence, Mean (which requires addition) and Median
2

(which requires order relations) cannot be calculated. Only available descriptive statistics
is Mode which is highest occurring score can be used. Percentages can be calculated and
represented as Pie charts.
2. As far as inferential statistics is concerned, Chi square can be used to test the hypothesis
for Goodness of Fit and tests of independence using crosstabs (Contingency coefficient).
3. Phi coefficient can be used as a measure of...