See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/358149225

Scales of Measurement in Research

Article · January 2021

CITATIONS READS

0 1,283

1 author:

Anjana B.S.
University of Kerala
2 PUBLICATIONS   0 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Anjana B.S. on 27 January 2022.

The user has requested enhancement of the downloaded file.

 Scales of Measurement in Research
Anjana B.S., Research Scholar, Department of Economics, University of Kerala
[email protected]

Introduction
Variable is a term that is frequently used in research. One of the significant and
challenging aspects of planning a study is identifying the variables. A variable, as the name
implies, is something that varies (Kaur, 2013). It can be defined as something that can change
and or can have more than one value. Income, education level, and height are examples of
variables. It varies for each person and has different values along a continuum.

In order to analyse data, the variables have to be defined and categorised using
different scales of measurements. There are four scales of measurements- nominal scale,
ordinal scale, interval scale, and ratio scale. The scale of measurement of a variable
determines the kind of statistical test to be used.

Scales of Measurement

Measurement can be defined as the assignment of numerals to objects or events

according to rules, and assigning such numbers resulted in different types of scales
(S.S.Stevens, 1946). Thus scaling is the development of systematic rules and meaningful
units of measurement for quantifying empirical observations. The scales or levels of
measurement describe the nature of the values assigned to the variables in a data set
(Formplus Blog, 2020). It helps to define and group variables into different categories. The
type of data being collected determines the kind of measurement scale to be used, and the
scale determines the type of statistical techniques to be used for analysis. There are four
scales of measurement. The variable can be defined as being one of the four scales. The four
types of scales are:

1. Nominal scale
2. Ordinal scale
3. Interval scale
4. Ratio scale

Nominal scale, also known as categorical variable scale, can be defined as a scale
used for labelling variables into different categories. The numbers are used to identify and
classify people, objects or events, like identity number, jersey number of sportspersons, and
vehicle registration number; thus, they have no specific numerical value or meaning. In
research, the nominal scale is used for analysing categorical variables such as gender, place
of residence, marital status, political party, blood group and so on. The interval between
numbers and their order does not matter on the nominal scale.

For example, while measuring gender on a nominal scale, labelling female as ‘1’,
transgender as ‘2’, and male as ‘3’ do not mean that one gender is superior or inferior to the
other. They are merely used to classify persons for data analysis. In fact, any other numbers
could be assigned, for instance, ‘20’ for female, ‘38’ for transgender and ‘11’ for male,
because the numbers do not indicate an amount or a quality.

It is the lowest level of measurement from a statistical point of view. A nominal scale
preserves only the equality property; there is no ‘more or less than’ relation in this
measurement. The statistical analysis that can be performed on a nominal scale is the
frequency distribution and percentage. It can be analysed graphically using a bar chart or a
pie chart. If there are two categorical variables, quantitative analysis techniques such as joint
frequency distribution and cross-tabulation can be used. Mode is the only measure of central
tendency which can be used in this scale. Since numbers do not have a quantitative
value, addition, subtraction, multiplication, division, and measures of dispersion cannot be
applied. It is also possible to perform contingency correlation. Hypothesis tests can be carried
out on data collected in the nominal form using the Chi-square test. It can tell whether there is
an association between the variables. However, it cannot establish a cause and effect
relationship or explain the form of relationship.

Ordinal scale is a ranking scale in which numbers are assigned to variables to

represent their rank or relative position in the data set. The variables are arranged in a specific
order rather than just naming them. So they can be named, grouped, and ranked. In research,
the ordinal scale is used for ranking students in a class (1,2,3), rating a product satisfaction
(very unsatisfied-1, unsatisfied-2, neutral-3, satisfied-4, very satisfied-5), evaluating the
frequency of occurrences (very often-1, often-2, not often-3, not at all-4), assessing the
degree of agreement (totally agree-1, agree-2, neutral-3, disagree-4, totally disagree-5), and
so on. In this scale, the attributes are arranged in ascending or descending order. The numbers
indicate rank or the order of quality or quantity.
 The origin of scale is absent because there is no fixed start or ‘true zero’ in the data.
Hence, it is impossible to find the magnitude of difference or distance between the variables
or their degree of quality. For example, while ranking students in terms of potential for an
award, a student labelled ‘1’ is better than the student labelled ‘2’, ‘2’ is better than ‘3’ and so
forth. However, this ordinal scaling cannot quantify or indicate how much better the second
student to the first student, or the difference between the potential of first and second
students, the same as the difference between the second and third. Similarly, very satisfied
will always be better than satisfied and unsatisfied will be better than very unsatisfied. Here,
the order of variables is of prime importance, and so is the labelling.
The ordinal scale is the second level of measurement from a statistical point of view.
These scales are unique up to a monotone transformation. A monotone transformation T is
one that assigns new values such that if f(X) > f(Y) in the ordinal scale, then T(f(X)) >
T(f(X)) in the newly transformed scale.

The ordinal data can be presented using tabular or graphical formats. The descriptive
analysis such as percentile, quartile, median and mode can be determined in ordinal scale
data. Since the interval between numbers is insignificant, addition, subtraction,
multiplication, division, and measures of dispersion cannot be applied. It is possible to test
for order correlation using Spearman's rank correlation coefficient. Non-parametric tests such
as Mann-Whitney U test, Friedman’s ANOVA, Kruskal–Wallis H test can also be used to
analyse ordinal scale data.

Interval Scale can be defined as a quantitative scale in which both the order and the
exact difference between categories are known. Thus it measures variables that can be
labelled, ordered, and have an equal interval. However, the point of beginning or zero point
on an interval scale is arbitrarily established and is not a ‘true zero’ or ‘absolute zero’. Thus
the value of zero does not indicate the complete absence of the characteristic being measured.
In Fahrenheit/Celsius temperature scales, 0°F and 0°C do not indicate an absence of
temperature. In fact, negative values of temperature do exist. Temperature, calendar years,
attitudes, opinions and so on fall under the interval scale. Likert scale, Net Promoter Score
(NPS), Bipolar matrix table, Semantic differential scale are the widely used interval scale
examples.
The major difference between ordinal and interval scale is the existence of meaningful
and equal intervals between variables. For example, 40 degrees is higher than 30 degrees,
and the difference between them is a measurable 10 degrees, as is the difference between 90
and 100 degrees. However, while ranking students on an ordinal scale, the difference between
first and second student might be 5 marks, and between second and third student is 8 marks.
Thus, with an interval scale, it is possible to identify whether a given attribute is higher or
lower than another and the extent to which one is higher or lower than another.

The interval scale is the third level of measurement scale. The arbitrary presence of
zero has implications in data manipulation and analysis. It is possible to add or subtract a
constant to all of the interval scale values without affecting the form of the scale but not
possible to multiply or divide the values. For instance, two persons with scale positions 4 and
5 are as far apart as persons with scale positions 9 and 10, but not that a person with score a
10 feels twice as strong as one with a score 5. Similarly, 100°F cannot be defined as twice as
hot as 50°F because the corresponding temperatures on the centigrade scale, 37.78°C and
10°C, are not in the ratio 2:1. So, unlike the ordinal and nominal scale, arithmetic operations
such as addition and subtraction can be performed on an interval scale. Any positive linear
transformation of form Y = a + bX will preserve the properties of an interval scale.

The arithmetic mean, median, and mode can be used to calculate the central tendency
in this scale. The measures of dispersion, such as range and standard deviation, can also be
calculated. Apart from those techniques, product-moment correlation, t-test, and regression
analysis are extensively used for analysing interval data.

Ratio Scale can be defined as a quantitative scale that bears all the characteristics of
an interval scale and a ‘true zero’ or ‘absolute zero’, which implies the complete absence of
the attribute being measured. Thus it measures variables that can be labelled, ordered, has
equal intervals and the ‘absolute zero’ property. Before deciding to use a ratio scale, the
researcher must observe whether the variables possess all these characteristics. The variables
such as length, age, weight, income, years of schooling, price etc., are examples of a ratio
scale. They do not have negative numbers because of the existence of an absolute zero point
of origin. For instance, a price of zero means the commodity does not have any price (it is
free); and there cannot be any negative price. Thus ratio scale has a meaningful zero. Also, it
allows unit conversions like metres to feet, kilogram to calories etc.
The ratio scale is the highest level of measurement scale. It is unique to a congruence
or proportionality transformation of form Y = bX. The ‘absolute zero’ property allows
performing a wide range of descriptive and inferential statistics on ratio scale variables. It is
possible to compare both differences in values and the relative magnitude of values. For
instance, the difference between 15cm and 20cm is the same as between 30cm and 35cm, and
30 cm is twice as long as 15 cm. So, arithmetic operations such as addition, subtraction,
multiplication, and division (ratio) can be performed in ratio scale data.
All statistical operations applicable to nominal, ordinal and interval scale can be
performed on ratio scale data as well. Besides, measures of central tendency such as
geometric mean and harmonic mean and all measures of dispersion, including coefficient of
variation, can be determined. Parametric tests such as independent sample t-test, paired
sample t-test, ANOVA etc., can also be performed. So, the ratio scale provides unique
opportunities for statistical analysis.
The properties of the four scales of measurement are summarised in Table 1.
Table 1: Scales of Measurement
Nominal Ordinal Interval Ratio Scale
Scale Scale Scale
Properties Equality Equality Equality Equality
Ordinality Ordinality Ordinality
Equal Equal
Intervals Intervals
Value Ratio
Variables Can be Labelled Yes Yes Yes Yes
Can be Ranked - Yes Yes Yes
Has Measurable - - Yes Yes
Difference
Has Absolute - - - Yes
Zero
Mathematical Group Structure Permutation Isotonic General Similarity
Group Group Linear Group Group
Y = f (X) Y = f (X) Y = a + bX Y = bX
f (X) means f (X) means
any one to any
one monotonic
substitution increasing
function
Mathematical Determination of Yes Yes Yes Yes
Operation equality
(=, ≠)
Determination of - Yes Yes Yes
greater or less
(>,<)
Determination of - - Yes Yes
equality of
intervals or
differences
(Addition,
Substraction)
Determination of - - - Yes
equality of ratios
(Multiplication,
Division)
 Measures of Mode Yes Yes Yes Yes
Central Median - Yes Yes Yes
Tendency Arithmetic Mean - - Yes Yes
Harmonic Mean - - - Yes
Geometric Mean - - Yes
Measures of Range - - Yes Yes
Dispersion Standard - - Yes Yes
Deviation
Coefficient of - - - Yes
Variation

Conclusion
In a nutshell, scales of measurement are various measures to quantify the variables.
There are four scales of measurement: nominal, ordinal, interval and ratio. These constitute a
hierarchy where the lowest scale of measurement, nominal, has fewer mathematical
properties than those further up this hierarchy of scales. Each scale of measurement has
specific properties that determine the statistical analysis technique to be used. Nominal scales
are used to label a series of values; ordinal scales give the order of values; interval scales
quantify the difference between each value, and ratio scales explain the identity, magnitude,
interval values, and absolute distance between any two points on the scale since a ‘true zero’
can be defined.
References
1. Formplus Blog. (2020, November 14). Retrieved March 15, 2021, from Seven types
of Data Measurement Scales in Research: https://www.formpl.us/blog/measurement-
scale-type

2. Kaur, S. P. (2013). Variables in Research. Indian Journal of Research and Reports in

Medical Sciences , 3 (4).

3. S.S.Stevens. (1946). On the Theory of Scales of Measurement. Science , 103 (2684).

View publication stats