Chapter 5

Measurement and
Scaling
 Outline of the presentation
• Introduction
• Types of data
• Measurement scales
• Sources of error in measurement
• Qualities of the good measurement
• Techniques for developing measurement tools
• Scaling
• Scaling classification
• Scale construction techniques
• We use measurement regularly in our daily lives
• Someone asks you of your favorite newspaper
• Your mind may create a list and you shall decide your
favorite most newspaper from that
• Your mind would have used several criteria such as your
reading pattern, content of the newspaper, writers
involved, format, color and pictures used
• The criteria your mind is using in deciding the favorite
newspaper is called measurement
• Scaling can be defined as an extension to the process of
measurement
• A fast food home delivery shop may wish to ask its
customers. How would you rate the service of our staff?
• They develop a scale where in 1 = excellent and 10 =
worst
• The consumers now can respond using these boundaries
• Carry out some statistical analysis and also provide
results which can easily be understood
• Measurement and scaling is highly important in
marketing and business research
 Measurement
• Measure physical objects as well as abstract concepts

• Standard unit of measurement (yardstick)

• Expect high accuracy

• Not easy to measure (complex)

• Less confident about the accuracy
• Measurement is process of mapping aspects of a domain
onto another aspects of a range according to the rule of
correspondence.
If the object in the domain appears to
be male, assign to “0” and if female assign to “1”.

Rule of correspondence
(Gender-Male/Female)

• Person’s marital status (single, married, widowed or divorced) – 1 to 4

• Yes or No answers to a question – 0 or 1
• Measurement is defined as the process of associating
numbers or symbols to objects or observations obtained
in a research study.
• Observed not measured
Quantitative Data
• Eye color, Gender, Marital status, Skin softness
• Non-numerical data
• Categorized

Quantitative Data
• Discrete data
• Continuous data

• Counted and measured

• Most analysis
• Mean, SD etc.
Note: clear understanding of type of characteristics of data
Feelings and
Perception
Types of data
 Nominal data
• Numerical in name only
• “Nominal” scales could simply be called “labels”
• Do not share any of the properties of the numbers
• Example:
– Marital status: single(1), married(2), widowed(3), divorced (4)
– No meaning on 4>2 or 3-1=4-2
 Ordinal data
• Cannot do anything except set up inequalities
Meaning:
5>2
No meaning:
10-9=5-4
Twice as hard (4 and 8)
 Interval data
• Addition to setting up inequalities, differences can also
be formed
• Example:
– Reading temperature in degree Fahrenheit
– 110⁰ is warmer than 70⁰
– 95⁰ is cooler than 135⁰
– 95⁰-70⁰=135⁰-110⁰
– No meaning on 126⁰/63⁰=2
– Not to say 40⁰ is twice as hot as 20⁰
– 0⁰ is not indicative of the absence
 Ratio data
• In addition to setting up inequalities, forming differences,
quotients can also be formed
• Example:
– Absolute zero
– Length, height, money, weight, volume, area, pressure
– Perform all customary operations of mathematics
– 30 cm is twice as 15 cm
Scale Measure Example Properties

Nominal Numbers • Convenient way of tracking

assigned to • Used in surveys and other ex-
runners (label) post-facto research
• Categorization
• Counting purpose
Ordinal Rank order of • Place events in order - ranking
winners • More precise comparison
cannot be made (greater/less
than)
• Real difference is not same
Interval Performance • Don’t have a “true zero”
rating in 0-10 • Equality of interval
scale

Ratio Time to finish • Absolute zero measurement

in seconds • Facilitates a kind of comparison
• Measure physical dimensions
Scale central measures of measure of statistical
tendency correlation dispersion significance

Nominal Mode Contingency Chi-square test

coefficient
Ordinal Median Percentile

Interval Mean Product Standard ‘t’ test and ‘F’

moment Deviation test
Ratio Geometric and All statistical
harmonic techniques
means

Physical sciences – ratio scale

Behavioral sciences – interval scale
Sources of Error in
Measurement
 Possible sources of error in
measurement
• Respondent
– Reluctant to express strong negative feelings or very little knowledge not
admit their ignorance – interview of guesses
– Factors like fatigue, boredom, anxiety limit the respondent
• Situation
– Strain on the interview
– Factors like presence of another person / lack of assurance of anonymity
• Measurer
– Interviewer can distort response by rewording /reordering questions,
carless mechanical processing, errors due to incorrect coding, faulty
tabulation
• Instrument
– Defective measuring instrument / complex words / ambiguity / poor
printing / inadequate space for replies
 Tests of sound measurement
1. Validity
– Refers to the extent to which a test measures what we
actually wish to measure
– Indicates the degree to which an instrument measures
what it is supposed to measure
– Three types: content, criterion-related and construct
• Content validity
– adequate coverage of topic under study
– Representative sample of the universe
– Eg: "the systematic examination of the test content to
determine whether it covers a representative sample of the
behavior domain to be measured"
• Criterion-related validity
– Ability to predict some outcome or estimate the existence of
some current conditions
– Eg: employee selection tests are often validated against
measures of job performance (correlation b/w test and
criterion)
– The concerned criterion posses the following qualities:
Relevance, Freedom from bias, Reliability, Availability
– Types: Predictive validity and concurrent validity

• Construct validity
– degree to which the measurement confirms to predicted
correlations with other theoretical propositions
– practical tests developed from a theory
– Eg: find out if an educational program increases emotional
maturity in elementary school age children
2. Reliability
– A measuring instrument needs to provide consistent
results
– Reliable instrument need not be a valid instrument, but
valid instrument is always reliable
– Two aspects of reliability: stability aspect – measure the
results of repeated measurement, equivalence aspect –
measure how much error introduced by different
investigators
3. Practicality
– Judged in terms of economy test (affordable), convenience
test (easy to administrable) and interpretability test
(interpretable by others)
4. Accuracy
– True representative of the observation of underlying
characteristics
 Techniques of developing
measurement tools
1. Concept development
– Researcher should understand the major concepts pertaining
to his study.
– Fundamental concepts are already established.
2. Specification of concept dimensions
– Accomplished by deduction or empirical correlation of the
individual dimensions with the total concept.
3. Selection of indicators
– Researcher must develop indicators (specific questions/scales)
for measuring each concept element.
– Adopt more than one indicator for stability and validity.
4. Formation of index
– Combine the various indicators into a single index.
– Overall index would provide better measurement than single
indicator
 Scaling
• When the concepts to be measured are complex and
abstract and we do not posses the standardized
measurement tools and to measure abstract concepts
(opinion/attitudes) more accurately.
• Need of scaling techniques.
• Scaling describes the procedures of assigning numbers to
various degrees of opinion, attitude and other concepts.
• Done in two ways:
1. Making a judgement about some characteristic of an
individual and then placing him directly on a scale that been
defined in terms of the characteristic
2. Constructing questionnaires in such a way that the score of
individual’s responses assigns him a place on a scale
• Scaling is used to determine quantitative measures of
subjective abstract concepts

Scaling refers “procedure for the assignment of numbers (or other

symbols) to a property of objects in order to impart some of the
characteristics of numbers to the properties in question.”
 Scaling classification bases
• The number assigning procedures or scaling procedures
classified into:
• SUBJECT ORIENTATION: A scale may be designed to
measure characteristics of the respondent who completes it
(between-stimuli variation is small) or judge the stimulus
object which is presented to the respondent (between-
respondent variation is small)
• RESPONSE FORM: Scales can be categorical (rating –
without reference to other objects) and comparative (ranking
– compare two or more objects) scale
• DEGREE OF SUBJECTIVITY: Scale based on measuring
subjective personal preferences (the respondent is asked to
select which person he favors to be employed) or make non-
preference judgement (the respondent is asked to judge
which person or solution will be more effective without
reflecting any personal preference).
• SCALE PROPERTIES: Scales based on Nominal (without
indicating order) / Ordinal (magnitude relationship) / Interval
(order and distance, no unique origin) / Ratio (all features)
• NUMBER OF DIMENSIONS: Scales can be Unidimensional
(measure only one attribute) / Multidimensional (described
better with ‘n’ attributes)
• SCALE CONSTRUCTION TECHNIQUES: Arbitrary approach
(adhoc basis); Consensus approach (panel of judges
evaluate); Item analysis approach (individual items analyzed
by a group of respondents and total scores calculated);
Cumulative scales (conforms to ranking of items); Factor
scales (inter-correlations of items indicating a common
factor )
Classification of scaling techniques

Ranking Rating
Comparative scaling
• Direct comparison of stimulus objects and make choices
• Measured on ordinal scale – cannot used for numeric
operations

• Disadvantage is inability to generalize beyond the

stimulus objects
Non-Comparative scaling
• Each object is scaled independently of others
• Respondent asked to give preference score on a 1 to 10
scale to product ‘A’
• Similar scores were obtained for other products
Types of comparative (ranking) scaling
 (i) Paired comparisons

2 3 1 4 ‘C’ is most popular

• Making a choice between two objects

• No. of paired comparison is
4x(4-1)/2=6
6x100=600
• Using the data in the above table, we work out the column
mean for A with the help of the formula given below:

• Like that do for A, B and D

• Z values for Mp secured from area under normal curve table.

Mp value is less than .5, the Z value is negative and for all Mp
values higher than .5, the Z values are positive
• Eliminate negative scale values by giving the value of zero to
the lowest scale value and then adding the absolute value of
this lowest scale value to all other scale items.
Question to Solve
(ii) Rank Order
• Ask to rank their choices
• Easier and faster than
paired comparison
• Problem of transitivity is
not there
• 10 instead 45
• Complete ranking is not
needed some times
• Problems: ordinal data,
careless in ranking when
there are many items
(iii) Constant sum

• Respondent gives
certain points to each
stimulus object out of
a fixed sum of points
• Take less time
• Not useful with large
number of objects
Types of non-comparative (rating)
scaling

• Judge the object against some criteria

• Like/dislike, excellent/poor etc.
• Three to seven points scales
 (i) Continuous rating (or) Graphic
rating scale

• Simple and used in practice

• Making a mark at the appropriate point on a line
• Limitation: Respondents may check at almost any position
along the line, which may increase the difficulty of analysis
 (ii) Itemized rating (or) numerical
rating scale
• Presents a series of statements from which a respondent selects
one as best reflecting his evaluation.

How well does a worker get

along with his fellow workers?

• Merit: provides more information

• Demerit: relatively difficult to develop and the statements may not
exactly what the respondent would like to express
• Three types of errors:
– Error of leniency - when certain respondents are either easy raters or hard
raters.
– Error of central tendency - when raters are reluctant to give extreme
judgements
– Error of hallo effect - when the rater carries over a generalized impression
of the subject
 (iii) Single/Multiple category scale

• Dichotomous scale
• Two or more mutually exclusive responses
• Can have more than one responses
 (iv) Verbal frequency scale
• When the respondent is
unable or unwilling to give
the exact numbers in
answer

How often do you eat out?

1. Frequently
2. Sometimes
3. Rarely
4. Never
 Scale Construction Techniques
• Different scales for measuring attitudes of people
 a) Arbitrary scales
• Arbitrary scales are developed on ad hoc basis and are
designed largely through the researcher’s own subjective
selection of items.
• The researcher first collects few statements or items which he
believes are unambiguous and appropriate to a given topic.
• Merit: developed easily, quickly and with relatively less
expense.
• Limitation: we do not have objective evidence that such scales
measure the concepts for which they have been developed
 b) Differential or Thurstone-type
scales
• Developed using consensus scale approach.
• The researchers gathers a large number of statements
belonging to the topic area
• Judges sort large number of statements into 11 piles that are
psychologically equidistant. (most favorable to unfavorable)
• Median scale values calculated for each statement.
• Most consistent or representative statements are chosen for
the final scale.
• Merits: Appropriate and reliable to measure a
single attitude
• Limitations: Cost and effort required, Judges may
reflect their own attitudes, Not completely
objective
http://www.socialresearchmethods.net/kb/scalthur.php
c) Summated or Likert scales
5X5 = 25 Favorable answer
5X3 = 15 Neutral answer
5X1 = 5 Unfavorable answer
• Summated scales
• Collect large number of
statements relevant to the
attitude being studied
• Merit: easy, reliable, less
time to construct
• Limitation: Cannot tell how
much more or less favorable
the respondents are
 d) Cumulative scales
• Louis Guttman’s scalogram analysis consist of series of
statements to which a respondent expresses his agreement or
disagreement
• Scalogram analysis refers to the procedure for determining
whether a set of items forms a unidimensional scale.

• Merit: easy to administer, measure single dimension of attitude

• Limitation: this analysis is a bit more difficult in comparison to
other scaling methods.
Question to solve
 e) Factor - Semantic differential scales
• Based on intercorrelations
of items which indicate
that a common factor
accounts for the
relationships between
items (+ve & -ve attitude)
• SD scale measure
psychological meanings of
an object to an individual
• Object have different
dimensions located in
multi-dimensional space
• Set of bipolar rating scales
Merit: efficient and easy way to secure attitudes from a large sample.
 e) Factor – Multidimensional scales
• MDS techniques can represent geometrically the locations
and interrelationships among a set of points.
• Given a list of city locations, and were asked to create a map
that included distances between cities.

• Not widely used due to computation complications

f) Staple scale
• Unipolar rating scale with
10 categories
• Does not have zero or the
neutral point
• + means the highest
degree of accuracy
• - means the highest
degree of inaccuracy
• The friendliness of the
cabin crew is good,
comfortable of interiors
are ok but timings is not
accurate