Attitude Measurement & Scaling

Unit 3
Measurement and Measurement Scales

 Measurement is the foundation of any

scientific investigation
 Definition: measurement is the
assignment of numbers to objects
 Scaling: Scaling is an extension of
measurement. Scaling involves creating a
continuum on which measurements on
objects are located.
Example:

When we use a personality test such as the

EPQ (Eysenck Personality Questionnaire) to
obtain a measure of Extraversion – ‘how
outgoing someone is’
we are measuring that personality
characteristic by assigning a number (a
score on the test) to an object (a person)
2.1 Data Types and
Levels of Measurement

The goal is to transform data into information,

and information into insight.”
Carly Fiorina (Executive and president of Hewlett-Packard Co. in
1999. Chairwoman in 2000)

LEARNING GOAL

Be able to identify data as qualitative or quantitative,

to identify quantitative data as discrete or continuous,
and to assign data a level of measurement (nominal,
ordinal, interval, or ratio).
Data Types

 Qualitative (or categorical) data consist of

values that can be separated into different
categories that are distinguished by some
nonnumeric characteristic.
 Quantitative data consist of values
representing counts or measurements.
Determine whether the data described are qualitative or
quantitative and explain why.

 A person’s social security number

 The number of textbooks owned by a
student
 The income of college graduates
 The gender of college graduates
Types of Quantitative Data

 Continuous data can take on any value in a

given interval. Continuous data values
results from some continuous scale that
covers a range of values without gaps,
interruptions, or jumps.
 Discrete data can take on only particular
distinct values and not other values in
between. The values in discrete data is
either a finite number or a countable number.
Discrete and Continuous
Variables

 Discrete variables have a relatively small set

of possible values
no of children
 Continuous variables can (theoretically)
assume any value between the lowest and
highest points on the scale

time, distance, weight

State whether the actual data are discrete or
continuous and explain why.

 The number of 500 RS note in

circulation
 The voltage of electricity in a power line
 The number of eggs that hens lay
 The time it takes for a student to
complete a test
Levels of Measurement

 Nominal
 Ordinal
 Interval
 Ratio
 Nominal and ordinal are qualitative
(categorical) levels of measurement.
 Interval and ratio are quantitative levels of
measurement.
Four Types of Measurement
Scales

 The scales are distinguished on the

relationships assumed to exist between
objects having different scale values
 The four scale types are ordered in that all
later scales have all the properties of earlier
scales—plus additional properties
 Nominal Scale
 Not really a ‘scale’ because it does not scale objects along any
dimension
 It simply labels objects. Nominal level of measurement—classifies
data into names, labels or categories in which no order or ranking
can be imposed

Gender is a nominal scale

Male = 1, Female = 2
Religious Affiliation

Catholic =1
Protestant =2
Jewish =3
Muslim =4
Other = 5
Categorical data are measured on nominal
scales which merely assign labels to distinguish
categories
Ordinal Scale

 Numbers are used to place objects in order.

 Ordinal level of measurement—classifies
data into categories that can be ordered or
ranked, but precise differences between the
ranks do not exist.
 Generally it does not make sense to do
calculations with data at the ordinal level.
Example—letter grades of A, B, C, D, and F.
What about symptoms of
depression from a psychiatric
assessment?

None =0
Mild =1
Moderate= 2
Severe =3
Interval Scale

 An interval scale is a scale on which equal

intervals between objects, represent equal
differences
 Interval level of measurement—ranks data, precise
differences between units of measure exist, but there
is no meaningful zero. If a zero exists, it is an arbitrary
point. Example—IQ scores, it makes sense to talk
about someone having an IQ 20 points higher than
another person, but an IQ of zero has no meaning
Fahrenheit Scale
 Interval relationships are meaningful

 A 10-degree difference has the same

meaning anywhere along the scale
 For example, the difference between 10 and
20 degrees is the same as between 80 and
90 degrees
 But, we can’t say that 80 degrees is twice as
hot as 40 degrees
 There is no ‘true’ zero, only an ‘arbitrary’ zero
Interval scale

 takes care of the limitation of the ordinal scale measurement where

the difference between the score on the ordinal scale does not have
any meaningful interpretation.
 In the interval scale the difference of the score on the scale has
meaningful interpretation.
 It is assumed that the respondent is able to answer the questions on
a continuum scale.
 The mathematical form of the data on the interval scale may be
written as
 Y=a+bX where a ≠ 0
 Ratio of the score on this scale does not have a meaningful
interpretation.
Ratio Scale

 Have a true zero point, Ratios are meaningful

 Physical scales of time, length and volume
are ratio scales
 We can say that 20 seconds is twice as long
as 10 seconds
 true ratios exist when the same variable is
measured on two different members of the
population. Example—weight of an
individual.
 Ratio scale: This is the highest level of
measurement and takes care of the limitations of
the interval scale measurement,
 where the ratio of the measurements on the

 scale does not have a meaningful interpretation.

 The mathematical form of the ratio scale data is
given by Y = b X.
 In ratio scale, there is a natural zero (origin).
 Example:How many chemist shops are there in
your locality?
 All mathematical and statistical operations can
be carried out using the ratio scale data.
CLASSIFY THE FOLLOWING AS TO QUALITATIVE
OR QUANTITATIVE MEASUREMENT. THEN STATE
THE LEVEL OF MEASUREMENT.

 Eye Color (blue, brown, green, hazel)

 Rating scale (poor, good, excellent)
 personality score
 Salary
 Age
 Ranking of high school football teams in Missouri
 Nationality
 Temperature
 Zip code
Figure 2.1 summarizes the possible data types and
levels of measurement.

Figure 2.1 Data types and levels of measurement.

Copyright © 2009 Pearson Education, Inc.

By the Way ...
Scientists often measure temperatures on the Kelvin
scale. Data on the Kelvin scale are at the ratio level
of measurement, because the Kelvin scale has a true
zero. A temperature of 0 Kelvin really is the coldest
possible temperature. Called absolute zero, 0 K is
equivalent to about -273.15°C or -459.67°F.
(The degree symbol is not used for Kelvin
temperatures.)

Copyright © 2009 Pearson Education, Inc.

Definition of Attitude

An attitude is viewed as an enduring

disposition to respond consistently in a given
manner to various aspects of the world,
including persons, events and objects.
Components of Attitude:
Cognitive component
Affective component
Intention or action component
Classification of Scales

Single item vs multiple item scale:

 In the single item scale, there is only one item to
measure a given construct.
 In multiple item scale, there are many items that
play a role in forming the underlying construct that
the researcher is trying to measure. This is
because each of the item forms some part of the
construct which the researcher is trying to measure.
Classification of Scales
Comparative vs non-comparative
scales
Classification of Scales
Comparative Scales
 Comparative scales – In comparative scales it is
assumed that respondents make use of a standard
frame of reference before answering the question.
Classification of Scales:
Non-Comparative Scales

 Non-Comparative Scales – In the non-

comparative scales, the respondents do not
make use of any frame of reference before
answering the questions.
Non-comparative scale

Graphic rating scale Itemized rating scale

Non-comparative scales:
Graphic Rating scale

Graphic Rating Scale

This is a continuous scale and the respondent
is asked to tick his preference on a graph.
Non-comparative scales

 Graphic rating scale (Examples)

 Please put a tick mark (•) on the following line to indicate your
preference for fast food.

Alternative Presentation of Graphic Rating Scale –

 Please indicate how much do you like fast food by pointing to the face
that best shows your attitude and taste. If you do not prefer it at all,
you would point to face one. In case you prefer it the most, you would
point to face seven.
Non-comparative scales:
Itemised Rating Scales
Itemized rating scale – In the itemized rating scale, the
respondents are provided with a scale that has a number of brief
descriptions associated with each of the response categories.
There are certain issues that should be kept in mind while
designing the itemized rating scale.
Number of categories to be used
Odd or even number of categories
Balanced versus unbalanced scales
Nature and degree of verbal description
Forced versus non-forced scales
Physical form
Non-comparative scales:
Itemised Rating Scales

Likert scale
 The respondents are given a certain number of items (statements) on which
they are asked to express their degree of agreement/disagreement.
 This is also called a summated scale because the scores on individual items
can be added together to produce a total score for the respondent.
 An assumption of the Likert scale is that each of the items (statements)
measures some aspect of a single common factor, otherwise the scores on
the items cannot legitimately be summed up.
 In a typical research study, there are generally 25 to 30 items on a Likert
scale.
Itemised Rating Scales
Likert Scale

Statements Strongly Disagree Indifferent Agree Strongly

Disagree Agree
i) Shopping online will save time for
customers.
ii) There will be more variety of products
offered online
iii) I will buy products online only when
the price is less compared to traditional
shopping

iv) Shopping online is always risky

v) The information provided about

products/ services in the websites is
accurate and complete
Non-comparative scales:
Itemised Rating Scales

Semantic Differential Scale

 This scale is widely used to compare the images of competing brands,
companies or services.
 Here the respondent is required to rate each attitude or object on a
number of five-or seven-point rating scales.
 This scale is bounded at each end by bipolar adjectives or phrases.
 The difference between Likert and Semantic differential scale is that in
Likert scale, a number of statements (items) are presented to the
respondents to express their degree of agreement/disagreement.
However, in the semantic differential scale, bipolar adjectives or phrases
are used.
 Itemised Rating Scales
Semantic Differential Scale

Positive Negative
Protects the environment Harms the environment
Pollution under control Increases pollution
Effective waste mgt. Ineffective waste mgt.
Conserving energy Wasting energy
Itemised Rating Scales:
Stapel scale

Stapel scale is an attitude measurement

scale that measures a single adjective in
the center of an even number of numerical
values.
Classification of Scales:
Formats of Comparative Scales
Paired comparison scales
Rank order scale
Constant sum rating scale
Q-sort technique
Paired Comparison

 In a paired comparison scale, a respondent

is presented with two objects and is asked to
select one according to whatever criterion
he/she wants to use.
 Suitable when the options are limited
 In reallife the options are always many
Paired Comparison
Z score of paired comparison
Rank order scaling:

 Respondents are presented with several

objects simultaneously and asked to rank
them according to some criterion
 Rank the following soft drinks the most
preferred soft drink as one and so on
 This scale is very commonly used to
measure preferences for brands as well as
attributes.
 The rank order scaling results in the ordinal
Constant sum rating scaling
 In constant sum rating scale, the respondents are
asked to allocate a total of 100 points between
various objects and brands.

 The respondent distributes the points to the various

objects in the order of his preference.
 Allocate a total of 100 points among the various
schools into which you would like to admit your child.
 The more the points you allocate to a school, more
preferred it is to be considered.
 The points should be allocated in such a way that the
sum total of the points allocated to various schools
adds up to 100.
Allocate 100 points to the schools
 Suppose Mother’s International is awarded
30 points, whereas Laxman Public School is
awarded 15 points, one can make a
statement that the respondent rates

 Mother’s International twice as high as

Laxman Public School.
 This type of data is comparative in nature
and could also result in ratio scale
measurement.
 This type of scale is widely used in allocating
weights which the consumer may assign to
the various attributes of a product.
Q Sort Technique
 The Q-sort technique was developed to
discriminate among a large number of
objects quickly.
 This technique makes use of the rank order
procedure in which objects are sorted into
different piles based on their similarity with
respect to certain criterion.
 Suppose there are 100 statements and an
individual is asked to pile them into five
groups, in such a way, that the strongly
agreed statements could be put in one pile,
agreed statements and so on.
 The data generated in this way would be
ordinal in nature.

 The distribution of the number of statement in

each pile should be such that the resulting
data may follow a normal distribution.
 The number of piles need not be restricted to
5.
 It could be as large as 10 or more as the
large number increases the reliability or
precision of the results.
Criteria for good measurement:
Reliability

Reliability
Reliability is concerned with consistency, accuracy
and predictability of the scale.
Methods to measures Reliability
 Test–retest reliability – repeated measurement of the same
person or group
 Split-half reliability – Number of items is randomly divided into two
parts and a correlation coefficient between the two is obtained
 Cronbach’s Alpha – it computes the average of all possible split-half
reliabilities for a multiple item scale. Thumb rule – alpha > .65
 Criteria for good measurement:
Validity
Validity

The validity of a scale refers to the question whether we are measuring what we
want to measure.
Different ways to measure Validity
 Content validity – face validity – subjective judgement by an expert for
assessing the appropriateness of the construct
 Concurrent validity – validity of the new measuring techniques by
correlating them with the established techniques
 Predictive validity – ability of measured phenomena at one point of time to
predict another phenomenon at a future point of time

Sensitivity
Sensitivity refers to an instrument’s ability to accurately measure the variability
in a concept.
Two Types of Measurement Error

 Random errors occur because of random

and inherently unpredictable events in the
measurement process.
 Systematic errors occur when there is a
problem in the measurement system that
affects all measurements in the same way.
 Measurement Error

 T = True value of the observation

 X = Measured value of the observation

Source: Research Methods Knowledge Base

http://www.socialresearchmethods.net/kb/
measerr.php
 What is random error?
 Caused by any factors that randomly affect measurement of the variable
across the sample.
 Each person’s mood can inflate or deflate their performance on any
occasion. In a particular testing, some children may be in a good mood and
others may be depressed. Mood may artificially inflate the scores for some
children and artificially deflate the scores for others.
 Random error does not have consistent effects across the entire sample. If
we could see all the random errors in a distribution, the sum would be zero.
 The important property of random error is that it adds variability to the data
but does not affect average performance for the group.
What is systematic error?
 Systematic error is caused by any factors that systematically
affect measurement of the variable across the sample.
 For instance, if there is loud traffic going by just outside of a
classroom where students are taking a test, this noise is liable to
affect all of the children's scores -- in this case, systematically
lowering them.
 Unlike random error, systematic errors tend to be consistently
either positive or negative -- because of this, systematic error is
sometimes considered to be bias in measurement.
 Reducing Measurement Error

1. Pilot test your instruments.

2. Thoroughly train people taking measurements.
3. Check and double check. If possible take multiple measurements.
4. You can use statistical procedures to adjust for measurement error.
These range from rather simple formulas you can apply directly to
your data to very complex modeling procedures for modeling the
error and its effects. Using multiple forms of measurement helps to
reduce systematic errors.
Copyright © 2009 Pearson Education, Inc.
Is the potential error systematic or random?

 Amtrak passenger trains are most often

late in arriving at their destinations.
 A recipe for grape jelly calls for 4
pounds of grapes. The jelly maker
estimates the 4 pounds of grapes by
standing on a bathroom scale with and
without the grapes. The scale only
shows the weight to the nearest pound.
Identify at least one likely source of random errors
and at least one likely source of systematic errors.

 You need to measure 50 meters for a sprint

workout. You don’t have a tape measure, so
you use a meter stick to measure the
distance.
 You are doing a survey about alcohol use
among college students. You ask students
to write down how many drinks they have
consumed in the last week.