Post on 20-Jan-2016
Chapter 4-Measurement in
Marketing Research
Chapter 4
Measurement & ScalingMeasurement – Process of assigning
numbers to objects to measure amount of attributes.
Rules – A guide.
Scale
Measurement Scales
Nominal - CountingOrdinal – Median / ModeInterval - MeanRatio – Geometric Mean
Difficulty in Measurement
needs to measure ‘internal’ subjective states of individual consumers
constructs: mental abstractions that form the basis for modeling the behavior of larger systems involving many individualsconstitutive definition defines a construct
with other constructs, as in a dictionary
operational definition specifies how a construct is to be measured
Example of concepts: mass , love, strength, advertising effectiveness, consumer attitude, market share
Example of constructs: consumer attitude, advertising effectiveness. Constructs are concepts used for research purpose.
Construct ‘attitude’ may be defined as a learned tendency to respond in a consistent manner with respect to a given object of orientation’ or as’ latent dispositions toward objects’
Constructs that can be measured and quantified are called variables. They…vary!
Operational definition
Height: Measure 1: in inches, with a ruler with the person wearing shoesMeasure 2: in inches, with a ruler w/out the person wearing shoesMeasure 3: Measured by an altimeter or barometerMeasure 4: Measured by the number of hands
Operational definition
Purchase intentions: measured as answer to:
I will definitely purchase Brand x…………… I probably will purchase Brand x……………. I probably will not purchase Brand x………… I definitely will not purchase Brand x…………
Can you think of an alternative operationalization?
Measurement Scales
Nominal - CountingOrdinal – Median / ModeInterval - MeanRatio – Geometric Mean
Scale Types
Scale Types
Scale TypesPlease divide 100 points among each of the
following soft drinks according to your degree of liking for each.
Coke ________Pepsi ________Dr. Pepper ________Sprite ________
TypicalExampl
es
Measures of
AverageScale
BasicComparison
sNominal
Ordinal
Interval
Ratio
Identity
Order
Comparisonof intervals
Comparison of absolutemagnitudes
Male-female User-nonuserOccupationsUniform numbers
Preference for brandsSocial classHardness of mineralsGraded quality of lumber
Temperature scaleGrade point averageAttitude toward brandsAwareness of advertising
Units soldNumber of purchasersProbability of purchaseWeight
Mode
Median
Mean
Geometric
meanHarmoni
c mean
ReliabilityReliability – consistency of measures–
should capture the core concept.Test-retest reliabilityEquivalent forms Internal consistency
Split-halfCronbach alpha
Reliability
Reliability
ValidityFace validity.Content validity.Criterion-related validity – the degree to which
an instrument can predictPredictive – FUTURE - SAT scores; Attitudes.Concurrent - CURRENT
Construct validity – the degree to which a measure confirms a theory based upon the conceptsDiscriminantConvergent
Reliability and Validity
Scaling IssuesGraphic Rating Scales
A graphic continuum typically anchored by two extremes.
Easily constructed and simple to use.Ability to discern fine distinctions.
Itemized Rating ScalesRespondents must select from a limited number of
ordered categories rather than placing a check mark on a continuous scale.
Three Types of Graphic Rating Scales
Uncomfortable
Scale A
Comfortable
Uncomfortable
Scale B
Comfortable
0 10 20 30 40 50 60 70 80 90 100
Neutral
1 2 3 4 5 6 7 8 9 10
Veryvery
Uncomfortable
Veryvery
Comfortable
Scale C
Three Types of Graphic Rating Scales
Exhibit 8.13 Graphic Rating Scale
Rank Order Scale
Respondent judges one item against another
Paired Comparison ScalesAsks a respondent to pick one of two objects from a set based upon some stated criteria.
The respondent makes a series of paired judgments between objects.
Constant Sum Scales
Requires the respondent to divide a given number of points, typically 100, among two or more attributes based on the importance to the person.
To learn about various types of attitude scales.Attitude Measurement Scales
Exhibit 8.14 Comparative Rating Scales
The Semantic Differential
• Begins with the determination of a concept to be rated.
• The researcher selects opposite pairs of words or phrases that describe the object
• Respondents rate on a scale.
• The mean is computed and plotted as a profile or image.
To learn about the various types of attitude scales.
Attitude Measurement Scales
Semantic Differential
Stapel Scale
Designed to measure both the direction and intensity of attitudes simultaneously.
Likert Scales
• A series of statements that express either a favorable or unfavorable attitude toward the concept under study.
• The respondent is asked the level of agreement or disagreement with each statement.
To learn about the various types of attitude scales.
Attitude Measurement Scales
Exhibit 8.9 Likert Scale Example
The Validity and Reliability Concepts
1. respondent characteristics, such as mood
2. situational factors, such as time of day
3. data collection factors, such as the influence of the interviewing method
4. measuring instrument factors, such as flawed survey questions
5. data analysis factors, such as coding error
Sources of measurement error:
total measurement error = systematic error (eS, consistent bias) plus random error (eR)
Estimating Validity
construct validity: compares measurements from the construct of interest and related constructs
content (face) validity: compares measurements with judgments by experts
concurrent validity: compares two different measurements of the same marketing phenomenon at the same point in time
predictive validity: compares measurements at one point in time with predicted measurements at a future point in time
Validity is the extent to which measurement is free from both systematic and random error – its overall accuracy. It is estimated with:
Estimating Reliability
test-retest reliability: compares repeated measurements using the same scaling device under similar conditions on the same subjects
alternative-forms reliability: compares measurements between two equivalent but not identical forms, administered to the same subjects
split-half reliability: compares measurements between equivalent groups of item responses in a multi-item measurement device
Reliability is the extent to which a measurement is free from random errors – its consistency, precision, and predictability. It is estimated with:
A Model of Behavioral Response
cognitive (or belief) component – respondent's awareness of and knowledge about object
affective (or feeling) component – respondent's liking of and preference for object
behavioral component – respondent's intention to buy and purchase behavior
The hierarchy-of-effects model hypothesizes that the buyer's response falls along a spectrum for each component of attitude.
A Model of Behavioral Response (cont.)
Figure 4-4 Model of Behavioral Response
Verbal Rating Scales-Issues to consider
1. overall number of categories
2. odd or even number of categories
3. balanced vs. unbalanced scale
4. extent of verbal description
5. category numbering
6. forced vs. non-forced scales
7. comparative vs. non-comparative scales
8. symmetric vs. asymmetric scales
9. scale direction
10. choice of endpoints
Issues to consider for design of verbal rating scales: