Marketing Research: CH 7: Survey Measurement Scales

Learning Objectives
Examine question response formats commonly used in marketing research.
Distinguish among the different levels of measurement.
Explain four types of measurement scales used by marketing researchers.
Explain validity and reliability in measurement.
Question-Response Formats
Three formats:
Open-ended
Aided, unaided
Categorical
Dual, multiple
Scaled
Natural, synthetic
Question-Response Formats
Three formats:
Open-ended
Aided, unaided
Categorical
Dual, multiple
Scaled
Natural, synthetic
Open-Ended Response Format:
 - Subject responds in own words
 - Good in exploratory - no limits on answers
 - Unaided open-ended format:
 - No prompt or probe - just answer
 - Aided open-ended format
 - Response probe as follow-up question
Open-Ended Response Format:
– Subject responds in own words
– Good in exploratory – no limits on answers
– Unaided open-ended format:
– No prompt or probe – just answer
– Aided open-ended format
– Response probe as follow-up question
“Are there any other brands you remember?”
This helps you form open-ended questions to help discover:
 - Situation-what do you need the car for?
 - Problem- what are you going to while the car is in getting service? How much will it cost?
 - Implication- how is this really impacting you? Had to take the bus,
 - Need payoff- whats in it for them, the client not you.This helps you form open-ended questions to help discover
This helps you form open-ended questions to help discover:
– Situation-what do you need the car for?
– Problem- what are you going to while the car is in getting service? How much will it cost?
– Implication- how is this really impacting you? Had to take the bus,
– Need payoff- whats in it for them, the client not you.This helps you form open-ended questions to help discover
Categorical Response Format:
– Answer options are provided
– Dual choice question – Respondent selects from two alternatives, yes or no
– you use this to force an answer or use as a pivit question so you know where to go next with your conversation.

– Multiple-choice question- Several options, respondent chooses one:
– Check all that apply:
– Looks like multiple-choice, but really dual for each
– Each has, in effect, yes/no answer
– If you check one and not the other its like a bunch of dual choice questions, yes or no
– All of these give a simplified answer. WE need to probe more.

Dual
Categorical-true/false, yes/no,
Metric/Scaled Response Format:

– Answer is a number expressing a quantity
– Specific number, or scale
Type 1: Natural metric/scaled-response:
– Give actual number, such as age, number of visits
– Its a response that can be measured
– “How many times did you use this?”
– “On a scale of 1-10 how did you enjoy this movie?”

Metric/Scaled Response Format:

Type 2: Synthetic metric/scaled-response:
– Artificial number measures property
– How do you measure loyalty? What does loyalty mean or look like? Its vague
– Indicate level of satisfaction from 1 to 10

Choosing a Question-Response Format
– Nature of property being measured
Subjective, varying degrees, choose scaled
Yes/no, simple choice, category
Previous research studies
Make sure you can compare to earlier studies
Better to borrow than invent
Choosing a Question-Response Format:

– Nature of property being measured
– Previous research studies
– Ability of the respondent: age level of language, ESL
– Scale level desired- what do I really want to know

Maria Slide
Writing a good questionnaire is very difficult, draft after draft
You can only downgrade a questionaire, you can always go more simple but not more complex. Aim for 1-level higher than you think you need.
Choosing a Question-Response Format:
– Nature of property being measured
– Subjective, varying degrees, choose scaled
– Yes/no, simple choice, category
– Previous research studies
– Make sure you can compare to earlier studies
– Better to borrow than invent
Choosing a Question-Response Format [cont’d]:

– Ability of the respondent
– Consider literacy, age, etc.
– Scale level desired
– Scaled response generally more informative
– Scales better for higher-level analysis
– Yes/no quite limited

Basic Concepts in Measurement:
– Measurement gathers if property possessed
– If so, measure quantity
How mjuch of a property is being posessed by an object:
Types of Properties:
– Objective properties are physically verifiable, e.g., height, gender
– Usually preset categories
Basic Concepts in Measurement [cont’d]
Typically measure properties
Often called attributes or qualities
Properties = distinguishing features/characteristics
Preferred brand, rating, demographics
Measurement helps describe marketing phenomena
Why the Measurement Level of a Scale is Important
– Determines information collected
– Affects which analysis should or should not be performed
– This then affects what many or many not be said about the property being measures
– use highest appropraite level
Intensity Continuum: THE LIKERT SCALE

A metric synthetic scale their is no subjectivity

- Interval scales capture range of properties
 - Extremely negative to extremely positive
 - Metric scale uses neutral
 - Instead use a dual choice tick off what implies and tick off if not applicable
Intensity Continuum: THE LIKERT SCALE

A metric synthetic scale their is no subjectivity

– Interval scales capture range of properties
– Extremely negative to extremely positive
– Metric scale uses neutral
– Instead use a dual choice tick off what implies and tick off if not applicable

Insert table 7.4 from page 242
Types of Propertie:

– Subjective properties:
– Cannot observe directly; mental constructs
Attitudes to brands, lifestyles
Need a “continuum of intensity”
Typically rating scale formats e.g., 1 to 5

Scale Characteristics:
– Description: unique label = agree/disagree
– Order: relative sizes of descriptors
– Greater than, less than
– Distance: differences between descriptors
– Family has three cars, one more than two-car family
– Origin: where true zero point exists
– Consumed no soda yesterday
Levels of Measurement Scales:
– Important to recognize this in scales
– Categorical
– Nominal
– Ordinal
– Metric
– Interval
– Ratio
Categorical Scales:
– Nominal =use only labels
– Gender, religion, brand last purchased
– Description is only characteristic
– Ordinal = rank order respondents’ choices
– Asked to rank brands first, second, etc.
– E.g. Have coffee once a day, several times a day
– Only relative size differences
Metric Scales:
– Interval = known distance between descriptors
– Assume each descriptor one unit away from next
– Somewhat important versus very important
– On a continuum, equal distances apart
– Ratio = true zero origin
– Number of purchases made, miles traveled
– Years of college education
Scale Measurement Level is Important:
– Level of measurement critical
– Determines what information collected
– What can be said based on that information
– Nominal scales crudest, ratio scales richest
– Nominal only identifies a property
– Ratio allows statistical analysis, extent of difference
– Generally, use scale at highest level
Synthetic Metric Scales:
– Marketing researchers often seek subjective views
– Attitudes, opinions, perceptions, intentions
– All unobservable, must measure
– Tailor-made scales for the purpose
– Often called workhorse scales of research
Likert Scale:

– Symmetric, with neutral in the middle
– Same number of units to left and right of neutral
– Use plain statements, let them agree/disagree
– Often include strongly, somewhat agree/ disagree
– Various versions of it

Issues with Synthetic Scale-Response Formats:

– Middle responses neutral in Likert scale
– Some prefer to leave that out
– Forces respondents to be for or against
– Symmetric, or one-way?
– Often only want positive, e.g. Importance
– If in doubt, pre-test

Lifestyle Inventory, Semantic Differential Scales:
 - Lifestyle inventory measure psychographics
 - Activities, interests and opinions
 - Attitudes to work, leisure, and purchases
 - Series of Likert scale questions
 - Semantic differential scale
 - Series of bipolar adjectives describing object
 - Respondents show impressions along continuum
 - Works well for images:
Lifestyle Inventory, Semantic Differential Scales:
– Lifestyle inventory measure psychographics
– Activities, interests and opinions
– Attitudes to work, leisure, and purchases
– Series of Likert scale questions
– Semantic differential scale
– Series of bipolar adjectives describing object
– Respondents show impressions along continuum
– Works well for images: “friendly – unfriendly”
insert Table 7.7 p. 246
Graphic Scales page 248 Table 7.8
-easy to interpret
thermonistors get used
face
Graphic Scales page 248 Table 7.8
-easy to interpret
thermonistors get used
face
More Synthetic Metric Scaled-Response Formats:

 - Many formats
 - May use:
 - Graphics 
 - Numbers
 - Percentages
More Synthetic Metric Scaled-Response Formats:

– Many formats
– May use:
– Graphics
– Numbers
– Percentages

Non
-Symmetric scales – “balanced” having an equal amounts of postiive and negative postions
Insert Figure 7.4 p 247
Non-symmetric scale – unbalanced scales that measure positive degrees of options of a construct.
Looking for a person’s opinion?
– use a metric and more like synthetic
Looking at product usage?
– use non synthetic metric scale and not just categories
Issues in the Use of Synthetic Scale- Response Formats:
– include neutral responses?
– symmetric, unbalanced or one-way?
– If in doubt should pre-test
Cultural differences
– Extremists versus moderates
– page 250
Nonsymmetric Synthetic Scales:

– One-way labelled scale
– Goes in one direction only
– Not at all important to extremely important
– Useful when negative makes no sense
– N-point scale: pick number to measure
– Usually 5 to 10 points on scale
– One =”not friendly”, five =”extremely friendly”
– Anchored at both ends often

Reliability, Validity of Measurements:

– Reliable measure:
– Similar answers to similar questions
– Too much variance, question is unreliable
– Valid measure:
– Truthful, measures what it should
– Problems: lying, poor memory, misconception
– Face validity: looks like it measures as it should

What Scale to Use When:

– Constructs = standard marketing concepts
– E.g., brand image
– Should always use highest level scale possible
– Yes-no scales appropriate at times
– Summated scale adds up several measures

What Scale to Use When:
– Operational Definition – when a researcher specifies the procedure to measure a property of an object
– constructs – standard marketing concepts
– – table 7.9 p. 252
– Should always use highest level scale possible
– Summated scale – adds up several measures
Reliability and Validity of Measurements
– Ideally should be reliable and valid
– reliability measure – similar answers to similar questions
– Too much variance, question is unreliable
– Valid measure – Truthful; measures what it should
– Problem include lying, poor memory
Measuring Atitudes:
1. ranking – gives you people’s perception of how they look at the world
2. rating – need a higher eduction of surveyers
3. sorting
4. choice
Constant-Sum Example
– Divide 100 points among each of the following brands according to your preference for the brand:
Sorting:
– respondents indicate their attitudes or beliefs by arranging objects on the basis some attribute or construct

Paired Comparisons:
– respondents are presented with two object at a time and asked to pick the one they prefer
– good way of doing value assessment: What’s more important to you, being popular or being successful