Docstoc

general concepts of measurement

Document Sample
general concepts of measurement Powered By Docstoc
					General Concepts
of Measurement
  Chapter 9
  Measurement and Analysis
• We have a problem involving “stuff”
• We try to measure “stuff” about “stuff”
• We use models (implicitly or explicitly) to
  conceptualize, find analogies, to figure out
  the “stuff”
• We analyze the “stuff” that measures other
  “stuff” and use it to predict or define other
  “Stuff”
   Why Understanding Modeling
   Basics is So Very Important
• Sensitizes managers to start thinking about the
  most important variables
• Forces researchers to Evaluate, Identify and
  Select the right variables
• Conceptualizes the notion of “What are the
  relationships between variables”
• Produces in a more structured approach to the
     •   Research activity
     •   Collecting of data
     •   Organization of information
     •   Organization of data
   Models Are Important When
 Preparing Management Reports
• Are important management concepts and
  propositions included in the model?
• Are assumptions clear?
• Are limitations stated?
• Does the model provide predictions?
• Does the model provide explanations?
• Does the model tell us what to do? (Are
  normative guidelines given)
• Is the model easily quantified?
• Can logical deductions based on the model lead
  to specific outcomes?
  5 Managerially Useful Rules for
      Models and Modeling
1. Models must be simple to understand and use
2. A model must be “robust” (not vary wildly)
3. A model must be easy to communicate with
   (descriptive, iconic, symbolic, predictive,
   normative, mathematical, statistical,
   estimation-inference)
4. A model must be adaptable to varying
   products, people or situations
5. A model must be complete on important details
    Areas of Measurement Definition
•    Model: Define what is to be
     measured
•    Measure: Decide how to make
     measurements
•    Collect: Decide how to conduct the
     measuring operations
•    Analyze: Determine how to analyze
     the resulting data
  Estimate New Product Trial
• Product Positioning   •   Family Brand
• Media Impressions     •   Consumer Promotion
• Consumer              •   Product Satisfaction
  Promotional Recall    •   Category Usage Rate
• Category Interest     •   Relative Price
• Distribution Extent   •   Relative Purchase
• Packaging                 Frequency
 What is Measurement Quality
Criteria to measure quality of the
   models:
1. Is it Valid?
2. Is it Useful?
3. Is it Repeatable (reliable)?
    When “More” is too much
• Model builders sometimes include so many
  variables that the basic structure of the model
  is buried, input data costs escalate, and
  confidence in them is lost.

• Models can be excused from not representing
  reality perfectly and will benefit… if they are
  simple enough for the managers to
  understand.
     Building Blocks for Models
• Concept       An abstraction formed by generalization about
  particulars
  Examples: Mass, strength, love,
  advertising effectiveness, customer attitude, price elasticity
     Building Blocks for Models
• Concept       An abstraction formed by generalization about
  particulars
  Examples: Mass, strength, love,
  advertising effectiveness, customer attitude, price elasticity
• Construct An Observable, Measurable Concept (conscious
  inventions of researchers)
  Example: Customer Attitude is a concept, but also can be observed
  and measured and is related to other constructs
      Building Blocks for Models
•   Concept        An abstraction formed by generalization about particulars
    Examples: Mass, strength, love,
    advertising effectiveness, customer attitude, price elasticity
•   Construct     An Observable, Measurable Concept (conscious inventions
    of researchers)
    Example: Customer Attitude is a concept, but also can be observed and
    measured and is related to other constructs
•   Variables   The constructs that researchers study. Variables can be
    measured and quantified.
    Example: Customer attitude can be measured as a function of attitude
    toward attributes of the product or service, importance of those attributes,
    normative beliefs, and motivations to comply to normative beliefs
    Example: Purchase intentions can be measured as a function of attitudes
    and enabling factors (income), or directly measured as intention to perform
    a future action.
       Building Blocks for Models
•   Concept       An abstraction formed by generalization about particulars
•   Construct     An Observable, Measurable Concept (conscious inventions of researchers)
•   Variables     The constructs that researchers study. Variables can be measured and
    quantified.

• Operational Definitions = a set of instructions that
  assigns meaning to a variable by specifying what and
  how is to be measured.
    Example: “Height”

    How would we measure height?
    Example: “Intention to Purchase”

    How would we measure purchase intentions?
       In Class Assignment
•   In your Group, Take 10 Minutes to
    Discuss how to measure:

1. Height
2. Purchase Intention

Develop the following:
  Concept, Construct, Variables, Scale
       Building Blocks for Models
•   Concept        An abstraction formed by generalization about particulars
•   Construct      An Observable, Measurable Concept (conscious inventions of researchers)
•   Variables      The constructs that researchers study. Variables can be measured and
    quantified.

• Operational Definitions = a set of instructions that
  assigns meaning to a variable by specifying what and
  how is to be measured.
    Example: “Height”
     – Measured in inches with a precision ruler with shoes
     – Without shoes
     – By altimeter, barometer, GPS
     – Number of hands
     – ???
      Building Blocks for Models
•   Operational Definitions = a set of instructions that assigns meaning to a
    variable by specifying what and how is to be measured.

    Example: “purchase intentions”
     – Defined as the answer to the question
       Please indicate your intention to purchase Windex brand window cleaner the
       next time you purchase a window cleaning product:

       I will definitely purchase Windex
       I will probably purchase Windex
       I will probably not purchase Windex
       I will definitely not purchase Windex

     – Measured as:
         P(x) = (Attitudes toward class of product) x (Beliefs about brand X)
    Building Blocks for Models
• Proposition: an explicit statement of the
  relationship between variables, the variables
  influencing this relationship and the form of the
  relationship.
      • Proposition are used to form models.
Example:    Advertising  Sales

            Advertising  Awareness  Knowledge  Sales
            Models: An Example
                                      This model shows the belief
                                        that sales will increase
                                        with increases in
                                        advertising up to a point,
                                        the showing diminishing
                                        returns

Outcome: Sales revenues
Variable: Advertising expenditures
Relationship: S-curve
While this model could be described explicitly by a mathematical
  formula, some models are understood implicitly
    Building Blocks for Models
• Proposition: an explicit statement of the
  relationship between variables, the variables
  influencing this relationship and the form of the
  relationship.
      • Proposition are used to form models.

• Model: a set of prepositions that are linked together
  to explain a system or process.
• Cause and effect: a relationship that is established
  if three conditions are met
             » Concomitant variation
             » The “cause” occurs first
             » Absence of competing explanations
     Measurement Concepts
• Measurement: “the assignment of numbers
  to objects to represent amount or degrees of
  a property possessed by all of the objects”
                                Torgeson, 1958
• Features of the real number series: order,
  distance and origin.
• Types of scales: nominal, ordinal, interval
  and ratio.
                           Scales
• Nominal scale: numbers serve as labels
       • Computations: modal value and contingency tests

• Ordinal scale: the elements are ranked
       • Computations: modal value, median and overall index rank

• Interval scale: a constant unit of measurement and
  properties of order and distance, but the zero point is
  arbitrary. (Fahrenheit and Celsius)
       • Computations: modal value, median, average
       • Comparisons can be made only in terms of differences, but not in
         terms of proportions

• Ratio scale: permits all arithmetic operations, possesses a
  unique zero point and scale values correspond to equal ratios
  among the entities being measured.
  There is Variation in Measurement…

• Situation: US Army just completed measures of
  financial health using the following questions:




When the survey is re-administered 6 months later, the results
for each individual change… Why?
Why is There Variation in Measurement?

 • True differences in the characteristic of a property at different points
   in time
 • Other relatively stable characteristics of individuals that affect scores
 • Transient personal factors (health, mood, immediate prior events)
 • Situational factors (occasion, usage situation)
 • Variations in administration of measuring instrument
 • Sampling of items included in instrument (poor items in survey)
 • Lack of clarity of measuring instrument (poorly worded questions)
 • Mechanical factors (ex: lack of space to record response)
 • Factors in the analysis (ex: scoring, tabulation)
 • Chance (ex: guessing the answer)
    How to Assess Validity and
    Reliability of a Measurement
• Does the scale really measure what we are
  trying to measure?

• Do subject responses remain stable over time?

• Are respondents consistent over the scales that
  are measuring the same thing?
        Validity and Validation
• Content validity: does the scale represent all
  dimensions of the property measured?
             » Face validity: Test with non-researcher (family, friends)
             » Logical validation: Intuitive, Common Sense
• Criterion validity: can better decisions be made with
  this construct? This “pragmatic validation” has two
  dimensions:
             » Predictive validity: Does the measure predict
             » Concurrent validity: Does a second measure correlate?

• Construct validity: does the construct work? Why and
  what deductions can be made?
             » Convergent validation (correspondence in results between
               independent methods)
             » Discriminant validation (the extent to which a measure is unique)
             » Nomological validation (understanding a concept)
                    Reliability
Reliability = Consistency of results over groups or over the
     same individual at different times.
Methods to measure a scale’s reliability:
1.   Test-Retest (a minimum of two weeks between
     measurements is required)
2.   Alternative forms (equivalent forms of measure
     administered to the same sample)
3.   Internal consistency (alternatives are formed by
     grouping variables; ex: split-half reliability)
        Concluding Comment
Reliability affects can affect:
   – The validity of a study
   – The ability to show relationships between variables
   – The making of precise distinctions among individuals
     and groups

				
DOCUMENT INFO
Shared By:
Stats:
views:165
posted:3/12/2008
language:English
pages:28