Customer Marketing Analytics Lecture 3

Customer Marketing Analytics Lecture 3: Measurement and Scaling

• Measurement in Marketing
  • How to measure concepts like price sensitivity, brand attitude, and customer loyalty?
  • Some concepts are easy to observe and measure (e.g., price, income) but others are more difficult (e.g., brand attitude, customer loyalty).
• Types of Scales
  • Comparative scales: compare stimuli to each other (e.g., rank ordering brands).
  • Non-comparative scales: evaluate stimuli independently (e.g., rating a brand on a scale of 1 to 5).
• Multiple Item Measurement Theory
  • Why use multi-item scales (e.g., questionnaires with several questions) instead of single-item scales?
  • To increase reliability by minimizing error.
  • To capture the complexity of constructs by improving validity.
  • Single-item scales are sometimes sufficient for concrete, objective constructs (e.g., age, income, purchase intention).
• Construct Measurement
  • Latent variable: an unobservable concept like "impulsive buying tendency."
  • Indicators: observable items that represent the latent variable (e.g., multiple questions about impulsive buying).
  • By summing scores from multiple indicators, we can estimate the latent variable.
• Scale Evaluation
  • Reliability: consistency of measurement.
  • Validity: whether the scale measures what it intends to measure.
• Reliability
  • Internal consistency: how consistently different scale items are measuring the same construct.
  • Test-retest reliability: how consistent results are over multiple administrations of the same scale.
• Validity
  • Criterion validity: how well the scale predicts related behavior or other variables.
  • Construct validity: aligns with theoretical understanding of the construct.
    • Convergent validity: high correlation with other measures of the same construct.
    • Discriminant validity: low correlation with measures of different constructs.
    • Nomological validity: confirms relationships between the scale and other constructs within a theoretical framework.
• Factor Analysis and Dimensionality
  • Factor analysis: Statistical technique to identify underlying factors from correlated variables (scale items).
  • Dimensionality: Number of underlying dimensions or factors being measured by a scale.
  • Scree Plot: visual representation of eigenvalues (variance explained by each factor) to determine the optimal number of factors.
  • Eigenvalues: Measure of variance explained by each factor.
  • "Elbow rule": Look for the steep drop in eigenvalues to determine the optimal number of factors.
• Communalities
  • How much variance of a single variable is explained by all extracted factors.
  • High communality implies the variable is closely related to the common factor structure.
  • Low communality (< .30) implies the variable is unique and may be a candidate for removal.
• Example: Measuring Consumer Innovativeness
  • The text discusses a study measuring consumer innovativeness.
  • Different techniques were used to assess the scale's reliability and validity.
  • Reliability: Alpha coefficient of .80 demonstrates internal consistency.
  • Validity:
    • The scale demonstrates criterion validity by correlating positively with rock music expertise.
    • It also demonstrates construct validity by correlating with overall trait innovativeness, but not with need for uniqueness (discriminant validity).
  • Example of Testing Reliability and Validity:
    • Deleting items: Ensuring that removing items does not negatively impact internal consistency.
    • Pretesting with experts: Verifying that the scale measures the intended construct.
    • Correlating with other measures: Confirming expected relationships with other variables.