Slide 14.1 Nonmetric Scaling MathematicalMarketing Chapter 14 Nonmetric Scaling Measurement, perception and preference are the main themes of this section. The sequence of topics is Reversing statistical reasoning Additive Conjoint Measurement Multidimensional Scaling Geometric Preference Models
Slide 14.2 Nonmetric Scaling MathematicalMarketing Let’s Turn Things On Their Heads ANOVA: Assume metric data, test for additivity Nonmetric Conjoint: Assume additivity, test for metric data
Slide 14.3 Nonmetric Scaling MathematicalMarketing Additive Conjoint Measurement Algorithmic Steps 0. Inititialize a metric version of y 1. Fit an additive model to the metric version 2. Rescale the metric version to improve the fit 3. If the model does not yet fit, go back to Step 2
Slide 14.4 Nonmetric Scaling MathematicalMarketing The Steps in Detail: Step 0. Copy the dependent variable, y* = y where y contains ordinal factorial data. Since the y i are ordinal, I can apply any monotonic transformation to the y i *
Slide 14.5 Nonmetric Scaling MathematicalMarketing Step 1 – The Usual Least Squares We use least squares to fit an additive model: where X is a design matrix for example,
Slide 14.6 Nonmetric Scaling MathematicalMarketing Step 1 Scalar Representation Looking at row effect i and column effect j we might have Note that since the y ij * are optimally rescaled anyway we do not have to worry about the grand mean. We modify I and j so as to improve the fit of the model
Slide 14.7 Nonmetric Scaling MathematicalMarketing Step 2 – Optimal Scaling Find the monotone transformation that will improve the fit of the above model as much as possible. We modify y * so as to improve the fit of the model
Slide 14.8 Nonmetric Scaling MathematicalMarketing Constraints on H m Assuming no ties, we arrange the original (ordinal) data in sequence: To honor the ordinal scale assumption, we impose the following constraints on the optimally rescaled (transformed) data:
Slide 14.9 Nonmetric Scaling MathematicalMarketing The Shephard Diagram Transformed Data y* Model data Ranked Values Scaled Values
Slide Nonmetric Scaling MathematicalMarketing Minimize Stress to Optimize H m whereis the average of the
Slide Nonmetric Scaling MathematicalMarketing Multidimensional Scaling – Proximity Data Collection The respondent’s job is to rank (or rate) pairs of brands as to how similar they are. Assume three brands A, B and C. Respondent ranks the three unique pairs as to similarity: ___ AB ___ AC ___ BC
Slide Nonmetric Scaling MathematicalMarketing The Geometric Model Brands judged highly similar (proximal) are represented near each other in a perceptual space. d ij - similarity judgment between brand i and brand j
Slide Nonmetric Scaling MathematicalMarketing i = [1 1] j = [2 2] Nonmetric MDS proceeds identically, but the model is a distance model, rather than an additive model
Slide Nonmetric Scaling MathematicalMarketing Brands Repondents Typical element: Consumer i rating brands j and k In individual differences scaling (INDSCAL), each subject’s data are analyzed as follows:
Slide Nonmetric Scaling MathematicalMarketing ABC DE F GHI ABC DE F GHI ABC DE F GHC Subject 1 Subject 2 Group SpaceSubject Weights Subject 1’s SpaceSubject 2’s Space Each subject’s perceptual map might be different:
Slide Nonmetric Scaling MathematicalMarketing Subject i Geometric Models of Preference: The Vector Model Subject i Brand A Brand B Two consumers who disagree Isopreference lines
Slide Nonmetric Scaling MathematicalMarketing Subjects i i Brands A B C D i C B A D The Ideal Brand ModelIs also called the Unfolding Model
Slide Nonmetric Scaling MathematicalMarketing A B C D i i Two consumers who disagree Isopreference circles The Ideal Brand Model in Two Dimensions