MODELING COMPLEX DECISIONS UNDER MULTIPLE CONSUMPTION SCENARIOS Open Access

Abstract

The objective of this dissertation is to understand complex consumer decisions under multiple consumption scenarios accounting for the interdependencies across product categories within the consumption portfolio in order to capture a full model of consumer behavior. The context includes multi-category/multi-channel media choice (Essay 1), and broader applications of consumption in seemingly disparate product categories (Essay 2).

Specifically, Essay 1 focuses on predicting media choice and time allocation, taking into account interdependencies between traditional and new media under a utility- maximizing framework. Using a rich database of individual-specific media activity diaries, it suggests that accounting for media interdependencies is extremely important and generates unique insights on consumer-level media switching, media multiplexing, potential sources of substitutability/complementarity resulting from media attention and penetration, and consumer heterogeneity often ignored in aggregate data.

Essay 2 takes a first step in modeling difficult-to-observe psychological processes that govern consumer decision making by examining consumption across seemingly disparate categories. This research proposes a hierarchical multinomial processing tree model to empirically examine the driver, which is defined as the "latent trait", which governs consumer choices across five seemingly disparate product categories: media consumption, automobile purchases, financial investments, soft drinks and cell phone plans through a dataset consisting of 5,000 consumers in the United States. Essay 2 further investigates how consumer behavior systematically varies from one category to another and suggests new approaches to segment and profile consumers based on behavior across multiple categories. Finally, by comparing the latent trait approach with the latent class approach, it contrasts discrete and continuous representations of consumer heterogeneity and discusses related substantive and empirical issues.

MODELING COMPLEX DECISIONS UNDER MULTIPLE CONSUMPTION SCENARIOS

Chen Lin
B.COMPUTING (Hon.), National University of Singapore, 2007
Advisors: Douglas Bowman, Ph.D.
Sandy D. Jap, Ph.D.
An abstract of
A dissertation submitted to the Faculty of the
James T. Laney School of Graduate Studies of Emory University
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy in Business
2012

Table of Contents

Table of Contents: Chapters
1 Introduction...
2 ESSAY 1: Media Multiplexing Behavior: Implications for Targeting and Media Planning...4

2.1 Introduction...4
2.2 Literature Review...9
2.3 The Empirical Model...14

2.3.1 Optimal Time Allocation...18
2.3.2 Unobserved Heterogeneity...21

2.4 Data...23

2.4.1 Media Consumption Data...24
2.4.2 Media Channel Switching...23
2.4.3 Time-varying Property...26
2.4.4 Consumer Heterogeneity...27

2.5 Results...28

2.5.1 Model Covariates and Constants...29
2.5.2 Model Selection and Goodness-of-Fit Measures...29
2.5.3 Baseline Constants...30
2.5.4 Time-of-Day and Day-of-Week Effect...32
2.5.5 Effects of Household and Individual Social-Demographics on Baseline Utility...32
2.5.6 Effects of Household Attention Level and Media Penetration on Baseline Utility...33
2.5.7 Satiation Estimates...34
2.5.8 Unobserved Heterogeneity...36
2.5.9 Robustness of the Time Period Selection...36

2.6 Media as Complements and Substitutes...37
2.7 Counterfactual Analysis...43
2.8 Discussion...46

2.8.1 Limitations...47
2.8.2 Implications for Management...47
2.8.3 Directions for Future Research...48

3 ESSAY 2: Modeling Consumer Decisions Across Seemingly Disparate Categories: A Latent-Trait Approach...49

3.1 Introduction...49
3.2 Literature Review...57
3.3 The Empirical Model...61

3.3.1 Person-Level Model...62
3.3.2 Prior Distribution...64
3.3.3 Hyperprior Distribution...64
3.3.4 Data Augmentation for the Gibbs Sampler...65
3.3.5 The Gibbs Sample...67
3.3.6 Implementation...71

3.4 Data...71
3.5 Results...74

3.5.1 Model Selection and Goodness-of-Fit Measures...74
3.5.2 Parameter Estimates...75
3.5.3 Latent Trait versus Latent Class and Nested Logit...76

3.6 Segmentation and Profiling Analysis...78
3.7 Discussion and Conclusion...79

4 References...81

Table of Contents: Tables
TABLE 1 : Media Switching Matrix...87
TABLE 2 : In-Sample Hit Rate Comparison...87
TABLE 3 : Estimates for the Proposed Model...89
TABLE 4 : Out-of-Sample Hit Rate Comparison...91
TABLE 5 : Marginal Rate of Substitution at the Point of Zero Consumption...91
TABLE 6 : Counterfactual Results from Print Media Exit in Los Angeles...92
TABLE 7 : Notations and Variables for the Latent Trait Model...93
TABLE 8 : Summary Statistics for Product Category Consumption...94
TABLE 9 : Parameter Estimates from the Latent Trait Model...94
TABLE 10 : Model Fit and Comparison...95
TABLE 11 : Parameter Estimates from the Latent Class Model...95
TABLE 12 : Parameter Estimates from the Latent Trait Segmentation...97
TABLE 13 : Out-of-Sample Prediction Results...97

Table of Contents: Figures
FIGURE 1 : Growth in Joint Consumption of Television and Computer Media...98
FIGURE 2 : Summary Statistics for Single and Multimedia Consumption...98
FIGURE 3 : Multinomial Process Tree Representation of the Latent Trait Model...99
FIGURE 4 : Multinomial Process Tree Representation of the Latent Trait Model with Augmented Data...100
FIGURE 5 : Sample Data Structure...101

Table of Contents: Appendices and Supplementary Materials
A1.FIGURE 1 : Media Choice by Time-of-Day (Aggregated across Households)...102
A1.FIGURE 2 : Average Share of Media Usage by Time-of-Day at the Household Level...102
A1.TABLE 1 : Media Choices by Day-Of-Week (Aggregated across Households)...103
A1.TABLE 2 : Results from Single Discrete-Continuous Model of Demand...104
A1.TABLE 3 : Results from Our Main Model (Half-hour Time Interval)...106
A1.TABLE 4A: Marginal Rate of Substitution at the Point of Zero Consumption: Baseline Constants...109
A1.TABLE 4B: Marginal Rate of Substitution at the Point of Zero Consumption: Attention for Computer...110
A1.TABLE 4C: Marginal Rate of Substitution at the Point of Zero Consumption: Attention for Television...111
A1.TABLE 4D: Marginal Rate of Substitution at the Point of Zero Consumption: Attention for Radio...112
A1.TABLE 4E: Marginal Rate of Substitution at the Point of Zero Consumption: Attention for Print...113
A1.TABLE 4F: Marginal Rate of Substitution at the Point of Zero Consumption: Home PC Ownership...114
A1.TABLE 4G: Marginal Rate of Substitution at the Point of Zero Consumption: Cable TV Ownership...115
A1.TABLE 4H: Marginal Rate of Substitution at the Point of Zero Consumption: Workplace PC Use...116
A1.TABLE 5 : Counterfactual Results from Print Media Exit in New York City...117

About this Dissertation

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Laney Graduate School
Department	Business
Degree	PhD
Submission	Dissertation
Language	English
Research Field	Business Administration, Marketing
Keyword	Underlying Processes Seemingly Disparate Categories Latent Trait New Media Media Consumption Multi-Category/Multi-Channel Choice
Committee Chair / Thesis Advisor	Bowman, Doug, Emory University
Committee Members	Lewis, Mike, Emory University Tripathi, Manish, Emory University

Last modified

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	MODELING COMPLEX DECISIONS UNDER MULTIPLE CONSUMPTION SCENARIOS ()	2018-08-28 16:33:15 -0400	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions