Essays in Consumer Preference Measurement

Open Access
Kim, Hye-jin
Graduate Program:
Business Administration
Doctor of Philosophy
Document Type:
Date of Defense:
March 14, 2013
Committee Members:
  • Min Ding, Dissertation Advisor
  • Rajdeep Grewal, Committee Member
  • Mosuk Chow, Committee Member
  • Fuyuan Shen, Committee Chair
  • Daniel Kifer, Committee Member
  • consumer preference measurement
  • conjoint analysis
  • speech analysis
This dissertation aims to propose new frameworks and marketing research methods to get a better understanding of consumer preferences. It is composed of four chapters, starting from a brief introduction in Chapter 1. Chapter 2 is an essay which proposes a new holistic framework of preference, PIE, which incorporates social influence in the formation of consumer preferences, as well as a measurement method to demonstrate the framework. I propose and test an incentive-aligned approach, a group-sourced mechanism, which mimics a real life consultation of a consumer with her friends in purchase decision making. The results provide support for the PIE framework, including superior predictive power. Chapters 3 and 4 explore the possibility of using human speech data to better understand consumers. Audio data are ubiquitous and contain rich information, and with the recent advancement of technology, marketers are able to collect, store, and analyze them. Because using the human voice to understand consumers is a relatively new attempt being done in the marketing field, in Chapter 3 I aim to identify opportunities for marketers by reviewing what has and is being done in other areas. In Chapter 4, I demonstrate an application of speech analysis in marketing. Marketing survey methods such as customer satisfaction surveys usually only consider the semantic aspect of responses, without accounting for uncertainty in such responses. I propose the use of human voice as an alternative data collection format that allows uncertainty to be inferred based on pitch, intensity, and temporal features extracted from a respondent’s voice. I show that a modified construct using the inferred uncertainty to reweight different survey items predict people’s behavior better than using the unmodified construct.