Universität Hamburg – Doktorandenkurs: Discrete Choice Experiments

Institution: Graduate School der Fakultät Wirtschafts- und Sozialwissenschaften der Universität Hamburg

Studiengang: Promotionsstudiengang

Dozent/in: Thorsten Teichert

25.11.2012, 11.00 – 19.00 Uhr
26.11.2012, 10.00 – 18.00 Uhr

Lehrveranstaltungsart: Workshop

Raum: n.a.

Semester-Wochen-Stunden: 1

Unterrichtssprache: englisch

Zielgruppe: BWL/VWL/Sozialwissenschaften/Sozialökonomie


Seminar Objectives:

Discrete choice experiments are a research method which is widely applied in the areas of management, consumer research and economic evaluation to estimate metric preference functions from choice input data. The goal of this research workshop is to give insights into newest developments of this method. Topics covered include: A tutorial on traditional conjoint – Choice context effects – Latent class – Hierarchical Bayes – Best-Worst Scaling

The workshop is oriented towards the exchange of research ideas, advanced concepts as well as preliminary and final research findings. There will be time for open discussion related to your research work.


DAY 1:
A tutorial on conjoint experiments: We will trace the versions of conjoint from its analytic roots to choice-based conjoint that is the standard in business. We will show how to use project individual-level results to a population using simulators.

Experimental Design techniques: Deriving designs from orthogonal arrays is an elegant way to generate conjoint design. You will learn how to quickly easily generate a design that is, if not optimal, robustly good for a particular application.

DAY 2:
Segmentation and Latent class: Latent class models are a finite mixture application for discrete choice experiments. They account for heterogeneity by positing a number of classes of “latent” respondents reflecting groups of people with similar values same way.

Hierarchical Bayes: Hierarchical Bayes has revolutionized the assessment of value functions. We examine why it works and explore other applications of Bayesian estimation in preference assessment.

Context effects: We examine the lessons from research from the last 20 years on the asymmetric dominance effect. What has been learned about the choice process? How large is the effect? How can the effect be parametrically modelled? What should be done to account for the effect in commercial studies?

Methods and Applications Revisited: As a synopsis of our workshop, we compare the different approaches and their contributions in specific application contexts. We investigate the emerging possibilities of applying Best-Worst Scaling Techniques. We discuss practical issues as survey and software issues.

Anmeldung über die Graduate School der Fakultät Wirtschafts- und Sozialwissenschaften der Universität Hamburg (Dr. Ulf Beckmann)