The analysis of rank order preference data based on Bradley-Terry type models


Type Research Project

Duration since Jan. 1, 1997

  • Mathematical Methods in Statistics AE (Former organization)

Tags

Press 'enter' for creating the tag
  • Dittrich, Regina (Former researcher) Project Head
  • Hatzinger, Reinhold (Former researcher)
  • Katzenbeisser, Walter (Former researcher)
 

Abstract (German)

statistische Analyse, Log-lineare Modelle von entsprechend definierten Kontingenztafeln


Abstract (English)

The intention of this work is to demonstrate how to analyse complete rank ordered data by means of Bradley-Terry type models. The main idea is to transform the rankings into paired comparison data, which can be modelled within a log-linear framework. This approach allows to estimate object-specific parameters, which in the marketing context can be interpreted as attractions of the analysed objects. Additionally, the effects of subject-specific covariates on the attractions can be estimated by the presented models. The estimated subject parameters offer a statistically motivated approach for customer segmentation. The outlined methodology will be applied to preference judgements within the Austrian daily newspaper market. It is shown that certain socio-economic characteristics of the customers have significant influences on their preference structure.

Publications

Journal article

2015 Grand, Alexandra, Dittrich, Regina, Francis, Brian. 2015. Markov models of dependence in longitudinal paired comparisons: anapplication to course design. Advances in Statistical Analysis 99 (2): S. 237-257. (Details)
  Grand, Alexandra, Dittrich, Regina. 2015. Modelling assumed metric paired comparison data: Application to learning related emotions. Austrian Journal of Statistics 44 (1): S. 3-15. open access (Details)

Classification

  • 1113 Mathematical statistics (Details)

Expertise

  • Bradley-Terry model