ISPOR Statistical Methods for the Analysis of Discrete-Choice Experiments
This guideline provides recommended statistical methods for analyzing data from discrete-choice experiments (DCEs), which are used to understand patient and public preferences in healthcare. It explains how to interpret results from such studies.
At a glance
Use when
Analyzing patient preference data from DCEs,Selecting appropriate statistical models for preference heterogeneity,Reporting DCE results in HTA submissions
Avoid when
Conducting non-DCE preference studies,Working with qualitative preference data without discrete choices
Inputs
Discrete-choice experiment (DCE) data, including individual responses to choice tasks, attribute levels, and experimental design structure
Outputs
Estimated preference weights, willingness-to-pay measures, identification of preference heterogeneity, model fit statistics
How it works
Third report from the ISPOR Conjoint Analysis Good Research Practices Task Force (2016) that details good-practice statistical approaches—including conditional logit, mixed logit, latent class models, and hierarchical Bayes—for the analysis and interpretation of discrete-choice experiment (DCE) data in health care.
- HTA domains
- Patient and Social Aspects
- Assumptions
- Respondents make consistent, utility-maximizing choices; the functional form of the utility model is correctly specified; the sample is representative of the target population
- Strengths
- Provides standardized, peer-reviewed guidance on advanced statistical methods; supports robust analysis of patient preferences; addresses heterogeneity through modern modeling techniques
- Limitations
- Assumes familiarity with econometric modeling; may not cover all emerging methods; limited guidance for very small sample sizes or complex adaptive designs
- Also known as
- ISPOR DCE Statistical Methods, ISPOR Conjoint Analysis Task Force Report 3
Questions this answers
- › What statistical models are appropriate for analyzing discrete-choice experiment (DCE) data?
- › How should researchers interpret results from mixed logit or hierarchical Bayes models in DCEs?
- › What are best practices for handling preference heterogeneity in DCE analysis?
- › How can model fit and validity be assessed in DCE studies?
- › What are the advantages and limitations of latent class models in preference analysis?
- › How should researchers report DCE analysis results in line with good research practices?
References & sources
Similar by meaning
- Discrete Choice Experiments in Health Economics: Past, Present and Future
- ISPOR Conjoint Analysis Experimental Design Good Research Practices
- DIRECT Checklist (Discrete Choice Experiment Reporting Checklist)
- ISPOR Roadmap for Patient-Preference Studies in Decision Making
- Distributional Cost-Effectiveness Analysis (DCEA)
Beta record. Based on the original catalogue summary; primary-source enrichment pending.