Course Level: Intermediate
Course Limit: 20
Format Requirements: This workshop will be able to host a maximum of 20 participants. There is no prerequisite on background knowledge of rank-based methods, however, a basic understanding of statistics and valuation of health (state) is recommended. The format will be 2 hours of presentations given by 2 presenters, followed by 0,5 hours of hand-on experimentation with the acquired knowledge with both presenters guiding the participants. This experimentation will be performed by an interactive software program featuring professional elements of state-of-art scaling methods. It will close with a 0,5 hours classical discussion. Participants will be required to have a laptop with them and be able to use Microsoft Excel 2003 files.
Background: This workshop will focus on rank-based methods and analyses in the field of preference-based health-related quality of life (HRQoL) measurement. Rank-based methods such as Thurstone scaling, discrete choice methods, best-worst scaling and multi-dimensional scaling, will be compared with traditional valuation techniques such as the standard game, time-trade off and the rating scale. Traditional valuation techniques in this field originated from (health) economics. However, these techniques have been shown to be biased and axioms have been shown to be violated. Scaling models are less prone to measurement biases than traditional valuation techniques and allow the investigator to transform data collected from individuals based on simple judgmental tasks (ranking, choices) into group data with interval or cardinal measurement properties (values). In this workshop, participants will use data from real experiments to develop personal experience within the field of health state measurement.
Description and Objectives: Objectives:
- Present an overview of utility/index HRQoL measurement
- Describe the pros and cons of each method
- Let participants develop a (basic) understanding of scaling models
- Let participants gain hands-on experience with collecting and modeling scaling data
The short course will start with an introduction to preference-based health-related quality of life measurement. The conventional approaches will be discussed in detail and evaluated critically on a theoretical level (axioms and assumptions). We will review the current literature and provide a detailed analysis of the weak and strong points of each of the conventional measurement methods.
We will continue the course by introducing scaling methods as an alternative to the conventional HRQoL preference-based methods. A historical overview of the scaling methods and their theoretical background will be provided, along with a critical evaluation of the assumptions.
Once participants have been introduced to the basics of scaling methods, we will go into detail on the various methods, such as pair wise comparisons (BTL model / Thurstone’s law of comparative judgment), Rasch modelling, rank ordering, multi-dimensional scaling, discrete choice modeling, and types of conjoint measurement. We will do this in a structured manner so that participants will not lose track of the broad picture.
When the different scaling methods have been discussed thoroughly we will start with the data examination of real research data. Participants will be led through an Excel developed program that clearly describes the different steps that researchers need to take when handling data from scaling models. Furthermore, they will be confronted with performing analyses themselves and discuss about their results. The workshop will end with a plenary discussion on the advantages and disadvantages of scaling models, and their applicability to health-care problems and research questions.