Bayesian Methods

Bayesian methods have been used to produce new utility values for the SF-6D. Kharroubi et al (2005) suggest that there are five principal advantages in using a non-parametric Bayesian approach over conventional parametric random effects modelling:

• Bayesian methods do not require additivity in the utiltiy function.
• Bayesian methods violate monotonicity less than standard methods.
• Bayesian methods allow full health to be valued at one without requiring a worse fit to the data.
• Bayesian methods can reflect that individuals have less disagreement over good health state values than for poor states.
• Bayesian methods can reflect that individual valuations of any state form a skewed distribution.

Reference:

Kharroubi, S. A., Brazier, J. E., Roberts J. and O'Hagan, A. (2005). Modelling SF-6D health state preference data using a nonparametric Bayesian method. CHEBS. Submitted to Journal of Health Economics.