John Stevens

Director of the Centre for Bayesian Statistics in Health Economics (CHEBS)

Author of: What is Bayesian statistics?

John graduated from Brunel University in 1982 with a BSc in Statistics/Mathematics. After graduating, he worked in the pharmaceutical industry for 24 years, including at SK&F Research Ltd, Wellcome Research Laboratories, GlaxoWellcome and AstraZeneca R&D Charnwood.  During this time, he gained experience in the design and analysis of pre-clinical studies and in all phases of drug development, and held a variety of management positions.

In 1998, he assumed a technical role in the development and application of statistical methods to improve the drug development process with a particular interest in Bayesian methods. During this time, he became interested in the application of Bayesian methods in health economics and was actively involved in the creation of CHEBS.

John’s PhD was on latent variable modelling of effectiveness and resource use data in cost-effectiveness analyses of clinical trials.  He moved to the University of Sheffield in November 2006 to provide leadership to CHEBS and further develop his Bayesian interests.  His main activities involve the development and application of Bayesian methods in health technology assessment.  He has served on the NICE Appraisal Committee.

Summary: What is Bayesian statistics?

  • Statistical inference concerns unknown parameters that describe certain population characteristics such as the true mean efficacy of a particular treatment. Inferences are made using data and a statistical model that links the data to the parameters.
  • In frequentist statistics, parameters are fixed quantities, whereas in Bayesian statistics the true value of a parameter can be thought of as being a random variable to which we assign a probability distribution, known specifically as prior information.
  • A Bayesian analysis synthesises both sample data, expressed as the likelihood function, and the prior distribution, which represents additional information that is available.
  • The posterior distribution expresses what is known about a set of parameters based on both the sample data and prior information.
  • In frequentist statistics, it is often necessary to rely on large-sample approximations by assuming asymptomatic normality. In contrast, Bayesian inferences can be computed exactly, even in highly complex situations.