Distributions

In statistics, distributions describe the spread of values for a particular characteristic measured in a population. For example, while we may know the mean age of a population, most individuals will be older/younger than this mean value – and not necessarily in a uniform manner.

Parameter distributions are frequently defined using a mean and standard deviation (for a normal distribution) or by “shape” and “scale” parameters for more complex distributions. Common distributions in economic modeling are symmetrical (such as the normal distribution; for age, intervention effectiveness) or skewed (like gamma or lognormal; for ratios or non-negative values like costs). Distributions for mutually exclusive probabilistic outcomes, such as binomial, Poisson, beta, or Dirichlet, are used for input parameters representing probabilities. Specifying model input parameters as distributions, rather than fixed values, enables probabilistic sensitivity analysis, which quantifies and assesses the uncertainty surrounding the model’s outputs (e.g. incremental cost-effectiveness ratios).