Three value estimates of input variables and health manpower projections

The introduction of variability and stochastic processes in health manpower projections may help health planners cope with the inherent problem of uncertainty in the future. However, limited data and the complex nature of health manpower make it very difficult or even impossible to estimate the probability distribution of input variables. In this paper, two statistical methods are discussed and compared for approximating a probability distribution based on imperfect data. The common feature of the two methods is that they use minimum, maximum, and most likely values, which can be estimated by people with little knowledge of statistics and probability. In addition, the methods can be used to analyse variables with a symmetrical as well as non-symmetrical probability distribution. An example is provided of the application of the methods to health manpower projections in China.