Abstract
This paper contains a description of the use of critical multiplism to evaluate the advantages and disadvantages of random sampling and deliberate sampling for heterogeneity, according to the ability of each type of sampling to (1) aid in establishing causal relationships; and (2) to generate accurate descriptions of population parameters. It is necessary to evaluate alternatives to random sampling because it is often not feasible to apply this methodology. Deliberate sampling for heterogeneity can itself be seen as an application of critical multiplism. A model is suggested for utilizing deliberate sampling for heterogeneity when the population parameters are unknown. The model, along with critical multiplism, is applied in a study of the psychosocial rehabilitation workforce. The data suggests that deliberate sampling for heterogeneity may be more useful for generating theories and hypotheses than for generating accurate estimates of population parameters.