The WEIGHT statement is used to perform a weighted analysis. Consult the section Log-Likelihood Functions for Response Distributions for expressions on how weight variables are included in the log-likelihood functions. Because the probability structure of a mixture model is different from that of a classical statistical model, the presence of a weight variable in a mixture model cannot be interpreted as altering the variance of an observation.
Observations with nonpositive or missing weights are not included in the PROC FMM analysis. If a WEIGHT statement is not included, all observations used in the analysis are assigned a weight of 1.