Under incomplete information about the parameters of the true distribution of the random coefficients, the optimal solutions to stochastic programs can be only approximated. This paper extends the previous results of the author to the case when strict complementarity conditions need not be assumed.