The expectation of the random compensating variation is the welfare change measure that is used in discrete choice models. The expectation of the equivalent variation is equally founded theoretically. When choices are income independent, the two measures take identical values. The case of income-dependent choices remains an area for exploration. The paper provides a twofold theoretical contribution. First, it derives the discrete choice model with n alternatives and a translog and power specification of the residual income term of the indirect utilities from a quasilinear direct utility in the consumption of the discrete goods and a numéraire. Second, it provides the equivalent variation counterparts of the formulas that are available for the expectation of the compensating variation: the one based on the random expenditure function, and the one based on conditional on observed transition expectations of the compensating variation. An example related to multinomial logit and choice of the transport mode of urban travel illustrates the insights.