Marginal generalized linear models for correlated data are often fit by generalized estimating equations (Liang, K.Y. and Zeger, S.L., 1986, Biometrika 73, 13-22), which requires the marginal expectation to be correctly specified while the covariance structure is allowed to be misspecified. This note is concerned with the correct specification of the marginal mean for data from a mixed log-linear regression model and the relationship between the true subject-specific parameters and the true marginal parameters.