Marginal model

In a marginal model, we collapse over the latent/unobserved level 1 & 2 residuals (R and U) and thus marginalize the joint distribution of the response variable (Y_ij) into an univariate distribution.

In multilevel modeling, we fit the marginal model to data.