In multilevel modeling, what does a random intercept mean?

• A. There is no intercept in the model.
• B. The slope is free to vary across groups.
• C. In multilevel modeling, the intercept is free to vary across groups.
• D. The intercept is fixed across all groups.

Answer: (C)

In multilevel modeling, a random intercept allows each group to have its own baseline level of the outcome. This is done by adding a group-specific term to the overall intercept, so different groups can start at different levels even when the predictor is the same. Mathematically, you’d see something like y_ij = β0 + β1 x_ij + u_j + ε_ij, where u_j is the random intercept for group j and ε_ij is the residual. The u_j terms capture the idea that some groups sit higher or lower on the outcome scale regardless of the predictor, and they are typically assumed to come from a distribution with variance τ^2.

This is distinct from a random slope, where the effect of the predictor (the slope) would vary by group, or from a fixed intercept, where all groups share the same baseline. So a random intercept means the intercept is free to vary across groups, reflecting group-level differences in the starting point.