Discovering Statistics Using IBM SPSS Statistics, 5th Ed. Practice Test

In multilevel modeling, the most flexible way to specify how random effects relate to each other is the unstructured covariance structure. It allows every variance on the diagonal and every covariance off the diagonal to be estimated freely from the data, with no preset pattern. This makes it completely general because no assumptions are imposed about how the random effects covary. The trade-off is that it involves many parameters, so it needs enough data to estimate reliably and can be harder to fit. The other options aren’t covariance structures that describe the random-effects relationships: one option is not a standard covariance pattern at all, another concept isn’t about covariances, and the last refers to priors rather than a covariance form.