ABSTRACT
In the application of statistical models across various scientific disciplines, data that are grouped or clustered in some way are commonly encountered. It is also extremely common to see random-effect models used to acknowledge such data structures. A cluster-specific random effect can reflect the reality that data observations arising in the same cluster are likely more similar to one another than data observations arising across different clusters. Moreover, by making these cluster-specific effects random rather than fixed, “borrowing of strength” is attained. Inference about what happens in a particular cluster is aided by information drawn from the other clusters as well.
