Exploration & description/Observational units

In mixed model analysis we have different types of units occurring at different layers – namely in this example: lambs, ewes, rams. The investigational or observational units defined within layers are assumed to be chosen independently of one another; usually they are chosen at ´random´. They will therefore be random effects in our mixed model.

Correctly identifying the layers of observational units and the different attributes assigned to units is crucial to a successful understanding of how hierarchical data should be analysed. This is what we aim to do with our mixed model tree.

We have two breeds. From within each of the two breeds a number of rams is selected. These are the observational units (ram shown as a random effect) against which the two breeds of rams should be compared.