Summary

Using the Linear Mixed Models procedure, you have successfully assessed the value of each promotion, adjusting for market-to-market variation. An alternative method for specifying simple random effects in this example would be to leave the subjects listbox empty, then specify Market ID as the factor in the Random Effects dialog.

An advantage of this method is that the market*promotion interaction could, if appropriate, be modeled with covariance structures other than the Identity structure. For an example of this specification, see the example of Using Linear Mixed Models to Analyze a Crossover Trial analysis.

You could also, alternatively, specify Market ID as a fixed factor. However, the results would then not be generalizable to all markets. That is, your estimates of the promotional effects would only apply to those markets represented in the study.