Summary

Using the GLM Repeated Measures procedure, you have found that there is considerable market-to-market variation in sales. Also, the profile plots suggest that the first promotion is more effective than the others. In order to generalize these results to other markets, you would like to be able to specify Market ID as a random effect. While the GLM Repeated Measures procedure does not allow for random effects, the GLM Univariate procedure does not allow for within-subjects effects.

However, another of your findings was that the within-subjects factor Week might not be contributing to the model. If you choose to ignore the possibility of week-to-week variation, you could fit Market ID as a random effect.

Another solution is to use the Linear Mixed Models procedure. That procedure will allow you to specify Market ID as a random effect and give you greater control over the specification of the covariance matrix for the within-subjects factor.

In any case, the Promotion * Market ID interaction should be removed, because it does not contribute to the model.