I think you should look at “in-brand” correlation. My hypothesis is that they would undergo similar preference trainings and hence tend to prefer “in-brand” res

I think you should look at “in-brand” correlation. My hypothesis is that they would undergo similar preference trainings and hence tend to prefer “in-brand” responses over “off-brand” models that might have more significantly different reward training.