Hi stata-experts!
I'm dealing with a dataset (survey experiment) with a long format where I have four responses (Y) per observation (id) and no time variable. Treatments (X) are randomly distributed by id (control group, treatment group 1, treatment group 2), not by each response. For now I have simply used:
reg responses i.treatment, cluster(id) robust
To be honest my stata knowledge is not that extensive which is why I'm now doubting my own methods. Would it make more sense to use the xtreg-command with some sort of effects?
Best regards,
Pekka
I'm dealing with a dataset (survey experiment) with a long format where I have four responses (Y) per observation (id) and no time variable. Treatments (X) are randomly distributed by id (control group, treatment group 1, treatment group 2), not by each response. For now I have simply used:
reg responses i.treatment, cluster(id) robust
To be honest my stata knowledge is not that extensive which is why I'm now doubting my own methods. Would it make more sense to use the xtreg-command with some sort of effects?
Best regards,
Pekka
