Accounting for Pre-existing differences between Control and Treatment groups

#1
I've got two sets of sales data. I want to compare the two groups after one has a sales program applied to it. However, the treatment group already consists of the best performing stores in the analysis.

My question is, how can I normalize/standardize the data so that I can accurately report the effect of the treatment?

To clarify, my treatment group already averages higher units sold than my control group. I want to be able to measure the effect of the sales program compared to the control.

Do I just subtract the original average difference? Use the means and standard deviations to adjust them individually to the same 0,1 standard scale? Any advice is greatly appreciated.

Thanks
 

rogojel

TS Contributor
#2
hi,
this sounds like a two sample t test or possibly a paired t test. If you compare the two groups, you can set up the null hypothesis so that the mean diffence between the two groups is larger the D, where D is the already existing difference before applying the treatment.

You could also completely ignore the untreated group an run a before-after paired t test with only the treated group.

regards
rogojel
 

hlsmith

Less is more. Stay pure. Stay poor.
#3
Good suggestions rogojel!

fatso101, is there any particular reason why one of the groups performs better than the other that may affect their adoption of the intervention or the generalizability of their results to other stores?