Apologies but as I am not well-versed in statistics, this might work best if I describe my data.Thanks for your patience.
The exercise is that I'm testing to see if an advertising campaign had a statistically significant (I'll go with the usual P = 0.05) effect on the number of transactions.
I have the number of transactions for two groups of customers: those who received the promotion (60,000), and those who did not (50,000) who are the control group. Although I could plot them as a continuous time-series, I've grouped the transaction counts into two distinct time periods: Pre campaign, and during the campaign:
No. Transactions
So, clearly the advertising campaign generated more transactions in the test group vs control group. My question is, what would be an appropriate test to confirm the significance?
The layout makes me think Chi Square test for independence, but would it be valid to say "The two groups are independent, therefore the uplift is significant"?
I was also looking at Chi Squared test for goodness of fit but, since I know the expected proportions from the group sizes:
But I'm not sure the test is actually valid for comparing two separate time periods with their own set of expected values.
Obviously I might be completely wide of the mark here. Any suggestions welcomed, use of lay language appreciated.
Thanks!