ANOVA for three different time periods

#1
I have three different time periods of sales data of 4 weeks apiece: weeks 1-4, 5-8, and 9-12.

During weeks, 9-12 we attempted a test that may have affected sales. Can I compare weeks 1-4, 5-8, and 9-12 with ANOVA to determine whether or not they have significantly different means? I don't have a control group for the test so I am trying to compare to the prior periods.
 

Miner

TS Contributor
#2
Do you only have the aggregate sales for those periods (3 numbers), or do you have the daily/weekly sales?
 

Miner

TS Contributor
#4
You could perform a 1-way ANOVA on either the daily or weekly data, but I personally would take a different approach. It is not well known outside of industrial statistics, so you probably have never heard of it.

I would plot the daily data on an Individuals - Moving Range control chart. If there are no outside influences, the sales should randomly vary within the control limits. If there is an outside influence, the data should shift at or about the time that you employed the test. The shift would cause the sales to be outside of the control limits. See Nelson's Rules for interpretation.

The advantage of this approach is that it uses the time series nature of the data better than the ANOVA.