2
$\begingroup$

I'm trying to measure the effect of an in-store media campaign on the sales. I have sales data along the time for test (treated) stores and control (not treated) stores.

Before comparing test and control, I have to rule out the price change and promotional factor for both time series. The decomposition of the time series into trend, seasonality and residual cannot isolate price or promotion effect.

My question is: how to model the price change component and remove it from the sales time series?

$\endgroup$
  • $\begingroup$ Do not remove them but incorporate those effects (price and promotion) into the model thus you will be evaluate the conditional impact without any assumption. Normally limited software is the excuse for mandatory up front adjustments/cleaning. $\endgroup$ – IrishStat Jul 16 '17 at 13:02
  • $\begingroup$ Do not remove them but incorporate those effects (price and promotion) into the model thus you will be able to evaluate the conditional impact without any assumption. Normally limited software is the excuse for mandatory up front adjustments/cleaning $\endgroup$ – IrishStat Jul 16 '17 at 14:28
  • $\begingroup$ Thanks for you answer. Could you please clarify what is "evaluate the conditional impact without any assumption" in a bit more detail? $\endgroup$ – Jingwei Zhang Jul 16 '17 at 14:56
  • $\begingroup$ When you incorporate k supporting/informational variables the coefficients for the other series are then conditional on the k variables. In this way you are examining the impact of the in-store media campaign GIVEN/CONDITIONAL UPON the effect of price and promotion rather than doing a 2 step procedure of adjusting Y for price/promotion first.. $\endgroup$ – IrishStat Jul 16 '17 at 15:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.