Forecasting using SAS Forecast Server, SAS/ETS, and more

time series event model with a control and treatment group? possible?

Reply
Contributor
Posts: 30

time series event model with a control and treatment group? possible?

Is there a way to compare two time series models statistically?

I have data from a pre-intervention period and a post-intervention period for both a control group and a treatment group. There was no intervention in the control group. So I expect to see no change in the time series for this group and I expect to see a change in the time series for the treatment group. Is there a way to test for differences in these two time series? Thanks.
Occasional Contributor
Posts: 15

Re: time series event model with a control and treatment group? possible?

You can use same statistics, for example MAPE (Mean Absolute Percentage Error) to compare two time series.
SAS Employee
Posts: 416

Re: time series event model with a control and treatment group? possible?

Hello -
Did you consider running a similarity analysis (PROC SIMILARITY of SAS/ETS)?
http://support.sas.com/documentation/cdl/en/etsug/63348/HTML/default/viewer.htm#etsug_similarity_sec... might point you in the right direction.
Thanks,
Udo
Contributor
Posts: 30

time series event model with a control and treatment group? possible?

The method in this document suggest putting two outcome values at one time point.  Can you do that?

See pages 48 to 51 of this link.  Let me know if you think this method is valid or not.  Thanks

http://www.oregoneval.org/ANALYSIS%20OF%20INTERRUPTED%20TIME%20SERIES%20FINAL.pdf

Occasional Contributor
Posts: 5

time series event model with a control and treatment group? possible?

Hi,

Consider using DID method rather then time series analysis as the time factor in your question seems to be 'before' and 'after'

you can start with this article:

http://www.nber.org/WNE/lect_10_diffindiffs.pdf

Ask a Question
Discussion stats
  • 4 replies
  • 463 views
  • 0 likes
  • 4 in conversation