How do I interpret Augmented Dickey Fuller test for stationarity?

[tirzac]

Hello,

I am using SAS 9.4 (TS1M5). I am trying to figure out how to model my time series data, using segmented regression analysis. I am looking at monthly proportions of kids up to date on vaccination, across a total of 96 months. The purpose of this analysis is to evaluate the impact of a policy change that occurred at month 84 on the rate of kids up to date. Upon visual inspection of the data, it appears that there is an overall decreasing trend. See attached graphic. 

I am having trouble understanding how to interpret the Augmented Dickey Fuller test for stationarity. All output is attached. I believe a sound approach is to use proc arima to run this test. Under "trend", Tau and Rho have a p-value<0.05 for lags 0 and 1. My understanding of this is that I should not have an issue with stationarity, as long as I include a trend variable in my model. However, I'm not understanding why Rho and Tau also have p-values<0.05 for lag 0 under "zero mean" and "single mean". Could you please advise? Thank you!

proc arima data=d.late;
identify var=latept stationarity=(adf=(0,1,2));
quit;run;

[rselukar · Accepted solution]

This is not an answer but I am providing doc links that might help you get the answer.  A new example was added to the ARIMA doc https://go.documentation.sas.com/?docsetId=etsug&docsetTarget=etsug_arima_examples08.htm&docsetVersi... .

It shows how to use the stationarity test options.  Also see a reference mentioned in the example: Dickey, D. A. (2005). “Stationarity Issues in Time Series Models.” In Proceedings of the Thirtieth Annual SAS Users Group International Conference. Cary, NC: SAS Institute Inc. http://www2.sas.com/proceedings/sugi30/192-30.pdf