ADF test in Proc Arima

[nstdt]

I have a query on interpreting ADF test results from Proc Arima. I am learning time-series myself, so please do bear with me if this is a basic question.

When I run the ADF test, I get 3 sets of results from proc Arima: for Zero mean, Single mean and Trend.

There are p-values listed for two parameters: Rho and Tau. From online examples, I learned that stationarity is indicated if the the Tau p-value is small.

My question is, shold I look under Zero/single mean or Trend? Does this depend on the mean value of my series? For ex, if my time series, say quarterly GDP from 2010 to 2014

has mean 0.03, should I look under Zero mean, and if it had mean of 1.1, say, I would look under Single mean?

Thanks for any help!

[SnurreJensen · Accepted solution]

The choice of which ADF test to use depend on the original series as you mention. A series with apparent trend should look to the test including trend etc. This is to ensure that you are testing the null hypothesis against the most relevant alternative.

Some more elaborate explanations can be found here: http://stats.stackexchange.com/questions/44647/which-dickey-fuller-test-should-i-apply-to-a-time-ser...