我只会用3.0做

Testing for nonstationarity with the Dickey-Fuller (DF) test (12.4.2):
Since the AC analysis in the previous section indicated that CO is most likely an AR(1) process,
the Dickey-Fuller (DF) test is valid. Follow these steps to conduct the Dickey-Fuller test of the
hypothesis that the CO series is non-stationary:
Step 1. Open the EViews workfile named Macro14.wf1.
Step 2. Open CO in one window by double clicking the series icon in the workfile window. Note that
EViews will probably display the correlogram view for CO since that was the last view selected
in the previous section. You can select View/Spreadsheet to view the series data or just proceed
with the next step to test for stationarity.
Step 3. To conduct the Dickey-Fuller (DF) test, select View/Unit Root Test… on the CO series
window menu bar.
Step 4. Four things have to be specified in the Unit Root Test dialog box to carry out a unit root test.
First, choose the type of test—either the Augmented Dickey-Fuller (ADF) test or the Phillips-

Perron (PP) test (select ADF for this example).7 Second, specify whether to test for a unit root in
the Level, 1st difference, or 2nd difference of the series (select level for this example).8 Third,
specify whether to include an Intercept, a Trend and intercept, or None in the test regression.
Select Trend and intercept for this example. To see why, read footnote 18, UE, p. 427. Fourth,
specify the number of lagged first difference terms to add in the test regression (0 for the DF
test). The theory behind each of these selections is beyond the scope of UE and this guide.
Advanced econometrics courses deal with these issues. When finished with the selections click

Step 5. The test fails to reject the null hypothesis of a unit root in the CO series at any of the reported
significance levels, since the ADF Test Statistic9 is not less than (i.e., does not lie to the left of)
the MacKinnon critical values.

你自己看吧，我也不记得了，当时好像就是按这个做的