在stata中用Fisher-type tests (combining p-values)进行单位根检验用dfuller选项时加入drift后由不显著变为显著，求教那采用哪种检验才是有效的呢？
做过时序图，显示是有截距项的。结果及命令如下，求教各位！！！
xtunitroot fisher  a , dfuller lags(1)

Fisher-type unit-root test for a
Based on augmented Dickey-Fuller tests
--------------------------------------
Ho: All panels contain unit roots           Number of panels  =     11
Ha: At least one panel is stationary        Number of periods =     10

AR parameter: Panel-specific                Asymptotics: T -> Infinity
Panel means:  Included
Time trend:   Not included
Drift term:   Not included                  ADF regressions: 1 lag
------------------------------------------------------------------------------
                                  Statistic      p-value
------------------------------------------------------------------------------
Inverse chi-squared(22)   P        22.7803       0.4142
Inverse normal            Z         0.8136       0.7921
Inverse logit t(59)       L*        0.5657       0.7131
Modified inv. chi-squared Pm        0.1176       0.4532
------------------------------------------------------------------------------
P statistic requires number of panels to be finite.
Other statistics are suitable for finite or infinite number of panels.
------------------------------------------------------------------------------

. xtunitroot fisher  a ,   lags(1)  dfuller drift

Fisher-type unit-root test for a
Based on augmented Dickey-Fuller tests
--------------------------------------
Ho: All panels contain unit roots           Number of panels  =     11
Ha: At least one panel is stationary        Number of periods =     10

AR parameter: Panel-specific                Asymptotics: T -> Infinity
Panel means:  Included
Time trend:   Not included
Drift term:   Included                      ADF regressions: 1 lag
------------------------------------------------------------------------------
                                  Statistic      p-value
------------------------------------------------------------------------------
Inverse chi-squared(22)   P        47.3193       0.0013
Inverse normal            Z        -3.5409       0.0002
Inverse logit t(59)       L*       -3.5257       0.0004
Modified inv. chi-squared Pm        3.8170       0.0001
------------------------------------------------------------------------------
P statistic requires number of panels to be finite.
Other statistics are suitable for finite or infinite number of panels.
------------------------------------------------------------------------------

. xtunitroot fisher a,   lags(1) pperron

Fisher-type unit-root test for a
Based on Phillips-Perron tests
-----------------------------------
Ho: All panels contain unit roots           Number of panels  =     11
Ha: At least one panel is stationary        Number of periods =     10

AR parameter:    Panel-specific             Asymptotics: T -> Infinity
Panel means:     Included
Time trend:      Not included
Newey-West lags: 1 lag
------------------------------------------------------------------------------
                                  Statistic      p-value
------------------------------------------------------------------------------
Inverse chi-squared(22)   P        22.2921       0.4426
Inverse normal            Z         0.3480       0.6361
Inverse logit t(59)       L*        0.3109       0.6215
Modified inv. chi-squared Pm        0.0440       0.4824
------------------------------------------------------------------------------
P statistic requires number of panels to be finite.
Other statistics are suitable for finite or infinite number of panels.
------------------------------------------------------------------------------