[Nick Cox]

Applicable to small samples only? Not at all. The results you show contradict that, with regressions for 400000 observations.

[Johnny YANG]

The results for the first identifier are for identifiers 2 up, those for the second 3 up, and so on; not what you want at all. See again #8 and the help. I guess that you want interval(sic_id 0 0).

Got it. Thank you ! I have another question, because I did repeated regression following Joseph's method. I did the drop if mi(sale oprix), the missing value has been deleted indeed, but some entries have only one observation, thus the reg keeps reporting insufficient value.
I have no idea how many observations does reg need, and how to do it properly, except for drop if mi().

[Nick Cox]

Stata will certainly tell you when the number of observations is too small for estimation, which is a matter of basic mathematics. If you’re asking something else there can’t be a precise answer. Note that e.g. R-square of 1 is often a bad sign....

Stata does not delete observations with missing values when doing regression: it necessarily ignores them.

[alessio lombini]

Dear all,

I am getting the same error (namely, ( invalid name r(198); ) in a similar situation. This is my code:

tempname anees_uroot
tempfile unitroots

postfile `anees_uroot'  str12 name dfuller_statistic dfuller_pvalue dfuller_lags  using `unitroots'


foreach var of varlist backward_total regulatory_distance {

xtunitroot  fisher `var', dfuller lags(1) demean trend
local dfuller_statistic = r(Zt)
xtunitroot  fisher `var', dfuller lags(1) demean trend
local dfuller_pvalue = r(p)
xtunitroot  fisher `var', dfuller lags(1) demean trend
local dfuller_lags = r(lags)

post `anees_uroot' ("`var'") (`dfuller_statistic') (`dfuller_pvalue') (`dfuller_lags')
}

Might you please tell me how to solve this issue?
Best regards