RE: Re:st: xtpcse - "no time periods are common to all panels"

[Nick Cox <n.j.cox@durham.ac.uk>]

First, -egen, total()- can work with an expression, so that 

egen howmany = total(1), by(year) 

would work and a new variable -iota- is not needed. 

Second, that in turn is another way of doing 

bysort year : gen howmany = _N 

Nick 
n.j.cox@durham.ac.uk 

Christopher Baum

What about

g iota = 1
egen howmany = total(iota), by(year)
tab year howmany
drop if howmany < (number of years)  where number of years refers to the entire dataset's range 
tab year howmany

On Nov 18, 2011, at 2:33 AM,Wayne wrote:

> Thanks to Kit, Gordon, and Nick for their helpful answers to my question 
> about -xtpcse-.  I had assumed that all of the panels included at least 
> some common years, but that was not the case.  There may be easier ways 
> to check this, but here is how I did it:
> 
> * how many panels are present in each year?
> tabstat year, by(year) stat(N)
> xtset panel_id year
> * what is the pattern of missing years?
> xtdescribe
> by panel_id: gen yrs=_N
> tab yrs
> sort yrs panel_id year
> * visually checking the sorted data, it was easy to see that countries 
> with fewer than 12 years of data were the problem
> drop if yrs<12
> drop yrs


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/