How to keep id that have two observations

[Raphael George]

Hello, I want to keep HHPBASE that has both dup 1 and 2. How do I delete observations that just have one copy in order to create a panel

bys HHPBASE : gen dup=_n

----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. To install: ssc install dataex
clear
input double HHPBASE float dup
101020102009 1
101020102009 2
101020103002 2
101020105003 1
101020116004 1
101020116004 2
101020117001 1
101020118005 1
101020120004 1
101020120004 2
101020204002 1
101020206004 1
101020206004 2
101020207004 1
101020207004 2
101020214002 1
101020215002 1
101020215002 2
101020216002 2
101020217002 1
101020217002 2
101020220002 1
101020305002 1
101020306009 1
101020306009 2
101020307002 1
101020310004 1
101020311002 1
101020314004 1
101020314004 2
101020316002 1
101020319004 1
101020319004 2
101020401002 1
101020401002 2
101020404002 1
101020408004 1
101020409002 1
101020413017 2
101020415002 1
101020417011 1
101020417011 2
101020501002 1
101020504004 1
101020507004 1
101020507004 2
101020512003 1
101020515002 1
101020517002 2
101020518002 2
101020520004 1
101020520004 2
101020601002 1
101020603002 1
101020603002 2
101020604002 1
101020606002 1
101020611002 1
101020612004 1
101020612004 2
101020615004 1
101020615004 2
101020701002 1
101020701002 2
101020702002 1
101020705002 2
101020706004 1
101020706004 2
101020707002 1
101020710002 1
101020711002 1
101020712002 1
101020803003 1
101020806002 1
101020807002 2
101020809002 2
101020812002 1
101020812002 2
101020813002 2
101020814002 2
101030101004 1
101030102002 1
101030104002 2
101030105003 1
101030109004 1
101030202002 2
101030205004 1
101030207002 2
101030213002 2
101030214002 1
101030214002 2
101030311004 1
101030311004 2
101030315004 1
101030315004 2
101030408002 2
101030411002 2
101030412002 2
101030414002 1
101030414002 2
end

------------------ copy up to and including the previous line ------------------

Listed 100 out of 18233 observations
Use the count() option to list more

[Carlo Lazzaro]

Raphael:
I'd try (not tested, though):

bysort HHPBASE (dup): keep if _N==2 & dup[1]!=dup[2]

[Nick Cox]

Carlo Lazzaro 's code should work fine if, as stated, the values can only be 1 and 2. If other values are possible, you may need to be more explicit:

bysort HHPBASE (dup): keep if _N==2 & dup[1] == 1 & dup[2] == 2

or

  
bysort HHPBASE (dup): keep if _N==2 & dup == _n 

[Raphael George]

Thank you for the help