I can't reproduce your results here. I get
the same results as Scott. Your logic that
0 & 0 & 1 == 1
is not the same as Stata's (or Boole's).
In other words, my code appears equivalent
to Scott's.
That's using in this example
gen nmark = natch < . & ednatch < . & marker != 0
Nick
[email protected]
Tony [= Peter A.] Lachenbruch wrote a note in the
STB on < .
STB-9 ip2 . . . . . . . . . . . . . . . . . . . . . . . A keyboard shortcut
. . . . . . . . . . . . . . . . . . . . . . . . . . P. A. Lachenbruch
9/92 p.9; STB Reprints Vol 2, p.46 (no commands)
keyboard shortcut to indicate nonmissing values
It's now recommended practice by StataCorp given that != . includes
any .a through .z.
Clive Nicholas
> Scott Merryman replied:
>
> [...]
>
> > . mark mark2 if edconch != .
>
> > . replace mark2 = 0 if conch == .
> (0 real changes made)
>
> > . replace mark2 = 0 if marker == 0
> (6 real changes made)
>
> [...]
>
> This worked perfectly: thanks!
>
> Nick Cox replied:
>
> [...]
>
> > gen mark = conch < . & edconch < . & marker != 0
>
> [...]
>
> Thanks for this: this also worked almost as well. Why the difference?
> Well, consider this block of data for Plaid Cymru (National Party of
> Wales) net votes, where -mark4- represents Scott's routine, -nmark-
> represents Nick's routine and -marker- was an -egen, tag()- generated
> indicator marking out unavoidable duplicate cases:
>
> natch ednatch marker mark4 nmark
> 1. -4.727228 21.4088 1 1 1
> 2. .7938523 9.794947 1 1 1
> 3. -1.833212 -4.599777 1 1 1
> 4. 2.022674 24.89381 1 1 1
> 5. 1.016729 5.512588 1 1 1
> 6. 3.982042 5.620074 1 1 1
> 7. -1.833212 -4.599777 0 0 0
> 8. 2.022674 -2.766565 0 0 0
> 9. 2.022674 -2.766565 0 0 0
> 10. -1.833212 21.08436 0 0 0
> 11. . . 1 0 1
> 12. . . 1 0 1
> 13. . . 1 0 1
> 14. . . 1 0 1
> 15. . . 1 0 1
> 16. . . 1 0 1
> 17. . . 0 0 0
> 18. . . 0 0 0
> 19. . . 0 0 0
> 20. . . 0 0 0
> 21. . . 0 0 0
> 22. . . 0 0 0
> 23. . . 0 0 0
> 24. -1.250048 9.806999 1 1 1
> 25. -1.039213 -1.973771 1 1 1
> 26. .0781536 -.9345583 1 1 1
> 27. .1249657 -1.012712 1 1 1
> 28. .6216435 . 1 0 0
> 29. 1.577682 5.098733 1 1 1
> 30. . . 1 0 1
>
> Scott's routine correctly marks out the missing observations
> 11-16 and 30;
> Nick's routine marks them _in_ because marker=1 (but I don't
> need these
> marked cases if they're missing on at least one of the vote
> variables).
>
> However, Nick's approach is attractive as it attempts to do
> it all in a
> single line, so the question is whether or not Nick's code
> can be tweaked
> to correctly mark out these cases. This isn't to disparage Scott's
> approach just because he does it in three.
>
> Thanks to Scott for introducing me to -mark- and to Nick for
> thinking up
> such a lateral way to -generate- (I would never have thought
> to use "<.").
> :)
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
