Re: st: flagging a specific pattern of values within observations

["Sergiy Radyakin" <Radyakin@aoek.uni-hannover.de>]

Hello Friedrich,

this will work only for small number of options, i.e. 3 as here. Notice that 
in the list
you specify not all the "options", but all "combinations". This will grow 
very fast (n!).
So it will take 120 combinations to list (and to check!) for n=5.
With just n=6 (n!=720) one can already forget about it.



The following will check if v1, v2, and v3 are different (c==1 if not):
gen c=1-inlist(v1, v2, v3)*inlist(v2,v1,v3)*inlist(v3,v1,v2)

or using built-in function diff:

egen c=diff(v1 v2 v3)

:)

(there will be 5 terms in the RHS for 5 variables, etc -- the number grows 
as n)
There is no check that 1<=vi<=n. It is unnecessary if the data stems from, 
say users choosing
options from drop-down lists. But if necessary, this can be easily 
accomplished by also checking
if inlist(v1, 1, 2,...,n)*inlist(v2, 1, 2,...,n)*...*inlist(vn, 1, 2,...,n) 
is equal to 1.
(n more conditions, totalling 2n (or n+1 if the shortcut is used), instead 
of n!)

But ideally I would like to write something like:
gen k= set(v1 v2 v3)=set(1 2 3)
(order of elements does not matter for sets)

or alternatively
gen k= sorted(v1 v2 v3)=(1 2 3)
(to have a sorted list compared with the benchmark).

Stata allows to compare lists:
di (1 2 3)==(1 2 3)

di (1 2 3)==(1 3 2)

but does not allow this:

gen k= (v1 v2 v3)==(1 2 3)


So some questions for the Gurus:
 -- how to subtract an element from a list?
 -- does Stata support sets? (e.g. at the level of Pascal sets)
 -- v1 v2 ...vn are chars: "a" "b" etc. Need X=concatenated sorted v1..vn. 
E.g:

v1    v2    v3    X
a    b    c    abc
b    a    c    abc
a    b    a    aab
a    a    a    aaa

  What is the shortest way? What if vi-s can be strings (not just chars)?
ab    aa    ca    aaabca

Best regards,
  Sergiy






----- Original Message ----- 
From: "Friedrich Huebler" <huebler@rocketmail.com>
To: <statalist@hsphsun2.harvard.edu>
Sent: Tuesday, April 24, 2007 8:08 PM
Subject: Re: st: flagging a specific pattern of values within observations



> This should do it.
>
> input fid q11a q11b q11c
> 55 2 3 1
> 21 1 2 1
> end
> gen q = string(q11a) + string(q11b) + string(q11c)
> gen tag = !(inlist(q,"123","132","213","231","312","321"))
>
> Friedrich
>
> --- Alan Neustadtl <alan.neustadtl@gmail.com> wrote:
> > Makes sense...thanks for finding the flaw.
> >
> > Best,
> > Alan
> >
> > On 4/24/07, Sergiy Radyakin <Radyakin@aoek.uni-hannover.de> wrote:
> > > No, this will not work(2-2-2):
> > >
> > > ----------------------------
> > > fid     q11a    q11b    q11c
> > > 55      2        3       1
> > > 21      1        2       1
> > > 22      2        2       2
> > > ----------------------------
> > >
> > > Prime numbers seem to be a better idea.
> > >
> > > Best,
> > >   Sergiy
> > >
> > >
> > > ----- Original Message -----
> > > From: "Alan Neustadtl" <alan.neustadtl@gmail.com>
> > > To: <statalist@hsphsun2.harvard.edu>
> > > Sent: Tuesday, April 24, 2007 7:30 PM
> > > Subject: Re: st: flagging a specific pattern of values within
> > observations
> > >
> > >
> > > > How about something like this:
> > > >
> > > > clear
> > > > input fid q11a q11b q11c
> > > > 55  2 3 1
> > > > 21  1 2 1
> > > > end
> > > >
> > > > tempvar checkit
> > > > egen `checkit'=rowtotal(q11a q11b q11c)
> > > > generate byte flag=1 if `checkit' !=6
> > > > list if flag==1
> > > >
> > > > Best,
> > > > Alan
> > > >
> > > >
> > > > On 4/24/07, Anirudh V. S. Ruhil <ruhil@ohio.edu> wrote:
> > > >> A colleague is trying to accomplish something in a different
> > stats
> > > >> package
> > > >> and asked if STATA could effect the solution. "Of course",
> > said I. Now I
> > > >> am
> > > >> scratching my head for the code.  Individuals were asked to
> > rank-order 3
> > > >> choices. Person 55 did fine, but not Person 21. How can I
> > generate a flag
> > > >> for such respondent-errors?
> > > >>
> > > >> ----------------------------
> > > >> fid     q11a    q11b    q11c
> > > >> 55      2               3       1
> > > >> 21      1               2       1
> > > >> ----------------------------
> > > >>
> > > >> thanks in advance
> > > >>
> > > >> Ani
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
> *
> *   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/ 
*
*   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/