Kate wrote
>
> Somebody sent me a dataset with variables like this
> one. There were 5 procedures and more than one could
> be chosen. Instead of making each a variable, a
> single variable was made with commas separating the
> procedure number.
>
> procedure | Freq.
> ------------+-----------
> 01,02,03 | 11
> 01,02,03,04 | 1
> 02 | 6
> 02,03 | 107
> 02,03,04 | 14
> 02,04 | 1
> 03 | 3
> 03,04 | 2
> 04 | 2
> 05 | 6
>
> I want to make 5 variables that record yes or no for
> each procedure. Index() seems to be the way to go but
> but I don't see how to use it when there's more than
> one instance of the index character. Can somebody
> point me in the direction?
James Hesford suggested
> In Stata, type "findit split"
> You will find a FAQ by Nick Cox on splitting a string variable into
> parts. The FAQ can also be found at
> http://www.stata.com/support/faq/data/splitstr.html
That should be
http://www.stata.com/support/faqs/data/splitstr.html
For Stata 8 this FAQ is superseded by the -split- command.
I will revise the FAQ ASAP. However, this isn't the answer
to Kate's question, as -split- will not create the
indicator variables Kate needs.
Shannon Driver suggested
> You might try something like the following:
> forvalues i = 1/5 {
> gen proc0`i' = cond(index(procedure,"0`i'"),1,0)
> }
which does seem the right way to go. Note in passing that
cond(index(procedure,"0`i'"),1,0)
could also be
index(procedure,"0`i'") > 0
There is a more general FAQ on multiple
responses by Ulrich Kohler and myself at
http://www.stata.com/support/faqs/data/multresp.html
which in turn has been revised and expanded
in Stata Journal 3(1), 81-99 (2003).
Nick
[email protected]
*
* 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/
