I posted a quasi-tutorial on this general area a short while
back. It may help understanding of the numeric <-> string
conversion problems (and they are problems, not a single
problem).
<http://www.hsph.harvard.edu/cgi-bin/lwgate/STATALIST/archives/statalist.0707/Author/article-469.html>
There is no reason why -encode- should not be used.
String variables can have embedded numeric characters.
That's absolutely no reason against using -encode-.
It may not help much but it is legal and a practical
possibility.
Otherwise, I think you need to look for substrings
like "200 MG/ML" using -strpos()- (-index()- in
old Stata versions). The string functions are
your friend here.
strpos(drug_1, "200 MG/ML")
will be positive if "200 MG/ML" is included and
zero otherwise.
-destring- is no use to you at all.
I don't understand what an "explore variable" is.
Explanatory?
Nick
[email protected]
P chrisman
> I am having a great deal of trouble changing strings to
> numeric. I have
> a variable "drug_1" that is a strg40 with 32 unique values
> (I did not
> create this data set!). Examples are:
>
> "TESTOSTERONE CYP 200MG/ML INJ (IN OIL) "
> "TESTOSTERONE ENANTH. 200MG/ML 1ML "
> "TESTOSTERONE ENANTHATE 200MG/ML INJ 5ML "
> "TESTOSTERONE ENANTHATE 200MG/ML INJ 5ML "
> TESTOSTERONE (ANDRODERM) 5MG/24HRS PATCH
>
> Drug dose is the explore variable and the PI wants to look at
> outcome by
> mode of administration and dose, so for the example here I
> need to have
> a drug_1 variable with mode 1=IM and 2=patch and dose 200mg/ml,
> 200mg/5ml and 5mg/24hr.
>
> I have tried encode, but I do not think I can use that because the
> strings also contain numbers(please correct if this is not true). I
> tried destring with the following errors:
>
> destring Drug_1, replace
> Drug_1 contains non-numeric characters; no replace
>
> . destring Drug_1, replace ignore("testosterone")
> Drug_1 contains characters not specified in ignore(); no replace
>
> I could not find anything in FAQ that was similar to this problem. I
> would appreciate ANY HELP as I have drug_1 through drug_5 for 384
> subjects and can do this the long and painful way, but would
> prefer not.
*
* 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/
