Thanks Nick. In actuality, I did the entire thing manually last night
and this morning - what a nightmare and carpal tunnel syndrome to boot.
But I am going to try your code to reproduce my output and for future
reference. I appreciate your help. I have all Stata 9 documentation, is
there a reference or info on forval=i (I've never come across this
before - I'm not familiar with it).
Nick Cox wrote:
Thanks for adding extra detail.
My comments still stand.
What you tried is legal,
but will not do what you hope it does.
As I pointed out in another posting,
Richard's suggestion is just illegal.
"71600"/"71699" is not a legal string
expression, so violates -replace-'s
syntax.
Your guess that there should be
a way of specifying a range of string
values was correct. In addition to
my earlier suggestion, consider
forval i = 1/4 {
replace newvar = 1 if inrange(var`i',"71600","71699")
}
Nick
[email protected]
Suzy
The do believe the variables are string variables - the
variable font
is in red and many other values within the variable look
like strings
(eg. v8123, 3401-, e78345).
Each of these string variables may or may not have the same
value for
particular observations. For example, stringvar1 may have "71611" for
observation #10 and stringvar2 might have "71611" for observation #2
and stringvariable3 might have "71611" for observation # 3,000. and
stringvar4 might not have any "71611" for any observation.
I'm creating a newvar which can account for all the "71611" values
across all the string variables. In other words, the newvar must
include all values from "71600" to "71699" across all the
stringvars (of
which any stringvar may or may not include these values).
The reason I thought it was impossible to have no changes is because
for example, the value of "71623" across all string variables
resulted
in 10 observations for my newvar. Newvar =44 is just a category I
created to account for the value of "71623" across all the string
variables. [output eg. replace newvar=44 if var1 =="71623" |
var2=="71623" | var33=="71623" | var4=="71623" (10 real changes made)]
So for example, newvar=1 could have represented all the
"71600" values
across all string variables, whereas newvar=2 could represent all
"71601" values across all string variables and newvar=3 could
represent
all "71602" values across all string variables, etc, etc...
These are just examples of what's been done - not actual
categories per
se. Thus, newvar is a categorical variable that is
accounting for all
the values "71600 to 71699" across all the string variables.
The code I
had used to account for all of it in one shot didn't work,
replace newvar=1 if var1 =="71600/71699" |
var2=="71600/71699" | var3=="71600/71699" |
var4=="71600/71699" (0 changes made)
but Richard's idea of closing each value in parenth might work...I
haven't tried it yet.
replace newvar=1 if var1 =="71600"/"71699" |
var2=="71600"/"71699" | var3=="71600"/"71699" | var4=="71600"/"71699"
Hope this e-mail explanation is easier to understand. I am
using Stata 9 (windows xp).
Thank you.
Nick Cox wrote:
It's not impossible at all. Stata takes literally
what you include between double quotes " ". Evidently
you had no strings with values like "71600/71699".
However, it may be that what you want is accomplishable
by something like
forval i = 1/4 {
replace newvar = 1 if inrange(real(var`i'),71600,71699)
}
but like Richard Williams I am not completely clear from your
examples what exactly you do want.
Nick
[email protected]
Suzy
I created a new variable (newvar) that is supposed to be a
collection
of sequential values from different string variables. The
values that
are included in the new variable go from 71600 to 71699. I've been
doing it manually for hours- replacing value after value.
see example
for value # 71623:
replace newvar=44 if var1 =="71623" | var2=="71623" |
var33=="71623" |
var4=="71623"
(10 real changes made)
I tried to do it this way - see example:
replace newvar=1 if var1 =="71600/71699" | var2=="71600/71699" |
var3=="71600/71699" | var4=="71600/71699"
(0 changes made)
But no changes occurred, which is impossible.
I'm hoping there is a simple and quick syntax for what I'd
like to do.
Ideally if newvar =1 with all the values from 71600 t0
71699 located
from the existing variables (1-4) included in newvar - then
that would
be less painful than what I've been doing.
Any simple solution?? explicit code is appreciated!!
*
* 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/
