<>
Well, Nick`s earlier advice remains completely valid, although you are right
that -csi- checks internally for integers via
-confirm integer number 2-
which passes the test
and
-confirm integer number 2.0-
which does not...
Here is an example to play around with for those who are interested:
***
clear*
set obs 100
gen x=cond(_n<=_N/2,0,1)
gen y1=cond(_n<=_N/2,1,3)
gen y2=cond(_n<=_N/2,3,5)
sum y1 if x==1, mean
local a1 = round(r(mean),0) // round to create integers
sum y1 if x==0, mean
confirm integer number `r(mean)'
di in red _rc
loc b1=r(mean)
/*local b1 = round(r(mean),0)
*/
sum y2 if x==1, mean
local c1 = round(r(mean),0)
sum y2 if x==0, mean
local d1 = round(r(mean),0)
// set tr on
csi `a1' `b1' `c1' `d1'
***
HTH
Martin
_______________________
----- Original Message -----
From: "David Elliott" <[email protected]>
To: <[email protected]>
Sent: Thursday, May 07, 2009 12:54 AM
Subject: Re: st: using computed values with the immediate command
Stata is telling you what is wrong. It has nothing to do with not
being able to store values. I assume since you are asking for r(mean)
after -summarize- that you are saving a scalar float with several
decimal places. However, -csi- the routine expects an integer based
on count data.
The following will probably be complete nonsense statistically but
illustrates the point:
sum stots if heart==1
local a1 = round(r(mean),0) // round to create integers
sum stots if heart==0
local b1 = round(r(mean),0)
sum stotd if heart==1
local c1 = round(r(mean),0)
sum stotd if heart==0
local d1 = round(r(mean),0)
csi `a1' `b1' `c1' `d1'
Giving csi a scalar, even if it is an integer, doesn't work. Contrast
the results of:
-csi 2 4 6 8- versus -csi 2.0 4.0 6.0 8.0-
D. C. Elliott
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
