You have identified the wrong problem. The problem is
that -cii- requires actual numbers of observations
and successes, so your syntax is illegal. There
is no problem in applying -ci- with -if-.
Is -treat- a string variable? Even if it is,
if treat = "`group'"
needs == not =.
A program -ciplot- from SSC does something
which sounds similar to your problem. If
you want something different, the code
can probably be cannibalised for your
problem.
Nick
[email protected]
Douglas Crawford
> In preparation for using rcap I want to create two variables:
> upper and
> lower confidence intervals for a proportion in my data. I
> have already
> collapsed the data set. I know that I can use
>
> cii obs successes, binomial
>
> to generate the appropriate information. My question is, Is
> there a way of
> generating a variable = to the r(ub) and r(lb)?
>
> I tried the following without success:
>
> levels treat, local(groups)
> gen pct_bag_ub = .
> gen pct_bag_lb = .
> gen pct_bag_se = .
> gen pct_bag_mn = .
> gen N = .
>
> foreach group of local groups {
> cii obs successes if treat = "`group'"
> replace pct_bag_ub = r(ub) if strain_treat == "`group'"
> replace pct_bag_lb = r(lb) if strain_treat == "`group'"
> replace pct_bag_mn = r(mean) if strain_treat
> == "`group'"
> replace pct_bag_se = r(se) if strain_treat == "`group'"
> replace N = r(N) if strain_treat
> == "`group'"
> }
>
> But cii does not support if (nor does ci).
>
> I also tried statsby, but ci does not appear to be supported command.
>
*
* 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/
