How to change default CI

[Tia Landry]

Hi community,

I am working on some survey data and need to use the Wilson CI. I am not sure how to specify this within Stata to change from the default Clopper–Pearson.

Any guidance on this would be greatly appreciated!

-T

[Tiago Pereira]

Can you clarify the command your are using?

[Tia Landry]

So when doing proportions in stata using survey data this is what we’d do:

svy: prop (var1), over (var2)
estat cv

[Tiago Pereira]

Type

help proportion

Then check the option

citype(citype)

[Tia Landry]

Thanks very much for the suggestion. I do run into some issues using citype(citype)


Set weights

code:

svyset [pweight=WeightN], brrweight(Wgt_1- Wgt_500) vce(brr) dof(500)

Proportions

code:

svy: prop var1, over (var2), citype (wilson)

Error: invalid 'citype'

however if I used the following code below it works. The only thing is if you want to look at 3 variables at a time or something you’d have to do a subpop each time rather than using “over” which makes this easier to look at all at once.

svy, subpop (if var2==1): prop var1, citype (wilson)
svy, subpop (if var2==2): prop var1, citype (wilson)

Any suggestions to help make this code work with the over option?

[Rich Goldstein]

the problem with the code in #5 that did not work is the extra comma before "citype"; options follow odd numbered commas but you have put a second comma and that is wrong

[Bruce Weaver]

You have two commas in the command that isn't working. Try this:

svy: prop var1, over(var2) citype(wilson)

[Tia Landry]

Thanks so much I don’t know how I missed that. Appreciate all the help. Thanks again everyone!

[Tia Landry]

Would you happen to know how you can specify a wilson ci within an espost svy command, citype(citype) does not work here.

estpost svy: tabulate var_1 var_2 if var_1==1 |var_1==0,col percent format(%11.3g) ci

[Tia Landry]

I have tried ci(wislon) with no luck when its included with estpost