-svy tab-, with a "ci" option uses the logit transform by default and for proportions, not for percentages. It can be used for ordinary, non-survey, data by preceding the -svy tab- command with
svyset _n
Steve
On Sep 9, 2011, at 3:42 PM, Bryan Sayer wrote:
While it is true that the mean of a 0/1 variable is equal to the proportion, the confidence interval obtained this way may go below zero or above one, which is generally an undesirable characteristic.
Some software will use a logit transformation to prevent this, but only when doing confidence intervals of actual percentages.
*
* 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/
