Problem with 95% Conf. interval

[suifang YE]

Hi All,

I am trying to calculate the proportion of two specific variables and compare the difference between these two population proportions. From the table below, In the 67 samples, the proportion of co-infection with vaccine virus is 2/7 (29%), while the proportion of non-vaccine virus with co-infection is 11/60 (18%).
I have chosen the Two-sample test of proportion. I am not sure about the 95% CI for the proportion of the co-infection with vaccine virus. Isn't the possible range of values is 0 to 100%? What will be the right test to workout the CI for the proportion? Many thanks.

Coinfection	vaccine virus absent (0)	present (1)	Total
absent (0)	49	5	54
present (1)	11	2	13
total	60	7	67

code:
. prtest coinfection, by(vaccine)

Two-sample test of proportion 0: Number of obs = 60
1: Number of obs = 7
-----------------------------------------------------------------------------------------------------------------
Variable | Mean Std. Err. z P>|z| [95% Conf. Interval]
---------------+------------------------------------------------------------------------------------------------
0 | .1833333 .0499537 .0854259 .2812408
1 | .2857143 .1707469 -.0489436 .6203721
---------------+------------------------------------------------------------------------------------------------
diff | -.102381 .1779042 -.4510667 .2463048
| under Ho: .1579452 -0.65 0.517
-----------------------------------------------------------------------------------------------------------------
diff = prop(0) - prop(1) z = -0.6482
Ho: diff = 0

Ha: diff < 0 Ha: diff != 0 Ha: diff > 0
Pr(Z < z) = 0.2584 Pr(|Z| < |z|) = 0.5169 Pr(Z > z) = 0.7416

[Clyde Schechter]

The method for calculating confidence intervals in -prtest- is based on a normal theory approximation. That approximation works poorly when there are proportions that are close to 0 or 1. You can get better confidence intervals with the -ci proportions- command.

[suifang YE]

Thank you very much. I will try the cii command.