* Chapter 7 Comparison of two means: confidence intervals, hypothesis tests and p-values 
* modified on Jan. 13, 2007 by Leena Choi

* Section 7.3 Example 7.1
input group n x_mean sd
1 36 4.7 0.6
0 64 5.0 0.6
end

gen z = invnormal(.975)
gen diff = x_mean[1] - x_mean[2]
gen se = sqrt(sd[1]^2/n[1] + sd[2]^2/n[2])

gen l_ci = diff - z*se
gen u_ci = diff + z*se
list l_ci u_ci in 1

*find 10%, 1% percent point
gen z_score = diff/se
* one-sided p-value
di normal(z_score)
* two-sided p-values
di 2*normal(z_score)

* Section 7.4 Table 7.2
* Example 7.2
use "C:\Teaching\IGP\data\birthweight.dta", clear
label data "Comparison of birth weights (kg) of children born to 14 heavy smokers with those of children born to 15 non-smokers." 
notes: "Data from EMS Table. 7.2"
notes: "Created on Jan. 14, 2007"
*label define smokelab 1 "Smokers" 0 "Non-smokers"
label values smoke smokelab

ttest weight, by(smoke)
sdtest weight, by(smoke)
di invttail(14+15-2, .025)

ttest weight, by(smoke) unequal

* Section 7.6 Table 7.3
label data "A placebo-controlled clinical trial to test the effectiveness of a sleeping drug" 
notes: "Data from EMS Table. 7.3"
notes: "Created on Jan. 14, 2007"

di invttail(9, .025)

use "C:\Teaching\IGP\data\sleepingdrug.dta", clear
ttest drug=placebo