version 9.2

log using C:\Teaching\IGP\log\13.transformation.log, replace

* Chapter 13 Transformation
* modified on Feb. 13, 2007 by Leena Choi

***************************************************************************************
* Section 13.2
***************************************************************************************

* Data in Table 13.1
use "C:\Teaching\IGP\data\betaTG.dta", clear

* put all label and define values
label variable diabetics "Diabetics"
label define diabeticslab 0 "Normals" 1 "Diabetics"
label values diabetics diabeticslab

* draw dotplot to compare the spreadness of measurements in each group
dotplot betaTG, over(diabetics)

* take log transformation (defaul is natural logarithm)
gen lbetaTG = log(betaTG)

* dotplot again using the log transformed values
dotplot lbetaTG, over(diabetics)

* observe the s.d of the values for the orinal scale and the log transformed
sort diabetics
by diabetics: sum betaTG
by diabetics: sum lbetaTG

* run t-test using the transformed values
ttest lbetaTG, by(diabetics)

* transform back the mean for the normal subjects to the orinal scale
di exp(2.433349)
di exp(2.811517)
di exp(2.055181)

* transform back the difference of menas to the orinal scale
di exp(-.9572788)
di exp(-.4353973)
di exp( -1.47916)

log close