Paried T-Test
In a matched pairs study, subjects are
matched in pairs and each treatment is given to one subject in
each pair randomly. The outcomes are compared within each other.
Another situation calling for matched pairs is before-and-after
observations on the same subjects. Paired t test is applied
to procedure the observed differences.
In Exercise
7.21 (Page 386 chapter 7 BPS) the healing rates of newts are studied.
In this experiment, each newts got razor cut in both hind limbs,
let one heal natually(the control) and change another one's electric
field to its half value(expermental). Two hours after, measure
the healing rate of both limb.Data are shown in worksheet.
They want to know if changing the electric
field reduces the mean healing rate for newts. Because the both
treatment and control were did on one subject, so we got paired
data here. To address the question we will test H0:
u=0 Ha: u<0. The 
denotes the mean of difference between experimental and control
data.
To do the
test we need to select Stat->Basic Statistics->Paired
t from the menu. Fill out the dialog box and then specify
the null hypothesis test value and alternative hypothesis by clicking
the Option button and filling out the Option subdialog
box also as shown below. Pay attention to the word on the
Paired t dialog box "Paired t evaluates the first sample
minus the second sample."
The result show that the P-value is 0.032, which is quite small, so we can reject null hypothesis.
