You are on page 1of 22

STATA

one-sample t-test
independent t-test/paired t-test
.


Email: nikom@kku.ac.th
WebPage: http://home.kku.ac.th/nikom

$ $

Research Process

$
1.

(objective)
2.
(study design)


3. Assumption
- -
-

K 1
-
One Sample t-test

t=

x
s/ n

95 %CI = x t / 2 , df (s/

n)

Assumption
-
- interval ratio scale

K 1
K
One Sample t-test t = x ; df = n 1

s/ n

Assumption Wilcoxon Sign Rank Test

-
- 7 sysBP 40
bmi 20-25 120 mmHg

: 7 sysbp
40 120 mmHg
120 130 130 140 140 140 150 150 160

. swilk sysbp
Shapiro-Wilk W test for normal data
Variable |
Obs
W
V
z
Prob>z
-------------+------------------------------------------------sysbp |
9
0.99549
0.066
-3.538 0.99980

$ sysbp
40

bmi 20-25 120 mmHg
1.
H0 : = 120
HA : 120
2. 0.05
3.
x
t=
s/ n

3.


t=

x
=
s/ n

140 120
12 . 25 / 9

= 4 . 899

120 130 130 140 140 140 150 150 160


. su
Variable |
Obs
Mean
Std. Dev.
Min
Max
-------------+----------------------------------------------------sysbp |
9
140
12.24745
120
160

95 %CI = x t / 2 , df (s/

n)

( t.05 / 2 ,9 )

95 %CI = 140 2 . 26 ( 12 . 25 /

9)

. ttest sysbp == 120


One-sample t test
-----------------------------------------------------------------------------Variable |
Obs
Mean
Std. Err.
Std. Dev.
[95% Conf. Interval]
---------+-------------------------------------------------------------------sysbp |
9
140
4.082483
12.24745
130.5858
149.4142
-----------------------------------------------------------------------------mean = mean(sysbp)
t =
4.8990
Ho: mean = 120
degrees of freedom =
8
Ha: mean < 120
Pr(T < t) = 0.9994

Ha: mean != 120


Pr(|T| > |t|) = 0.0012

Ha: mean > 120


Pr(T > t) = 0.0006

P-value T=4.8990 ;df=9-1 ;p-value = ?


5.
4.

 sysBP 40 bmi 20-25

95% ci

120 mmHg

130.59-149.41 mm.Hg

%CI

H0
1
2
3

CI

 = non-significant

(2) H0

 = significant

(1,3) H0

%CI

H0 : = 120
120
130.59

149.41
CI

130.59-149.41 mm.Hg
95% ci
7 H0 = 120 $$ = significant

$
ttest varname , mu= #
varname = K
#
= KK

sysbp

. ttest sysbp == 120


One-sample t test
-----------------------------------------------------------------------------Variable |
Obs
Mean
Std. Err.
Std. Dev.
[95% Conf. Interval]
---------+-------------------------------------------------------------------sysbp |
9
140
4.082483
12.24745
130.5858
149.4142
-----------------------------------------------------------------------------mean = mean(sysbp)
t =
4.8990
Ho: mean = 120
degrees of freedom =
8
Ha: mean < 120
Pr(T < t) = 0.9994

Ha: mean != 120


Pr(|T| > |t|) = 0.0012

Ha: mean > 120


Pr(T > t) = 0.0006

7 sysBP 40 bmi 20-25


120 mmHg
95% ci 130.59-149.41 mm.Hg

$ 2 $
$
-Independent t-test

2

2


F-Test
-

K 2 K
K
(independent t-test pooled variance)

x x
1 2
t=
1 1
S
+
p n n
1 2

Standard error

n s2 + n s2
2 2
S = 11
n +n 2
1
2
2
p

2
(x

x
)
n
S 2i = i
i = 1 n 1

df = n + n 2
1 2

K 2 K
K
(Independent t-test unequal variance)

Standard error

x x
t= 1 2
s12 s 22
+
n n
1 2

df =

s12
n1

n (x x) 2
Si = i
i = 1 n 1
2

s 22
n2

s22
n2

n2 1

s12
n1

n1 1

95%CI = ( x1 x2 ) t/ 2,df (se)


df = n + n 2
1 2

df =

s12
n1

2
s12
n1

n1 1

+
+

s 22
n2

2
s22
n2

n2 1

(Assumption)
-

Assumption Wilcoxon Rank Sum Test
Mann-Whitney Test


F-Test

H 0 : 12 = 22 , H A : 12 22

0.05
2

s1 2 2
F = 2 ; s1 > s2 ; df1= n11,df2 = n2 1
s2
70.447762
F=
= 3.011
2
40.5984

10

 chol 40


:
 ( 2 )

( 1)
idno

gr

chol

154

262

299

438

176

193

193

336

336

10

245

 chol 40

( 1)



( 2 )

1.7
H0 : 1 = 2

H 0 : 1 2 = 0

HA :

1 2

H A : 1 2 0
2. 0.05

11


3.

variance
. sdtest chol, by(gr)
Variance ratio test
-----------------------------------------------------------------------------Group |
Obs
Mean
Std. Err.
Std. Dev.
[95% Conf. Interval]
---------+-------------------------------------------------------------------0 |
5
330.8
31.55852
70.56699
243.1795
418.4205
1 |
5
195.6
18.07927
40.42648
145.4039
245.7961
---------+-------------------------------------------------------------------combined |
10
263.2
28.31443
89.53807
199.1483
327.2517
-----------------------------------------------------------------------------ratio = sd(0) / sd(1)
f =
3.0470
Ho: ratio = 1
degrees of freedom =
4, 4
Ha: ratio < 1
Pr(F < f) = 0.8470

Ha: ratio != 1
2*Pr(F > f) = 0.3060

Ha: ratio > 1


Pr(F > f) = 0.1530

independent t-test pooled variance

x x
1 2
t=
1 1
S
+
p n n
1 2

Standard error

2 + n s2
n
s
2 2
S2p = 1 1
n +n 2
1 2
2
(x

x
)
n
S 2i = i
i = 1 n 1

df = n + n 2
1 2

12

Manual K
. su chol if gr==1
Variable |
Obs
Mean
Std. Dev.
Min
Max
-------------+----------------------------------------------------chol |
5
195.6296
40.5984
153.9819
262.3016
. su chol if gr==0
Variable |
Obs
Mean
Std. Dev.
Min
Max
-------------+----------------------------------------------------chol |
5
330.8275
70.44776
245.3274
437.8791

t=

330.8 195.6
= 3.7
1 1
S
+
p 5 5

4. P-value t-value = 3.7 ;df = 5+5-1


;p-value <.05
5.
 chol 40

13


. swilk chol if gr==0
Shapiro-Wilk W test for normal data
Variable |
Obs
W
V
z
Prob>z
-------------+------------------------------------------------chol |
5
0.97679
0.274
-1.383 0.91674
. swilk chol if gr==1
Shapiro-Wilk W test for normal data
Variable |
Obs
W
V
z
Prob>z
-------------+------------------------------------------------chol |
5
0.91232
1.035
0.046 0.48164

variance F-TEST
7 sdtest varname , by(groupvar)

14

chol

gr

. sdtest chol, by(gr)


Variance ratio test
-----------------------------------------------------------------------------Group |
Obs
Mean
Std. Err.
Std. Dev.
[95% Conf. Interval]
---------+-------------------------------------------------------------------0 |
5
330.8
31.55852
70.56699
243.1795
418.4205
1 |
5
195.6
18.07927
40.42648
145.4039
245.7961
---------+-------------------------------------------------------------------combined |
10
263.2
28.31443
89.53807
199.1483
327.2517
-----------------------------------------------------------------------------ratio = sd(0) / sd(1)
f =
3.0470
Ho: ratio = 1
degrees of freedom =
4, 4
Ha: ratio < 1
Pr(F < f) = 0.8470

Ha: ratio != 1
2*Pr(F > f) = 0.3060

Ha: ratio > 1


Pr(F > f) = 0.1530

15

$
ttest varname1 == varname2 , unpaired [unequal]
ttest varname , by(groupvar)

16

. ttest

chol, by(gr)

Two-sample t test with equal variances


-----------------------------------------------------------------------------Group |
Obs
Mean
Std. Err.
Std. Dev.
[95% Conf. Interval]
---------+-------------------------------------------------------------------0 |
5
330.8
31.55852
70.56699
243.1795
418.4205
1 |
5
195.6
18.07927
40.42648
145.4039
245.7961
---------+-------------------------------------------------------------------combined |
10
263.2
28.31443
89.53807
199.1483
327.2517
---------+-------------------------------------------------------------------diff |
135.2
36.37032
51.3299
219.0701
-----------------------------------------------------------------------------diff = mean(0) - mean(1)
t =
3.7173
Ho: diff = 0
degrees of freedom =
8
Ha: diff < 0
Pr(T < t) = 0.9971

H 0 : 1 = 2

Ha: diff != 0
Pr(|T| > |t|) = 0.0059

Ha: diff > 0


Pr(T > t) = 0.0029

H A : 1 2

95 %CI = ( x1 x 2 ) t / 2 , df (se)

%CI

H0 : 1 = 2
0
H 0 : 1 2 = 0
51.33

219.07
CI

51.33-219.07 mm.Hg
95% ci
7 H0 = 0 $$ = significant

17

K 2
K
K paired t-test

t=
Standard error

di

2
(d

d
)
n
S 2i = i
i = 1 n 1

sd
n

df = n - 1

95 %CI = d t / 2 , df (se)

(Assumption)

-
(di)

cho1

cho2

300

240

-60

320

300

-20

330

330

330

300

-30

340

312

-28

Assumption Wilcoxon match pair sign rank

18

: chol 40
3
cho1

cho2

300

240

-60

320

300

-20

330

330

330

300

-30

340

312

-28

chol 40

1. `
H0 : d = 0

HA :

d 0

2. 0.05

19

3.
. swilk d
Shapiro-Wilk W test for normal data
Variable |
Obs
W
V
z
Prob>z
-------------+------------------------------------------------d |
5
0.95382
0.545
-0.721 0.76445

. su d
Variable |
Obs
Mean
Std. Dev.
Min
Max
-------------+----------------------------------------------------d |
5
-27.6
21.65179
-60
0

t =

di
sd
n

27.6
21.7
5

= 2.9

4. P-value T=-2.9 ;df=5-1;p-value <.025


5.
chol 40

$
ttest varname1 == varname2

20

. ttest cho1 == cho2


Paired t test
-----------------------------------------------------------------------------Variable |
Obs
Mean
Std. Err.
Std. Dev.
[95% Conf. Interval]
---------+-------------------------------------------------------------------cho1 |
5
324
6.78233
15.16575
305.1692
342.8308
cho2 |
5
296.4
15.13142
33.83489
254.3884
338.4116
---------+-------------------------------------------------------------------diff |
5
27.6
9.682975
21.65179
.7157522
54.48425
-----------------------------------------------------------------------------mean(diff) = mean(cho1 - cho2)
t =
2.8504
Ho: mean(diff) = 0
degrees of freedom =
4
Ha: mean(diff) < 0
Pr(T < t) = 0.9768

Ha: mean(diff) != 0
Pr(|T| > |t|) = 0.0464

Ha: mean(diff) > 0


Pr(T > t) = 0.0232

95 %CI = d t / 2 , df (se)
chol 40

7 95% CI 0.72-54.48

21

%CI

H0 : d = 0
0
.72

54.48
CI

.72-54.48
95% ci
7 H0 = 0 $$ = significant

%CI

H0 : d = 0
0
-54.48

-.72
CI

-54.48 -.72
95% ci
7 H0 = 0 $$ = significant

22

You might also like