T F m n

Fm,n T (U/m)/(V /n)

2 2
U V m n
F
H0 : 2 /⌧ 2  1 H1 : 2
/⌧ 2 > 1
2
X1 , ..., Xm Y1 , ..., Yn N (µ, )
2
N (⌫, ⌧ ) µ ⌫

↵0 = 0.05

X1 , ..., X32 Y1 , ..., Y20

x̄ = 163 ȳ = 158 sx = 7.8 sy = 9.0
 5%

↵0 = 0.1

X1 , ..., X10 Y1 , ..., Y10

x̄ = 101.5 ȳ = 99.2
s2x = 5.1 s2y = 5.2
 5%

↵0

↵0 = 0.05

↵0 = 0.05
↵0 = 0.05
Data :

-NoteMs Ogi O asse e B eas we

for the test ,

we assume os are the same
Ho :

Ms = Munl Ms-Mn =0
He :
Ms Mul (smoking leads to
higher had levell
Ms-Mn>0
Test statistic :
T(x =
,
-Y = M Antm-2
⑭
⑧ n
1
-
sample
cause we rescale
with sample variance
mean
Ms=Mn
T =
x -
Y ~turm-2
student table
Sxx + -biggest of in
5 0 95
↓ , ,

ki =

(2 20),
+
=

guno1 20 + I
/trim-e ,
=
/1 68 +2)
.
,

Sx Y (m-1/Sx2+ (n-11542
-

test :
=TCx 1 ,
=

i-y n +
m 2
-

SI In th
52 =
.

107 x3, 32 333 , 3

=
+

=
0, 9-0, 1 =

2 09 108 334 +
-

2
,
in,
92 (5zud 11 92
=

S
.

t 2 09 ckT
=

,
=
reject Ho
Conclusion reject to - there is that
:
statistically significant evidence
blood had level
higher in children
of
smoking mothers

pate" X
, NIM 02) ,
assumptions both unknown

Hypothesis :

at least
-
Ho :

M7, 800 Mo =
Lo =
0, 05

H :

M
< 800 =

Mo

Test statistic :
cause we do not know 0 (we divide
by sxl
M
Mo
=

T(X1 =

X -

Mo I -

⑰
ounknown - use Sx

critical
region
:
KT = 1 -
c
, ty c)
,
=
< 0
,
-tr, -co]
 O I you pair the data - same
experimental
conditions

(i) Data :
↳
Awe assume O= OB
X ,
. .

.,
No NIMA OY ,

independant
NCMB j
*
Y..... Ys
,

Test statistic :
Hypothesis :

Ho :

MA= MB T2x % ,
= X- Y
712-2
0
SxY
Its
He MA F MB
unknown
:

1)V(tane,
Critical KT ( -tanz
region
=
:
-
0
,
,

k= =
1 -
8:
-

to, . sided It 10 0 95 1
+

0)
kT
,
81]u(1
.

=
1 - c
,
-
1,
, 81 +201
,
Test :
1 =
T(x 11,
= X Y
-

X =
57500
* 59000
Say
=

+ = -
0, 4587 ki Sx =

308 -
105
1 - 2
,
-1, 81]v(1 81 tes/
, ,
Sy =

336 185
x

Ayki - retain Ho
Sx
=

(m-11S+ In-11SY-
Conclusion m+n- 2
SxY = 332= 105
*enough significant stat - evidence
Sx y = 332x 105
,

that MA F MB .