means u, and 42 of the
"Tesii whether thesignificantly2). two
totest
We want
(i.e,
eual or not.
dotheydiffer
populations S
th
It has been proved mathematically that the standard error of (1-
whereS²=49t
tn -2
freedom of the
number of degrees of statistic.
isthe for testing the significance of the difference beetween t
The test statistic
information is given by
means based on sample
population
t=
|1 1
From t-table, we can determine, the table value of t, for ( +nn -2) dt.
Ifthe calculated value of <the table. value ,we accept the
hypothesis H,
If the calculated value of > the table value, we reject th
hypothesis H,
Note: The above procedure is only for a two-tail test. In the case of one-al
test, we state the hypothesis as follows:ethoa 5ibn h
(OR)
In this case, the table value of t for the significance level a is got
to the column 0.05 probability for 5% and reterk
0.01
conclusion is drawn on the same lines as given probability for 1% leve
Example 1: earlier:
Two salesmen Aand Bare working in a certain district. From asampk
survoy conducted by the Head office, the following results were [Link]
whether there is
two salesmen: any significant difference in the average betweel
sales
No. of sales A B
Average sales (in Rs.) 20 18
Standard De viation (in Rs.) 170
20
205
25
2616
Statistical and Numerical
samples. Let us apply t-test forMethods
Solution:
given samples are small
Thetwo
themeans.
(there is no significant testing
Ho:M1= ,
salesmen).
difference average sales ofthe two
in the
( difference in the average sales of two salesmen is
H: =205
F=170
S =25
significant)
G=
20
n=20 n, =18
statistic is
The test
-
+where s S ty
ntn-2
20 x20 +18× 25 8000+11250 19250
20+18-2 36 36
s= 534.72. .. S=23.12
170-205 -35
.t=
23.12,/+ 23.12 x0.325 35
-=4:65
20 18 7.52I
ndf =n + n, -2 = 36.
level=2.58.
Table value of tfor 36 df at 1%
Conclusion:
1% level since the calculated value of>the table
H, is rejected at
two salesmen differ significantly with regard to their average
Value. Hence the
sales.
Example 2: tested for their mean life. Assuming
Two batches of the same product are distribution with an unknown
follow a normal
that 'the lives of the product both the batches,
for
lifeisthe same
variance, test the hypothesis that the mean S.D (in hrs.)
given the following information: Mean life (in
hrs.) 12
Batch Sample size 750 14
10 820
1 8
oftat5%level 18
value
Choose a Significance level of 5% ; 17
2.101
df 16
14 15 2.110
2.120
2.145 2.131
t Test .
Solution: 2617
The two given samples are small samples and so we apply1t-test for
the means.
o Ho:u = u, (mean life is the same for both the batches) testing
batches)
H:u, *u, (mean life is not the same for both the
x=170 , =205
S = 20 n, = 18
n = 10 n, = 8
SE of the difference between the means is
2 2
S=
ntn-2
|10x 144+8x 196 1440+ 1568
10+8-2 16
2LEAO008
3000 =V188 =13.71 0S
16
The test statistic is
750-820
1 1
10 18
-70 -70
13.71 x 0.474 6.5
=-10.77,=10.77tha a osiley altist
ndf =n +n,-2 16
Table value of tfor 16df at 5% level =2.12.
Conclusion:
H, is rejected since the calculated value of > the table value.
.. The mean life of the two batches is not the same.
Example 3:
Two samples of 6 and 5 items respectively gave the following data.
Mean of the first sample 40
SD of the first sample 8
Mean of the second sample 50
SD of the second sample 10
Is the difference between the means significant? The value oftfor 9dfat5%
level is 2.26.
2618
Solution:
given samples are small Statistical and Numerical
Thetwo
sampl es. Let us apply Methods
mean.
HÍMi =Hz
(The vloviiaet
means of t
the two t-test for testing the
(The means ofthe
two populations areare equal).
m=6 populations not equal)
i= 40 , =50
S=8 S =10
The SE ofthe difference between the means is
2 2
S=
6x64 +5x 100 384+500
6+5-2 9
884
=/98.22 =9.91
The test statistic is
40-50
t=
1 1
S
-10 -10
9.91x 0.61 6.05
=-1.65,j| =1.65
ndf= n n, -2 = 9.
Table value of t for 9df at 5% level = 2.262.
Conclusion: significant differencebetween
,1S accepted at 5% level. Hencethere is no
the means of the populations.
length of lifeandthe
Example 4: weretested for
Samples of two types of electric bulbs
fol owing data were obtained. Tlype- I
Type- I 7
No. of samples 8
I134
1024
Mean of the samples (in hrs.) 40
S.D of the samples (in hrs.) 35
t Test
Test at 5% level, whether the difference in sample meanssiis 26.19
significant..(table
value oft for 13 df- 2. 16, for 14 df= 2. 15 and for 15 df-2. 13 at 5% level f
for two
tailareas and are 1.77, 1.76and 1.75 respectively for one-tailarea).
Solution:
Thetwo given samples are small samples and so we apply t-test
for
the equality of the population means.
H =H2
testing
=7
7-1134 =1024
S=35 S = 40
SEof the difference between means is given by
2 2
S=
Yn+-2
8x35 +7x40
= 40.19
8+7-2
The test statistic is
t= Z-4 1134-1024
1 1
40.19+:
8
110 110
40.19 x 0.52 20.9 -= 5.26
ndf =nth-2 = 13.
Table value of tfor 13df at 5%
Coclusion:
level=2.16.
H, isrejected since the
calculated value of t > the table value of t.
i. The two types of electric bulbs
mean life. differ significantly with respect to thel
Example 5:
The average number of articles
and 250 with s.d. 20 and 25 produced by two machines per day are 200
respectively on the basis of records of 25 day
production. Can you regard both the machines equally efficient at 1%
significance? level
2620
Statistical and Numerical
Solution:
I=200
=250 Methods
sj=20 S = 25
4-25
The samplesarare small samples.
difference between the means is
SEofthe
2 2
S=, tn-2
25x 400+ 25 x 625 10000+ 15625
48 48 =23.11
The test statistic is
200-250
f=
S,+
11
23.11 V1 1
25 25
-50 -50
23.11 x 0.28 6.47
=-7.73, l=7.73
ndf =n +n-2 = 25+25-2 = 48
Table value oft at 1% level =2.58.
Coclusion:
rejected since the calculated value of t > the table value of .
H, is the machines as equally efficient.
Hence we can notregard both
Example 6: towns Aand Bregarding a newproduct
Sample of sales in similar. shops in
yielded the following information:
=38, EX=228, =1I
For Town A, X =3.45, 2X
=40, X,=222, n =9 Test at 5%
For Town B, X2 =4.44, X, inthetwo
towns.
sales
Is there any evidence of difference in 2.101).
level, (the value oft at 5%%levelfor 18dfis t - testfor
can apply
SolutTheion:two given samples are
aresmall samples. So we
testing the difference means.
between the
M4=11, X=3.45
t Testtbh selh
6
=
228_(3.45) = 20.73 -11.90 =8.83
11
n, =9, X, =4,44
2
222 e0o
-(4.44)' =24.67- 19.71= 4.96
9
2
n +n,-2
11x 8.83+9x4.96 97.13+44.64
=7.88
l1+9-2 18
..S=2.81
The test statistic is
X-X 3.454.44
S 1 1
2.81x
0.99
2.81x 0.45
0.99
1.26
=-0.786
9)CS8Si
=0.786
ndf= 11+9-2= 18
Table value of t for 18df at 5%
level =2.101.
Conclusion:
H is
value of [Link]-at 5% level since the calculated value of t< the tabit
:. There is no
Example 7:
evidence of difference in
their sales.
Below are
Gain in weightsgiven(inthe gains in weights(bs) of f cows fed on diets Xand Y.
lbs) two
Diet X: 25 32 30
Diet Y: |24 34 2. 3230 24 14 32
Test at 5% level, 42 31 40 30 35
in whether
increase weight (table value oft for diets
the two
32
1Sdf atdiffer
as regards their effect mean
on
5% is
12639
Statistical and Numerical
Solution:
Thesizes of
the samples are less
than 30. So the Methods
the ttest.
Letus apPply samples are soañ samples.
H: calculatethe
mean and S.D off
Let us the two
samples.
(-27)
24
(-32)
25 -8 64
25 34
32 2
13 9 22
30 -10 100
32 5
o m o 25b 30 2 4
24b-3 b o o 9 42 10 100
14 -13vg 16908 31 -1
5 25 40 8 64
32
30 |22 4
32 0 0
35 3 9
266 320 350
189
189 320
X=27 : Y==32
10
7
Let d, =x- 27 ; d, =y- 32
2
266 -0=38
350 -0=35
10
7x38 +10×35 266+350 =41.07
7+10-2 15
S=41.07 =6.41
The test statistic is
27-32
t= X-X,
s 1 1 6.41x,
tTest
-5 -5
3.14
=-1.59 8
6.41 x 0.49
=1.59
ndf= n +n, -2=7+ 10-2 =
15
Table value oft for15dfat 5% level =2.131.
Conclusion:
of t.
The calculated value of t < the table value
: H, is accepted at 5% level.
Hence there is no significant difference between the mean increase in
due to two diets. weight
Example 8:
Two sets of ten students selected at random from : a college were
set was given memory test as they were and the other was given the
taken: one
test after two weeks of training and the scores are given below:
memory
Set A :10 7 98 10 9 6 8
Set B: 12 8 8 10 8 11 9 9 9
Do you think there is any significant effect due to training? (null and
alternative hypothesis should be stated) (Given t-2.10 at df =18, a =0.05)
Solution:
The samples are small samples. Let us apply t-test for the data given.
H:4 <p, (one tail test)
For SetA For Set B
y
10 100 12 144
64
7 49
88 64
64
9 81 10 100
8 64 8 64
10 100 11 121
81 9 81
36 8 64
49 9 81
8 64 9 81
82 688 92 864
X-8.2
10 ; Y=9.2
2624
82
Statistical and Numerical Methods
688 = 68.8-67.24 = 1.56
2
10 10
864 -
2
10
864-84.64 = 1.76
2
2
10x1.56+10x1.76 15.6+ 1.76 33.2
=184
10+10-2 18 18
ndf=18.
Table value ofttfor 18dfat 5% level = 1.845
Conclusion:
His accepted at 5% level since the calculated value of <the able
valueoft.
There is no significant effect due to training.
Example 9: