Você está na página 1de 24

HYPOTHESIS TESTING

Test

HYPOTHESIS TESTING FOR


POPUOLATION MEAN WITH SMALL
SAMPLES
If small samples i.e., of size n 30 are
drawn from normal population with
mean and for each sample we
compute sample statistic t.
Various use of t distribution are
1. Hypothesis for population mean
2. Hypothesis testing for difference
between two population means with
independent sample
3. Hypothesis testing for difference
between two population means with
dependent samples

HYPOTHESIS TESTING FOR


SINGLE POPULATION MEAN

HYPOTHESIS TESTING FOR SINGLE


POPULATION MEAN(small sample)
Test statistic for determining the
difference between the sample mean x
and population mean is given by
t= x-/sx = x / (s/n )
[if s not given then = (x-x)/n-1
Where s is an unbiased estimation of
unknown population standard
deviation . The test statistic has n-1
degrees of freedom
[Degree of freedom = the number of
values that are free to vary in a
random sample = n-1]

HYPOTHESIS TESTING FOR SINGLE


POPULATION MEAN
Interval estimation estimate of
the population mean when
unknown population std
deviation is estimated by
sample std deviation s is given
by
x tcritical (s/n)

HYPOTHESIS TESTING FOR SINGLE


POPULATION MEAN (small sample)
Decision Rule
Null hypothesis Ho : =o
Alternate Hypothesis
One tailed test

Two tailed test

H1: >o or
<o

H1 : =o

HYPOTHESIS TESTING FOR SINGLE


POPULATION MEAN (small sample)
Decision Rule
Reject Ho at given degrees of freedom
n-1 when

One tailed test

Two tailed test

tcal> t [For
H1:>o] or
tcal < - t [For
H1:< o]

tcal > t/2


or
tcal < - t/2

Example 1
The average breaking strength
of steel rods is specified to be
18.5 thousand kg. For this a
sample of 14 rods was tested.
The mean and standard
deviation obtained were 17.85
and 1.955 respectively . Test the
significance of the deviation.
Critical values of t at df =13 and
/2 =0.025 is t/2 = 2.16

Example 1-Solution
Let us take the null hypothesis that
there is no significant deviation in the
breaking strength of rods that is
Ho: = 18.5 , H1: = 18.5 (two tailed
test)
Given n=14 , x= 17.85, s = 1.955
= 0.05. Critical values of t at df =13
and /2 =0.025 is t/2 = -2.16
t=x-/s/n = 17.85 18.5/ (1.955/14)
= -1.24
Since tcal is more than critical value t/2 ,
the null hypothesis is accepted

Example 1-Solution
Interval Estimation is
= x t/2 (s/n)
Given x= 17.85 , t/2 = 2.16,
s=1.955, n= 14
= 17.85 2.16 (1.955 / 14)
= 17.85 2.16 (.522)
= 17.85 1.12
= 18.94 and 16.73

Example 2
An automobile tyre company
manufacturers claims that the average
life of a particular grade of tyre is more
than 20,000 km when used under
normal condition. A random sample of
16 tyres was tested and a mean and
standard deviation of 22,000 km and
5,000 km respectively were computed.
Assuming the life of tyres in km to be
approximately distributed decide
whether the manufacturer claim is valid
?

Example 2-Solution
Let us take null hypothesis that the
manufacturers claim is valid
Therefore Ho : 20,000 and H1: <
20,000
(left tailed test)
Given n=16, x=22,000 , s= 5000, df =15,
=0.05
t = x-/s/n = 22,000-20,000/500016
= 2000/1250 =1.60
Since tcal is less than its critical value t
=1.753 at =0.05 and df=15, Ho is
accepted

HYPOTHESIS TESTING FOR


DIFFEERNCE OF TWO
POPULATION MEANS
(small samples )

HYPOTHESIS TESTING FOR DIFFERNCE OF


TWO POPULATION MEANS (small sample)
(population variance unknown & equal)
Test statistic is
t = {(x1-x2) / s} { n1 n2/ n1+n2}
[ s= (n1-1)s1 + (n2-1) s2/ (n1+n2-2) ]
Sampling distribution of this t statistic is
approximated by t distribution with n 1+n2-2
degrees of freedom
Null hypothesis and alternate hypothesis
Two tailed Ho : 1- 2 =0
H1 :1- 2 =0
One tailed Ho : 1- 2 =0
H1 : 1- 2 > do
H1 : 1- 2
< do

HYPOTHESIS TESTING FOR DIFFERNCE


OF TWO POPULATION MEANS(small
sample)
(population variance unknown & equal)

Decision Rule
Reject Ho at df = n1+n2-2 and at
specified level of significance when
One Tailed Test
tcal > t or tcal
< -t
For H1:(1-2)
< do

Two tailed test


tcal > t/2 or tcal
< - t/2

HYPOTHESIS TESTING FOR DIFFERNCE


OF TWO POPULATION MEANS (small
sample)

Interval Estimation
= x1-x2 t/2 sx1-x2
= x1-x2 t/2s1/n1 + s2/n2

EXAMPLE 3
The mean life of a sample of 10 electric
light bulbs was found to be 1456 hours
with a standard deviation of 423 hours. A
second sample of 17 bulbs chosen from a
different batch showed a mean life of
1280 hours with standard deviation of 398
hours. Critical value t/2 =2.06 at df =25
Is there a significant difference between
the means of two batches?
Also calculate interval estimation?

EXAMPLE 3 - Solution
Let us take the null hypothesis
that there is no significant
difference between the mean life
of electric bulbs of two batches
that is
Ho:1- 2=0
H1 :1- 2=0
Given n1=10, x1=1456,s1= 423

n2 =17, x2=1280, s2 =398 , =


0.05

EXAMPLE 3 - Solution
Applying t test
t = {(x1-x2) / s} { n1 n2/
n1+n2}
={1456-1280/407.18}
{10x17/10+17} =1.085
Since the calculated value tcal
=1.085 is less than its critical
value t/2 =2.06 at df =25 and
=0.05 the null hypothesis is

EXAMPLE 3 - Solution
Interval Estimation
Interval Estimation
= x1-x2 t/2 sx1-x2
= x1-x2 t/2s1/n1 + s2/n2
= 1456-1280 2.06 423/ 10 +
398/17
=1456-1280 2.06(164.95)
= 176 339.8 = 515.8 and -163.8

Example 4
The manager of a courier service believes that the
packets delivered at the end of the month are
heavier than those delivered early in month. As an
experiment he weighed a random sample of 20
packets at beginning of month . He found that the
mean weight was 5.25 kg and standard deviation
of 1.2 kg. Ten packets randomly selected at end of
the month had a mean of 4.96 kgs and a standard
deviation of 1.15 kg. At 0.05 significance level can
it be concluded that the packets delivered at end
of month weigh more?
At = 0.01 and df =28, the critical value =1.701

Example 4- Solution
Let us take the null hypothesis
that the mean weight of packets
delivered at end of month is
more than the mean weight of
packets at beginning of the
month
Ho :e b
H1: e < b
Given n1= 20, x1=5.25, s1=1.20
n2=10, x2=4.96, s2= 1.15
s= (n1-1)s1 + (n2-1) s2/

Example 4 - Solution
Applying t test
t = {(x1-x2) / s} { n1 n2/ n1+n2}
= (0.29/5.16 ) 200/30 =
0.056x2.58 =0.145
Since at = 0.01 and df =28, tcal
= 0.145 is less than critical value
=1.701 the null hypothesis is
accepted