2. Definition
Any statement or observation or specification about
population parameter or population is called hypothesis.
Example:
The daily talk time charge of 15 randomly selected customers of Grameenphone
is given below:
110, 118, 130, 140, 142, 146, 112, 100, 95,98,96,122,123,124,130.
Do you think the average talk time of customers in the population is 110?
To a make logical conclusion, on this situation we use
hypothesis testing.
3. Related Terms
Parameter: The unknown characteristic of a population.
Such as- µ,σ.
Statistic: Any function of sample observation. Such as-
𝑥, s.
Null Hypothesis: is the hypothesis which is picked up
for the test. (denoted by Ho)
Alternative Hypothesis: is any statement against the
null hypothesis. (denoted by Ha)
If, Ho: µ= µo Then, Ha: (i) µ≠µo or (ii) µ<µO (iii) µ>µO
Level of significance or Error: The probability level for
which the null hypothesis is rejected even it is true.
(denoted by 𝛼 )
4. Cont.
Test statistic: The function of sample observations
which is used to verify the null hypothesis.
Three cases for test statistic:
i. If the population variance is known
T statistic: z =
𝒙−µ
σ/√𝒏
ii. If the population variance is unknown but large sample
(≥ 30)
T statistic: z =
𝒙−µ
σ/√𝒏
5. Cont.
iii. If the population variance is unknown and small
sample.(<30)
T statistic: t =
𝒙−µ
𝐬/√𝐧
Degrees of Freedom (d.f.): The number of independent
values which are used in deriving a test statistic.
Formula: ( n-1) n= Total size of sample.
6. Cont.
One Tailed and Two Tailed Test:
(i) HA: µ>µO
(ii) HA: µ<µO
(iii) HA: µ≠µO
When, (i) and (ii) condition appears ,it will be called one
tailed test.
When, (iii) condition appears, it will be called two tailed test.
See T- Table
8. Acceptance Region & Critical
Region
The decision regarding the null hypothesis is made as
follows:
1. If T- statistic ≥ Zά/2,(n-1) , HO is rejected in favour of HA: µ<µO
2. If T- statistic (-)≤ -Zά/2,(n-1) , HO is rejected in favour of HA : µ>µO
3. If T- statistic ≥ Zά/2,(n-1 , HO is rejected in favour of HA : µ≠µO .
Zά/2,(n-1) = Acceptance region. This value can be sought out with the
help of d.f. and level of significance from the T- table.
9. Steps of Hypothesis Testing
Develop the null hypothesis and alternative hypothesis.
Specify the level of significance.
Select the test statistics that will be used to test the
hypothesis.
Collect the sample data and compute the value of the
test statistic.
Use the level of significance to determine the critical
value.
Reject the null hypothesis if the test statistic value is
greater than the critical value.
Go to previous example!
10. Types of Hypothesis Testing
Test regarding mean
Test of equality of two means
Test of equality of two correlated means
Test regarding proportion
Test regarding proportion with independent sample
Test regarding variance etc.
11. Test regarding mean
HO : µ= µo against
HA : µ≠ µo
T- statistic, t =
ẋ−µ
𝐬/√𝐧
z =
ẋ−µ
σ/√𝒏
12. Test of equality of two means
HO : µ1= µ2 against
HA : µ1≠ µ2
T- statistic, Z=
t =
ẋ1− ẋ2
𝑠(
1
𝑛1
+
1
𝑛2
Where, 𝑆2 =
𝑛1 −1 𝑠1
2
𝑛1+𝑛2−2
+
(𝑛2−1)𝑆2
2
𝑛1+𝑛2−2
ẋ1− ẋ2
𝜎1
2
𝑛1
+
𝜎2
2
𝑛2
13. Test Equality of Two Correlated Mean
HO : µ1= µ2 against
HA : µ1≠ µ2
𝑑𝑖= 𝑥𝑖 − 𝑦𝑖 Mean 𝑑= 𝑖=1
𝑛
𝑑 𝑖
𝑛
, Variance, 𝑆 𝑑
2
=
(𝑑 𝑖− 𝑑)2
𝑛−1
Test statistic, t =
𝑑
𝑠 𝑑
√𝑛
14. Test regarding proportion
HO : P= PO (given) against
HA : P≠ PO
Sample proportion, p=
𝑎
𝑛
=
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑒𝑙𝑒𝑐𝑡𝑒𝑑 𝑖𝑡𝑒𝑚𝑠
𝑠𝑖𝑧𝑒 𝑜𝑓 𝑠𝑎𝑚𝑝𝑙𝑒
T- statistic, Z=
𝑝−𝑃 𝑂
𝑃 𝑜𝑄 𝑜
𝑛
where, (Qo =1- Po )
Level of significance 5%
15. Test regarding proportion with independent sample
HO : P1= P2 (given) against
HA : P1 ≠ P2
P1=
𝑎1
𝑛1
= proportion of population-1
P2=
𝑎2
𝑛2
= proportion of population -2
T- statistic, Z =
𝑃1−𝑃2
𝑃𝑄(
1
𝑛1
+
1
𝑛2
)
Where, P =
𝑎1+𝑎2
𝑛1+𝑛2
Q = (1-P)
16. Test regarding variance
HO : 𝜎2
= 𝜎𝑜
2
against
HA : 𝜎2
≠ 𝜎𝑜
2
Since, 𝜇 𝑖𝑠 𝑢𝑛𝑘𝑛𝑜𝑢𝑛 , 𝑠2=
(𝑥 𝑖− 𝑥)2
𝑛−1
T- statistic, χ2 =
(𝑛−1)𝑠2
𝜎 𝑜
2
χ2 is distributed as chi- square with (n-1) d.f.