Testing of Hypothesis

Single sample test

Dr. T. T. Kachwala

1
Objective of Single Sample test for Arithmetic Mean

We can use X as an estimator of μ and it can be shown that X has all the
desirable properties of an estimator. In particular, X is an unbiased and
efficient estimator of μ.

μ (Population mean)

X (Sample mean)

The objective of the single sample test is to assess whether or not there is a
significant difference between Sample Mean X & hypothesized population
Mean μ H
0

2
Step 1: Setting up a Hypothesis Statement (Define H & H ) 0 1

Define Null Hypothesis H0

In hypothesis testing, we must state the assumed or hypothesized value of

the population parameter before we begin sampling. The assumption we
wish to test is called the null hypothesis and is symbolized H0.

If we use a hypothesized value of population mean in a problem, we

would symbolically represent it as ‘ μ H0 ’
i.e. H0 :   μ H0

Suppose, we want to test the hypothesis that the population mean is equal
to 75. We would symbolize it as follows: H0 :   75

3
Why is it called null hypothesis?

The term null hypothesis arises from earlier agricultural and

medical applications of statistics.

In order to test the effectiveness of a new fertilizer or drug, the

tested hypothesis (the null hypothesis) was that ‘it had no effect’
i.e. the tested hypothesis (the null hypothesis) was that there was
no difference between treated and untreated samples.

The approach is to start with the hypothesis that drug has no

effect. Using the testing procedure prove that your assumption is
false, i.e. Reject H0 & in the process Accept Alternate hypothesis
H1 that the drug is effective.

4
Alternate Hypothesis

Whenever we reject the null hypothesis, the conclusion we do accept is called

the alternative hypothesis and is symbolized H1:

For H0:  = 75

The following are the three possible alternative hypothesis:

H1:   75 Population mean is not equal to 75.

H1:   75 Population mean is greater than 75.

H1:   75 Population mean is less than 75.

5
Selecting Alternate Hypothesis & Procedure of Test

The selection of alternate hypothesis is based on the objectives of the

research.

Example 1: if the researcher does not want to deviate significantly from

 = μ H0 in either direction, then the appropriate alternate hypothesis is
H1:   μ H0 and the corresponding procedure is two tailed test.

Example 2: if the researcher does not want to deviate significantly from

 = μ H0 in the left direction, then the appropriate alternate hypothesis is
H1:   μ H0 and the corresponding procedure is left tailed test.

Example
μ H0 3: if the researcher does not want to deviate significantly from
 = in right direction, then the appropriate alternate hypothesis is
μ H0
H1:   and the corresponding procedure is right tailed test.

6
Step 2: Defining critical (table) values

Defining critical value or table value depends on:

(1) Level of Significance

(2) Probability Distribution (Normal or t distribution)

(3) Single Tailed test or Two Tailed test.

7
Selecting the Level of Significance  (alpha)

The purpose of hypothesis testing is not to question the computed

value of the sample statistic but to make a judgment about the
difference between the sample statistic and a hypothesized population
parameter.

The next step after stating the null and alternate hypothesis is to
decide what criterion to use for accepting or rejecting the null
hypothesis. The confidence with which a researcher rejects or accepts
a null hypothesis depends on the significance level selected. It is
customarily expressed as a percentage such as 5% and is indicated by
letter  (alpha).
8
Meaning of Level of Significance

1. Level of Significance α mathematically = 1 – Confidence level,

For example 95% confidence level is the same as 5% level of
significance

2. Level of Significance α signifies the total area under the tail of the
distribution. It signifies the size of the rejection area (total area of the
two tails for a two tailed test).

3. Level of Significance also signifies the probability of committing

Type I error. For e.g.:    simply means that the total rejection
area is 5% & that there is a 5% chance of committing Type I error. 9
Type I and Type II Errors

There are two type of errors in testing hypothesis as tabulated below:

States  Decision 
Accept H0 Reject H0

Correct Accept Type I

H0 True Decision Error

Type II Correct Reject

H0 False Error Decision
Type I & Type II Error

Type I error is the error committed by way of rejecting a True hypothesis.

For e.g.:   .  There is a 5% chance of rejecting a true hypothesis.
(Alternatively, a true hypothesis will be accepted 95% of the time – Confidence
level).
  .  5% level of Significance is the same as 95% level of
Confidence.
  .1  1% level of Significance is the same as 99% level of Confidence.
Type II error is the error committed by way of accepting a false hypothesis

There is a tradeoff between these two errors: The probability of making one type
of error can be reduced only if we are willing to increase probability of making
the other type of error.

To deal with this tradeoff in professional situations, decision makers decide the
appropriate level of significance by examining the cost or penalties attached to
both types of errors.
11
Examples of tradeoff of Type I & Type II Error

i) Suppose Type I error involves reworking of chemicals by manufacturer

and Type II error involves user of chemical getting poisoned, then
manufacturer will prefer Type I error to Type II error and will set a very
high level of Significance.

ii) Suppose Type I error involves expensive disassembling an entire engine

by manufacturer at factory and Type II error involves inexpensive
repairs by the customer then manufacturer will prefer Type II error and
set a low level of Significance.

12
Selecting the Distribution for testing of hypothesis:

After deciding the level of significance we must determine the

appropriate probability distribution. We have a choice between
Normal distribution & t - distribution and we use guidelines similar
to those used in estimation.

1. Normal distribution if  is known

2. Normal distribution if n >  and if  is not known

3. ‘t’ if n   and if  is not known

13
Selecting Two Tailed / Single Tailed

The selection of Two Tailed / Single Tailed is based on the alternate

hypothesis.

H1:  
μ H0 and the corresponding procedure is two tailed test.

μ H0
H1:   and the corresponding procedure is left tailed test.

μ H0
H1:   and the corresponding procedure is right tailed test.

14
Examples of critical (table) values

Using Standard tables, Compute the table values for:

1.  , Normal distribution, & two tailed

2.  , t distribution, & two tailed (Degree of Freedom 

= 15)

3.  1, Normal distribution, & right tailed

4.  , t distribution, left tailed (Degree of Freedom  =

15)

15
Step 3: Calculation of test statistics (z or t)

Step (iii) Calculation of test statistics depends on type of distribution

X - μ H0
z
x


x 
n
or
X - μ H0
t
ˆ x

s
ˆ x  16
n
Step 4 & 5: Decision Rule & Conclusion

Assuming Normal Distribution & two tailed test, the decision rule can
be illustrated diagrammatically for the following two possibilities (as
indicated below):

x x
z stat z stat
Accept H0 Reject H0
Accept H1
There is no significant There is a significant
difference between Sample difference between Sample
Mean X & μ H0 Mean X & μ H0 17
Summary of the Testing Procedure for Single Sample test for Mean

The following is the summary of the testing procedure:

Step (i) Define H0, H1 (depending on one tailed or two tailed test).

Step (ii) Decide critical values corresponding to distribution (normal

or t) and level of Significance and Single or Two Tailed test.

Step (iii)Calculate Standard error  x & the standardized value z or t

statistics

Step (iv) Decision Rule (Sketch the distribution and mark the position
of Standard value and critical values).

Step (v) Conclusion and Interpretations of the results.

18
Testing Procedure for Single Sample test for Mean using Normal
distribution ( z test)

Assuming population Standard Deviation is known:

Step (i) H 0 : μ  μ H0
H1 : μ  μ H0
(Assuming two tail test)

Step (ii)   , Normal distribution, two tailed

ztable = 

Step (iii) X - μ H0
z
x


x  19
n
Testing Procedure for Single Sample test for Mean using Normal
distribution ( z test)

Step iv & v : Decision Rule & Conclusion

The following are the two possibilities as indicated in diagram below:

x x
z stat z stat
Accept H0 Reject H0
There is no significant Accept H1
difference between Sample There is a significant
Mean X & μ H0 difference between Sample
Mean X & μ H0 20
Testing Procedure for Single Sample test for Mean using t distribution ( t test)

Note: If Sample Size is larger than 30, we continue to apply Normal

distribution. However, if n  30 we apply t  Distribution.
Step (i) H 0 : μ  μ H0

H1 : μ  μ H0

(Assuming two tail test)

Step (ii) t distribution (critical value depends on Degree of Freedom , level

of significance  and single tailed or two tailed test).

21
Testing Procedure for Single Sample test for Mean using t distribution ( t test)

Step (iii) Calculation of test statistics

X - μ H0
t
ˆ x

s
ˆ x 
n

x 2
s
n -1

22
Testing Procedure for Single Sample test for Mean using t distribution ( t test)

Step (iv)& (v) Decision Rule & Conclusion: There are two possibilities as
indicated in diagram below:

x x
t stat t stat
Accept H0 Reject H0
Accept H1
There is no significant
There is a significant
difference between Sample
difference between Sample
Mean X & μ H0
Mean X & μ H0 23
A word of Caution

If Sample statistic falls in the acceptance region, it does not prove that
our H0 is true. It simply means that there is no statistical evidence to

reject it. Therefore, whenever we say we accept the null hypothesis,

we actually mean that there is no sufficient statistical evidence to
reject it.

The only way in which ‘H0’ can be accepted with certainty is for us to

know the population parameter, which unfortunately is not possible.

There is no single standard level of significance for testing hypothesis.

However, published research results are often tested at 5% and 1% 24
Objective of Single Sample test for Proportion of success

We can use p as an estimator of p and it can be shown that p has all the
desirable properties of an estimator. In particular p is an unbiased and
efficient estimator of p.

P (Population Proportion of Success)

p (Sample Proportion of Success)

The objective of the single sample test is to assess whether or not there is a
significant difference between Sample proportion of success p &
hypothesized population proportion of success p H0

25
Hypothesis Testing for Proportion of Success

Single Sample test Summary Procedure for proportion of success:

Step (i) H 0 : p  p H0 H1 : p  p H0 ( assuming two tailed)

Step (ii)    0.05, two tailed, Normal   z critical  1.96

Step (iii) Calculate  p and z statistics

p H0 q H0
p 
n
p - p H0
z stat 
p
26
Hypothesis Testing of Proportions

Step (iv)& (v) Decision Rule & Conclusion: There are two
possibilities as indicated in diagram below:

x x
z stat z stat
Accept H0 Reject H0
If z statistic is in the Accept H1
acceptance area, If z statistic is in rejection area
Accept H0: p = pH0 Reject H0: p = pH0
Conclusion: that there is no Accept H1: p ≠ pH0
significant difference between Conclusion: there is a significant
statistic p and hypothesized difference between sample statistic p
and hypothesized population proportion 27
population proportion of
of success pH0
Hypothesis Testing of Proportions

On similar lines single tailed test can be summarized as follows:

Left Tailed test

H0: p = pH0
H1: p < pH0

Right tailed test

H0: p = pH0
H1: p > pH0

28
Thanks and Good Luck

Dr. T. T. Kachwala

29