Probability and Statistics

Testing Hypothesis

Dr. Yasir Ali ([email protected])

DBS&H, CEME-NUST

December 22, 2022

Dr. Yasir Ali ([email protected]) Probability and Statistics

A statistical hypothesis is a conjecture about a
population parameter. This conjecture may or may not be
true.

If we want to decide whether a given coin is biased, we formulate

the hypothesis that the coin is fair, i.e., p = 0.5, where p is the
probability of heads.
If we want to decide whether one procedure is better than
another, we formulate the hypothesis that there is no difference
between the procedures
Such hypotheses are often called null hypotheses.

Dr. Yasir Ali ([email protected]) Probability and Statistics

The null hypothesis states that there is no difference between a
parameter and a specific value, or that there is no difference between
two parameters. It is denoted by H0 .

The alternative hypothesis, symbolized by H1 , states the existence of a

difference between a parameter and a specific value, or states that
there is a difference between two parameters.

Any hypothesis that differs from a given null hypothesis is called an

alternative hypothesis.

For example, if the null hypothesis is p = 0.5, possible alternative

hypotheses are p > 0.7, or p 6= 0.5, or p < 0.5.

Procedures that enable us to decide whether to accept or reject hypotheses

or to determine whether observed samples differ significantly from expected
results are called tests of hypotheses

Dr. Yasir Ali ([email protected]) Probability and Statistics

A researcher reports that the average salary of assistant professors is more than
$42,000. A sample of 30 assistant professors has a mean salary of $43,260. Test
the claim that assistant professors earn more than $42,000 per year.

Dr. Yasir Ali ([email protected]) Probability and Statistics

A researcher reports that the average salary of assistant professors is more than
$42,000. A sample of 30 assistant professors has a mean salary of $43,260. Test
the claim that assistant professors earn more than $42,000 per year.
H0 : µ = $42000 and H1 : µ > $42000 (claim).

Dr. Yasir Ali ([email protected]) Probability and Statistics

A researcher reports that the average salary of assistant professors is more than
$42,000. A sample of 30 assistant professors has a mean salary of $43,260. Test
the claim that assistant professors earn more than $42,000 per year.
H0 : µ = $42000 and H1 : µ > $42000 (claim).

The mean lifetime of 100 bulbs produced by is 1570 hours with a σ =120
hours. Test the claim of company that average life time of bulb is 1600.

Dr. Yasir Ali ([email protected]) Probability and Statistics

A researcher reports that the average salary of assistant professors is more than
$42,000. A sample of 30 assistant professors has a mean salary of $43,260. Test
the claim that assistant professors earn more than $42,000 per year.
H0 : µ = $42000 and H1 : µ > $42000 (claim).

The mean lifetime of 100 bulbs produced by is 1570 hours with a σ =120
hours. Test the claim of company that average life time of bulb is 1600.
H0 : µ = 1600 and H1 : µ 6= 1600

Dr. Yasir Ali ([email protected]) Probability and Statistics

A researcher reports that the average salary of assistant professors is more than
$42,000. A sample of 30 assistant professors has a mean salary of $43,260. Test
the claim that assistant professors earn more than $42,000 per year.
H0 : µ = $42000 and H1 : µ > $42000 (claim).

The mean lifetime of 100 bulbs produced by is 1570 hours with a σ =120
hours. Test the claim of company that average life time of bulb is 1600.
H0 : µ = 1600 and H1 : µ 6= 1600

It is claimed that automobiles are driven on average more than 20,000 kilometers
per year. To test this claim

Dr. Yasir Ali ([email protected]) Probability and Statistics

A researcher reports that the average salary of assistant professors is more than
$42,000. A sample of 30 assistant professors has a mean salary of $43,260. Test
the claim that assistant professors earn more than $42,000 per year.
H0 : µ = $42000 and H1 : µ > $42000 (claim).

The mean lifetime of 100 bulbs produced by is 1570 hours with a σ =120
hours. Test the claim of company that average life time of bulb is 1600.
H0 : µ = 1600 and H1 : µ 6= 1600

It is claimed that automobiles are driven on average more than 20,000 kilometers
per year. To test this claim
H0 : µ = 20, 000 & H1 : µ > 20, 000

Dr. Yasir Ali ([email protected]) Probability and Statistics

 Type I error A type I error occurs if you reject the null hypothesis when it
is true.
Type II error A type II error occurs if you do not reject the null hypothesis
when it is false.

Dr. Yasir Ali ([email protected]) Probability and Statistics

The maximum probability to risk a Type I error is
called the level of significance of the test.

The maximum probability of committing a type I error is denoted by α.

That is,
 
reject H0
α = P (type I error) = P
H0 is true

Dr. Yasir Ali ([email protected]) Probability and Statistics

In practice a level of significance of 0.05 or 0.01 is
customary, although other values are used.

If for example a 0.05 or 5% level of significance is chosen in designing a

test of a hypothesis, then there are about 5 chances in 100 that we
would reject the hypothesis when it should be accepted,
i.e., whenever the null hypotheses is true, we are about 95% confident that
we would make the right decision.

In such cases we say that the hypothesis has been rejected at a 0.05 level
of significance, which means that we could be wrong with probability 0.05.

Dr. Yasir Ali ([email protected]) Probability and Statistics

 One tailed

Two tailed

The critical value divides the area under the probability distribution curve
in rejection region(s) and in non-rejection region.

Dr. Yasir Ali ([email protected]) Probability and Statistics

α = 0.01 (Left-Tailed Test) α = 0.01 (Right-Tailed Test)

α = 0.01 (Two-Tailed Test)

Dr. Yasir Ali ([email protected]) Probability and Statistics
Dr. Yasir Ali ([email protected]) Probability and Statistics
Significant Level and Critical Region

If the significance level α, then the critical region will consist of all
values of Z which are
(i) less than −z α2 and greater than z α2 in case of two-tailed test;

Significane level α Two-tailed test One-tailed test

0.10 ± 1.645 ± 1.28
0.05 ± 1.96 ± 1.645
0.01 ± 2.58 ± 2.33

Dr. Yasir Ali ([email protected]) Probability and Statistics

Find the critical value(s) for each situation and draw the appropriate figure,
showing the critical region.
1 A left-tailed test with α = 0.10.

2 A two-tailed test with α = 0.02.

3 A right-tailed test with α = 0.005.

Dr. Yasir Ali ([email protected]) Probability and Statistics

Find the critical value(s) for each situation and draw the appropriate figure,
showing the critical region.
1 A left-tailed test with α = 0.10.

2 A two-tailed test with α = 0.02.

3 A right-tailed test with α = 0.005.

Dr. Yasir Ali ([email protected]) Probability and Statistics

Approach to Hypothesis Testing with Fixed Probability of Type I Error
1. State the null and alternative hypotheses.
2. Choose a fixed significance level α.
3. Choose an appropriate test statistic and establish the critical region
based on α.
4. Reject H0 if the computed test statistic is in the critical region.
Otherwise, do not reject.
5. Draw scientific or engineering conclusions.

Significance Testing (P-Value Approach)

1. State the null and alternative hypotheses.
2. Choose an appropriate test statistic.
3. Compute the P-value based on the computed value of the test statistic.
4. Use judgment based on the P-value and knowledge of the scientific
system.

Dr. Yasir Ali ([email protected]) Probability and Statistics

A researcher reports that the average salary of assistant professors is more than
$42,000. A sample of 30 assistant professors has a mean salary of $43,260. At a
α = 0.05, test the claim that assistant professors earn more than $42,000 per year.
The standard deviation of the population is $5230.

H0 : µ = $42000 and H1 : µ > $42000 (claim). At a α = 0.05, z = 1.65.

z Test for mean
Observed value - Expected value X −µ
Test value= =⇒ z = √ = 1.32
standard error σ/ n

Since 1.32 < 1.65, and is not in the critical region, the decision is to not
reject the null hypothesis.
Summarize the results. There
is not enough evidence to sup-
port the claim that assistant
professors earn more on aver-
age than $42,000 per year.
Dr. Yasir Ali ([email protected]) Probability and Statistics
The mean lifetime of 100 bulbs produced by is 1570 hours with a σ =120 hours.
Test the hypothesis µ = 1600 against the alternative hypothesis µ 6= 1600 (
µ <1600), using a level of significance of 0.05

H0 : µ = 1600 and H1 : µ 6= 1600

(Two tailed Test). At a α = 0.05,
z = ±1.96.
(1) Reject H0 if the z score of the
sample mean is outside the range
to 1.96.
(2) Accept H0 otherwise.

µ = 1600, σ = 120 and n = 100

−1600
gives z = X120/10 = −2.50 lies
outside the range -1.96 to 1.96, we
reject H0 at a 0.05 level of signifi-
cance.
Dr. Yasir Ali ([email protected]) Probability and Statistics
The mean lifetime of 100 bulbs produced by is 1570 hours with a σ =120 hours.
Test the hypothesis µ = 1600 against the alternative hypothesis µ 6= 1600 (
µ <1600), using a level of significance of 0.05

H0 : µ = 1600 and H1 : µ 6= 1600

(Two tailed Test). At a α = 0.05, H0 : µ = 1600 and H1 : µ < 1600
z = ±1.96. (One (left )tail Test). At a α =
(1) Reject H0 if the z score of the 0.05, z = −1.645.
sample mean is outside the range (1) Reject H0 if the z is less than
to 1.96. -1.645.
(2) Accept H0 otherwise. (2) Accept H0 otherwise.

µ = 1600, σ = 120 and n = 100 µ = 1600, σ = 120 and n = 100

−1600
−1600
gives z = X120/10 = −2.50 lies gives z = X120/10 = −2.50 less
outside the range -1.96 to 1.96, we 1.645, we reject H0 at a 0.05 level
reject H0 at a 0.05 level of signifi- of significance.
cance.
Dr. Yasir Ali ([email protected]) Probability and Statistics
The mean lifetime of 100 bulbs produced by is 1570 hours with a σ =120 hours.
Test the hypothesis µ = 1600 against the alternative hypothesis µ 6= 1600 (
µ <1600), using a level of significance of 0.001.

H0 : µ = 1600 and H1 : µ 6= 1600

(Two tailed Test). At a α =
0.001, z = ±2.58.
(1) Reject H0 if the z score of the
sample mean is outside the range
to 2.58.
(2) Accept H0 otherwise.

µ = 1600, σ = 120 and n = 100

−1600
gives z = X120/10 = −2.50 lies in-
side the range -2.58 to 2.58, we
accept H0 at 0.001 level of signif-
icance.
Dr. Yasir Ali ([email protected]) Probability and Statistics
The mean lifetime of 100 bulbs produced by is 1570 hours with a σ =120 hours.
Test the hypothesis µ = 1600 against the alternative hypothesis µ 6= 1600 (
µ <1600), using a level of significance of 0.001.

H0 : µ = 1600 and H1 : µ 6= 1600

(Two tailed Test). At a α = H0 : µ = 1600 and H1 : µ < 1600
0.001, z = ±2.58. (One (left )tail Test). At a α =
(1) Reject H0 if the z score of the 0.001, z = −2.33.
sample mean is outside the range (1) Reject H0 if the z is less than
to 2.58. -2.33.
(2) Accept H0 otherwise. (2) Accept H0 otherwise.

µ = 1600, σ = 120 and n = 100 µ = 1600, σ = 120 and n = 100

−1600
−1600
gives z = X120/10 = −2.50 lies in- gives z = X120/10 = −2.50 less -
side the range -2.58 to 2.58, we 2.33, we reject H0 at a 0.001 level
accept H0 at 0.001 level of signif- of significance.
icance.
Dr. Yasir Ali ([email protected]) Probability and Statistics
Decision Rule When Using a P-Value

If P-value ≤ α, reject the null hypothesis.

If P-value > α, do not reject the null hypothesis.
There are two schools of thought on P-values. Some researchers do not
choose an a value but report the P-value and allow the reader to decide
whether the null hypothesis should be rejected.1
If P-value ≤ 0.01, reject the null hypothesis. The difference is highly
significant.
If P-value > 0.01 but P-value ≤ 0.05, reject the null hypothesis. The
difference is significant.
If P-value > 0.05 but P-value ≤ 0.10, consider the consequences of
type I error before rejecting the null hypothesis.
If P-value ≥ 0.10, do not reject the null hypothesis. The difference is
not significant.
1
In this case, the following guidelines can be used, but be advised that these guide
lines are not written in stone, and some statisticians may have other opinions.
Dr. Yasir Ali ([email protected]) Probability and Statistics
The mean lifetime of 100 bulbs produced by is 1570 hours with a σ =120 hours.
Test the hypothesis µ = 1600 against the alternative hypothesis µ 6= 1600, using a
level of significance of 0.001 and find the P value of the test.

H0 : µ = 1600 and H1 : µ 6= 1600, α = 0.001

X ± 1600
µ = 1600, σ = 120 and n = 100 gives z = = ±2.50
120/10

P−value
z }| {
P (z ≤ −2.5) + P (z ≥ 2.5) = 0.0124

P − value > α
Since the P-value is more than 0.001, the
decision is not to reject the null hypothesis.

Dr. Yasir Ali ([email protected]) Probability and Statistics

Comparison Between α and P − value

H0 : µ = 1600 and H1 : µ 6= 1600

H0 : µ = 1600 and H1 : µ 6= 1600
α = 0.001 =⇒ z α2 = ±2.58 .
α = 0.001 .
µ = 1600, σ = 120 and n = 100 µ = 1600, σ = 120 and n = 100

X − 1600 X − 1600
gives zo = = −2.50 gives zo = = −2.50
120/10 120/10
P(−2.5 ≤ z ≤ 2.5) = 0.0124
Reject H0 if zo ∈
/ [−z α2 , z α2 ]
Accept H0 if zo ∈ [−z α2 , z α2 ] If P-value ≤ α, reject H0
If P-value > α, accept H0 .
Since −2.5 ∈ [−2.58, 2.58], we Since the P-value is more than
accept H0 at 0.001 level of 0.001, the decision is accept H0 .
significance.

Dr. Yasir Ali ([email protected]) Probability and Statistics

It is claimed that automobiles are driven on average more than 20,000 kilometers
per year. To test this claim, 100 randomly selected automobile owners are asked to
keep a record of the kilometers they travel. Would you agree with this claim if the
random sample showed an average of 23,500 kilometers and a standard deviation
of 3900 kilometers? Use a P-value in your conclusion.

H0 : µ = 20, 000 & H1 : µ > 20, 000

X = 23, 500; σ = 3900 and n = 100 Rule

If P-value ≤ 0.01, reject the
23, 500 − 20, 000
zo = 3900
= 8.97 null hypothesis. The
√
100 difference is highly significant.
one tailed
z }| {
P( z > z0 ) = P(z > 8.97) ' 0

Yes I will agree with the claim that automobiles are driven on average more
than 20,000 kilometers per year.
Dr. Yasir Ali ([email protected]) Probability and Statistics
To test the hypothesis that a coin is fair, using decision rules are adopted:
(1) Accept the hypothesis if the number of heads in a single sample of
100 tosses is between 40 and 60 inclusive,
(2) Reject the hypothesis otherwise.
(a) Find probability of rejecting the hypothesis when it is correct.
(b) Interpret graphically the decision rule and the result of part (a).
(c) Draw conclusions if 100 tosses yielded 53 heads? 60 heads?

The mean and standard deviation of the number of heads in 100 tosses are
1

√
given by µ = np = 100 2 = 50 and σ = npq = 5. As np > 5, nq > 5,
the normal approximation to the binomial distribution can be used.
On a continuous scale, between 40 and 60 heads inclusive is the same as
between 39.5 and 60.5 heads.
P (39.5 < X < 60.5) = P (−2.10 < Z < 2.10) = 0.9642

Dr. Yasir Ali ([email protected]) Probability and Statistics

(a) the probability of not getting between 40 and 60 heads inclusive if the
coin is fair equals 1 − 0.9642 = 0.0358. Then the probability of
rejecting the hypothesis when it is correct equals 0.0358.
(b)

If a single sample of 100 tosses yields a z score between -2.10 and 2.10, we
accept the hypothesis; otherwise we reject the hypothesis and decide that
the coin is not fair.
The error made in rejecting the hypothesis when it should be accepted is
the Type I error of the decision rule; and the probability of making this
error, equal to 0.0358
Dr. Yasir Ali ([email protected]) Probability and Statistics
(c) According to the decision rule, we would have to accept the hypothesis
that the coin is fair in both cases. If only one more head had been
obtained, we would have rejected the hypothesis. This is what one
must face when any sharp line of division is used in making decisions.

Dr. Yasir Ali ([email protected]) Probability and Statistics

Design a decision rule to test the hypothesis that a coin is
fair if a sample of 64 tosses of the coin is taken and if a level
of significance of (a) 0.05, (b) 0.01 is used.

First Method:
If the level of significance is 0.05, it means that non-rejection region area is
95% implies that −1.96 ≤ z ≤ 1.96.
(1) Accept the hypothesis that the coin is fair if −1.96 ≤ z ≤ 1.96.
(2) Reject the hypothesis otherwise.

√
−1.96 ≤ z ≤ 1.96 =⇒ 24.16 ≤ x ≤ 39.84, µ = np = 32, σ = npq = 4

(1) Accept the hypothesis that the coin is fair if the number of heads is
between 24.16 and 39.84, i.e., between 25 and 39 inclusive.
(2) Reject the hypothesis otherwise.

Dr. Yasir Ali ([email protected]) Probability and Statistics

Second Method:
With probability 0.95, the number of heads will lie between µ − 1.96σ and
µ + 1.96σ. As
√ p
µ = np = 64(0.5) = 32, and σ = npq = 64(0.5)(0.5) = 4.

Thus, with probability 0.95, the number of heads will lie between

32 − 1.96(4), and 32 + 1.96(4) =⇒ 24.16 and 39.84

(1) Accept the hypothesis that the coin is fair if the number of heads is
between 24.16 and 39.84, i.e., between 25 and 39 inclusive.
(2) Reject the hypothesis otherwise.

Dr. Yasir Ali ([email protected]) Probability and Statistics

A fabric manufacturer believes that the proportion of orders
for raw material arriving late is p =0 .6. If a random sample
of 10 orders shows that 3 or fewer arrived late, the
hypothesis that p = 0.6 should be rejected in favor of the
alternative p < 0.6. Use the binomial distribution. (a) Find
the probability of committing a type I error if the true
proportion is p = 0 .6. (b) Find the probability of
committing a type II error for the alternatives p = 0 .3, p =
0 .4, and p = 0 .5.
H0 : p = 0.6 and H1 : p < 0.6. To test this hypothesis a random sample of
n = 10 is selected if 3 or fewer arrived late, H0 should be rejected.
3
X
α = P (Rejecting H0 ) = P(0 ≤ X ≤ 3) = Cx10 (0.6)x (0.4)10−x = 0.0548
x=0

Dr. Yasir Ali ([email protected]) Probability and Statistics

(b)
H0 : p = 0.6 and H1 : p < 0.3.
 
  4 to 10 orders
not rejecting H0
β=P =P  for raw material when p = 0.3
When it is false
arriving late
For n = 10 and p = 0.3, we have
10
X 3
X
10−x
β= Cx10 (0.3)x (0.7) =1− Cx10 (0.3)x (0.7)10−x = 0.3504
x=4 x=0

H0 : p = 0.6 and H1 : p < 0.4.

For n = 10 and p = 0.4, we have
10
X 3
X
10−x
β= Cx10 (0.4)x (0.6) =1− Cx10 (0.4)x (0.6)10−x = 0.6177
x=4 x=0

H0 : p = 0.6 and H1 : p < 0.5.

For n = 10 and p = 0.5, we have β = 0.8281.
Dr. Yasir Ali ([email protected]) Probability and Statistics
The average height of females in the freshman class of a certain college has
historically been 162.5 centimeters with a standard deviation of 6.9 centimeters. Is
there reason to believe that there has been a change in the average height if a
random sample of 50 females in the present freshman class has an average height
of 165.2 centimeters? Use a P-value in your conclusion. Assume the standard
deviation remains the same.

Dr. Yasir Ali ([email protected]) Probability and Statistics