Discussion - 9 & 10

1 Small sample tests

For small sample tests (n < 30), we will assume that the population has a normal distribution
with mean µ and standard deviation σ.

Theorem 1.1. Let x1 , x2 , . . . , xn be a random sample of size n drawn from a normal proportion
x1 + x2 + . . . + xn
with mean µ and standard deviation σ, which is unknown. Let x = be the
n
n
1 X
sample mean and s2 = (xi − x)2 be the unbiased sample variance. If the sample size n
n − 1 x=1
x−µ
is small, then z = ! follows students t-distribution with n − 1 degrees of freedom.
s
√
n

1.1 Testing of hypothesis concerning the mean of a normal population

Case (i) When σ is known.
Let the null hypothesis be H0 : µ = µ0 .
The test statistic
x − µ0
z=
σ
!
√
n
follows N(0, 1).
If H1 : µ > µ0 is the alternative hypothesis, then the critical region is z > zα .
If H1 : µ < µ0 is the alternative hypothesis, then the critical region is z < −zα .
If H1 : µ , µ0 is the alternative hypothesis, then the critical region is |z| > zα/2 .
Case (ii) When σ is unknown.
The test statistic
x − µ0
t= !
s
√
n
n
1 X
where s2 = (xi − x)2 follows t-distribution with n − 1 degrees of freedom.
n − 1 x=1
If H1 : µ > µ0 is the alternative hypothesis, then the critical region is t > tα .
If H1 : µ < µ0 is the alternative hypothesis, then the critical region is t < −tα .
If H1 : µ , µ0 is the alternative hypothesis, then the critical region is |t| > tα/2 .

Example 1.2. The time taken for a machine to produce a certain plastic bowl has a normal
distribution with mean 4.8 seconds. Recently the machine was readjusted to reduce the mean.

1
To test whether the adjustment was successful the time taken to produce 8 bowls are measured
and the results, in second, are follows.

4.2, 4.5, 4.9, 4.3, 4.5, 4.7, 4.6, 4.4

Test at 5% level of significance, whether the adjustment had the desired effect.

µ = 4.8, n = 8
H0 : µ = 4.8
H1 : µ < 4.8
4.2 + 4.5 + 4.9 + 4.3 + 4.5 + 4.7 + 4.6 + 4.4
x= = 4.5125
8
1
s2 = [(4.2 − 4.5125)2 + (4.5 − 4.5125)2 + (4.9 − 4.5125)2 + (4.3 − 4.5125)2
8−1
+ (4.5 − 4.5125)2 + (4.7 − 4.5125)2 + (4.6 − 4.5125)2 + (4.4 − 4.5125)2 ]
= 0.0498
x − µ0
z= !
s
√
n
4.5125 − 4.8
= !
0.0498
√
8
= −3.64

For α = 0.05 and n − 1 = 7 degrees of freedom, from the t-table, zα = 1.895. Since z = −3.64 <
zα = −1.895, we reject H0 . That is, the mean time has been reduced by the adjustment.

1.2 Testing the equality of means of two normal populations with the
same standard deviation
Let x1 and x2 be the means, and s21 and s22 are the unbiased sample variances of two independent
random samples of sizes n1 and n2 taken from the two normal populations with means µ1 and
µ2 respectively. Let σ2 be the common variance and sample sizes n1 and n2 are small (< 30).
Case (i) When σ is known.
Let the null hypothesis be H0 : µ1 = µ2 or H0 : µ1 − µ2 = 0 and α be the level of significance.
The test statistic is
x1 − x2
z= s !
1 1
σ2 +
n1 n2
and it follows N(0, 1).
If H1 : µ1 > µ2 is the alternative hypothesis, then the critical region is z > zα .
If H1 : µ1 < µ2 is the alternative hypothesis, then the critical region is z < −zα .

2
If H1 : µ1 , µ2 is the alternative hypothesis, then the critical region is |z| > zα/2 .
Case (ii) When σ is unknown.
The pooled estimate for the common variance is given by

(n1 − 1)s21 + (n2 − 1)s22

s =
2
n1 + n2 − 2
The test statistic is
x1 − x2
t= s !
1 1
s2 +
n1 n2
and it follows t-distribution with n1 + n2 − 2 degrees of freedom.
If H1 : µ1 > µ2 is the alternative hypothesis, then the critical region is t > tα .
If H1 : µ1 < µ2 is the alternative hypothesis, then the critical region is t < −tα .
If H1 : µ1 , µ2 is the alternative hypothesis, then the critical region is |t| > tα/2 .
Example 1.3. A researcher is testing the effect of plant food on plant growth. Nine plants have
been given the plant food. Another nine plants have not been given the plant food. The heights
of the plants are recorded after eight weeks. The mean height of plants fed with plant food is
found to be 16 inches with a standard deviation of 2.5 inches while that of plants without food
is 14 inches with a standard deviation of 1.5 inches. Test at 1% level of significance whether
food makes the plants grow taller. Assume that both populations have normal distributions with
identical variance.
Let µ1 and µ2 denote the means of the heghts of plants fed with plant food and without food
respectively.

n1 = 9, x1 = 16, s1 = 2.5
n2 = 9, x2 = 14, s2 = 1.5
H0 : µ1 = µ2
H1 : µ1 > µ2
(n1 − 1)s21 + (n2 − 1)s22
s =
2
n1 + n2 − 2
(9 − 1)2.52 + (9 − 1)1.52
=
9+9−2
= 4.25
x1 − x2
z= s !
1 1
s2 +
n1 n2
16 − 14
= s !
1 1
4.25 +
9 9
= 2.058

3
For α = 0.05 and n1 + n2 − 2 = 16, from the t−table, zα = 2.583. Since z = 2.058 < zα = 2.583,
we accept H0 . Thus the samples do not provide sufficient evidence to conclude that the plant
food makes plant grow taller.

2 Exercise
1. The mean lifetime of a sample of 25 bulbs is found as 1550 hours with a SD of 120 hours.
The company manufacturing the bulbs claims that the average life of their bulbs is 1600
hours. Is the claim is acceptable at 5% level of significance.
2. The mean spot speed of 15 vehicles observed on a Sunday at a particular location of a
highway was 81.2 km/hour and the standard deviation was 10.2 km/hour. The mean speed
of all vehicles at this location as per previous record was 77.5 km/hour. Is there sufficient
evidence to show that the speed of vehicles of the particular Sunday was higher than the
average speed ?
3. The manufacturer of a certain type of metal wire claims that the mean breaking strength
of the wire is more than 575 kg. Tests made on the breaking strength of 10 samples of the
wire gave the results

Sample No. 1 2 3 4 5 6 7 8 9 10
Breaking strength(kg.) 578 572 570 568 572 570 570 572 596 590

Test whether the manufacturer’s claim can be accepted at 5% level of significance.

4. A certain injection administered to each of 12 patients resulted in the following increases
of blood pressure 5, 2, 8, −1, 3, 0, 6, −2, 1, 5, 0, 4. Can it be concluded that the injuction,
in general, accompanied by an increase in blood pressure.(Use 5% level of significance).
5. Two types of batteries are tested for their length of life and the following resullts were
obtained

No. of Sample Mean (Hrs.) Variance

Battery A 10 500 100
Battery B 15 560 121

Is there a significance difference in the two means. Assume that both populations have
normal distributions with identical variance.
6. Two persons A and B were tested according to the time (in seconds) to run a particular
race with the following results

Person A 28 30 32 33 33 32 34
Person B 29 30 30 26 39 27

Can you conclude that the two persons differ significantly as far their running ability is
concerned. Assume that both populations have normal distributions with identical vari-
ance.

4
7. To test whether the sex of a driver had any effect on the speed of a vehicle, a spot speed
study was conducted . A sample of 8 male drivers yielded a mean speed of 70.2 km/hour
with a variance of 98.2 km/hour and a sample of 12 female drivers yielded a mean speed
of 65.5 km/hour with a variance of 84.2 km/hour. Test whether the mean speed of female
drivers is significantly below the mean speed of male drivers. Assume that the samples
come from normal population with the same variance.

8. The following table gives the biological values of protein contained 6 randomly collected
samples of cow’s milk and buffalo’s milk. Examine whether the average values of protein
in the two types of miks differ significantly (use 5% level of significance)

Cow’s milk (x1 ) 1.82 2.02 1.88 1.61 1.81 1.54

Buffalo’ s milk (x2 ) 2.00 1.83 1.86 2.03 2.19 1.88

Assume that both populations have normal distributions with identical variance.