WILP Advanced Statistical Methods

BITS Pilani
Semester 1 (2024-25)
Mid-Semester Exam (Regular) Solutions

1. As part of the testing program conducted by a survey-group, the life-length of a mini-

laptop’s battery has been tested. A random sample from a market place yielded the
following results (in months):

10 12 11 16 23 16 19 10 18.

(a) Assuming that the life time distribution is approximately normal, does the data
suggest that the mean life time is different from 16 months ? Take α = 0.05.
(b) Determine a 90% confidence interval for the mean lifetime.

Solution: Let µ denote the mean lifetime (in months) of the laptop’s battery.

(a) We want to test the following hypotheses :

H0 : µ = 16 vs Ha : µ ̸= 16. (1)
The sample mean and the sample standard deviation are given by
v
9 u 9 r
1 X 135 u1 X 166
x= xi = = 15, s = t (xi − 15)2 = = 4.5552. (1)
9 9 8 8
i=1 i=1

So the value of the test statistic under H0 is

x − µ0 15 − 16
t= √ = √ = −0.6586.
s/ n 4.5552/ 9
Under H0 , t ∼ tn−1 = t9−1 = t8 , that is, a t distribution with 8 d.f. (1)
Since this is a two-sided test, we reject H0 if t ≤ −tα/2,n−1 = −t0.025,8 or if t ≥
tα/2,n−1 = t0.025,8 . Now t0.025,8 = 2.306. Since −0.6586 > −2.306, we can’t reject H0
at 5% level of significance. Hence the data does not suggest that the mean lifetime
is different from 16 months.
(b) Given the confidence level is 90% so 100(1 − α) = 90 or α = 0.10. The upper t-critical
value at α = 0.10 is tα/2,n−1 = t0.05,9−1 = t0.05,8 = 1.86. (1)
Hence the 90% confidence interval for the mean lifetime or µ is
s 4.5552
x ± tα/2.n−1 √ = 15 ± 1.86 × √ = 15 ± 2.8242 = (12.1758, 17.8242). (1)
n 9
2. The following 2 × 3 contingency table contains observed frequencies for various levels of
the variables X and Y . Test whether the variables are independent assuming α = 0.05.

Y
X 1 2 3
1 20 44 50
2 30 26 30

1
 Solution: We want to test the following hypotheses.

H0 : The variables X and Y are independent.

Ha : The variables X and Y are not independent. (1)

Under H0 the expected frequencies in the contingency table are given by

Row i Total × Column j Total
eij = .
Total Sample Size
Let fij denote
P the observed frequencies in the contingency table. Then the total sample
size is n = i,j fij = 200. (1)
The table of expected frequencies computed using the above formula is as follows.

Y
X 1 2 3
(2)
1 28.5 39.9 45.6
2 21.5 30.1 34.4

The value of the test statistic under H0 is given by

2 X
3
X (fij − eij )2
χ2 = = 7.8627.
eij
i=1 j=1

Under H0 , the test statistic has a chi-square distribution with (2 − 1)(3 − 1) = 2 degrees
of freedom, that is, χ2 ∼ χ22 under H0 . (1)
Since 7.8627 > χ20.05 (2) = 5.991, we reject H0 at 5% level of significance and conclude that
the variables X and Y are not independent. (1)

3. Consider the following sample drawn from a normal population

1 6 2 4 3 6 1 4

(a) Test whether the population variance is different from 5. Take α = 0.10.
(b) Construct a 90% confidence interval for the population variance.

Solution: Given the sample size is n = 8. Let x1 , . . . , x8 denote the sample observations.
Then the sample mean and the sample variance are
n n n
1X 27 1 X 1X 27.875
x= xi = = 3.375, s2 = (xi −x)2 = (xi −3.375)2 = = 3.9821.(1)
n 8 n−1 7 7
i=1 i=1 i=1

(a) Let σ 2 denote the population variance. We want to test the hypotheses
H0 : σ 2 = 5 vs Ha : σ 2 ̸= 5. (1)

2
 The value of the test statistic under H0 is
(n − 1)s2 7 × 3.9821
χ2 = = = 5.575.
σ2 5
Note that χ2 ∼ χ2n−1 = χ27 under H0 . (1)
Given α = 0.10 so 1 − α/2 = 0.95, χ27 (0.95) = 2.1673 and χ27 (0.05) = 14.0671. Since
χ27 (0.95) < χ2 < χ27 (0.05), we can’t reject H0 at 10% level of significance and conclude
that the population variance is not different from 5. (1)
(b) Given the confidence level is 90% so 100(1 − α) = 0.90 or α = 0.10. (1)
Hence a 90% confidence interval for the population variance is
2 (n − 1)s2 7s2 7s2
     
(n − 1)s 27.875 27.875
, = , = , = (1.9816, 12.8616).(1)
χ2n−1 (α/2) χ2n−1 (1 − α/2) χ27 (0.05) χ27 (0.95) 14.0671 2.1673
4. An experiment has been conducted for five treatments with three blocks. The following
data was recorded - sum of squares due to total (SST) = 430, sum of squares due to
treatments (SSTR) = 310, sum of squares due to blocks (SSBL) = 85. Test whether there
is any significant difference among the treatment means using α = 0.05. Also show the
analysis of variance (ANOVA) table.
Solution: Here the sum of squares due to error is given by
SSE = SST − SSTR − SSBL = 430 − 310 − 85 = 35.
ALso k = 5, b = 3 and n = bk = 15 where k is the number of treatments, b is the number
of blocks and n is the total sample size. The mean sums of squares are given by
SST 430 SSTR 310
MST = = = 30.7143, MSTR = = = 77.5,
n−1 14 k−1 4
SSBL 85 SSE 35
MSBL = = = 42.5, MSE = = = 4.375. (2)
b−1 2 (k − 1)(b − 1) 8
Let µ1 , µ2 , µ3 , µ4 and µ5 denote the population means of Treatments 1, 2, 3, 4 and 5
respectively. We want to test the following hypotheses.
H0 : µ1 = µ2 = µ3 = µ4 = µ5 vs Ha : µi ̸= µj for at least one pair (i, j). (1)
The F-statistic for testing the above hypotheses is
MSTR 77.5
F = = = 17.7143.
MSE 4.375
Here F ∼ Fk−1,(k−1)(b−1) = F4,8 under H0 . (1)
Since 17.7143 > F4,8 (0.05) = 3.8379, we reject H0 at 5% level of significance and conclude
that all the treatment means are not equal. (1)
The ANOVA table for this experiment is shown below.

Source of variation Sum of squares Degrees of freedom Mean square F-statistic

Treatments SSTR = 310 4 MSTR = 77.5 17.7143
Blocks SSBL = 85 2 MSBL = 42.5
Error SSE = 35 8 MSE = 4.375
Total SST = 430 14 MST = 30.7143
(1)

3
 5. A large automobile insurance company selected samples of single and married male policy
holders and recorded the number who made an insurance claim over the preceding three-
year period. Out of 400 single policyholders, 76 made claims whereas 90 out of 900 married
policyholders made claims.

(a) Determine whether the claim rates differ between single and married male policy-
holders using α = 0.05.
(b) Provide a 95% confidence interval for the difference between the proportions for the
two populations.

Solution: Denote the single male policyholders by Population 1 and the married male
policyholders by Population 2. Let p1 and p2 denote the population proportions of single
and married male policyholders respectively who made insurance claims. Let n1 and n2
denote the sample sizes from populations 1 and 2 respectively. Then the sample prportions
are
76 90
p1 = = 0.19, p2 = = 0.1. (1)
400 900
(a) We want to determine whether the claim rates differ between single and married male
policyholders. So the hypotheses are

H0 : p1 − p2 = 0, p1 − p2 ̸= 0. (1)

Under H0 , the pooled sample proportion is

n1 p1 + n2 p2 400 × 0.19 + 900 × 0.1 166
p= = = = 0.1277.
n1 + n2 400 + 900 1300
The test statistic is
p1 − p2 0.19 − 0.1
z=q =q = 4.4874. (1)
p(1 − p)( n11 + 1
n2 )
1
0.1277(1 − 0.1277)( 400 + 1
900 )

Under H0 , Z ∼ N (0, 1). This is a two-tailed test, so the critical values are z0.025 = 1.96
and −z0.025 = −1.96 for α = 0.05. Since z > 1.96, we reject H0 at 5% level of
significance and conclude that the insurance claim rates differ between single and
married male policyholders.
(b) Given the confidence level is 95% so 100(1 − α) = 0.95 or α = 0.05. (1)
Hence a 95% confidence interval for the difference between the proportions for the
two populations is
s r
p1 (1 − p1 ) p2 (1 − p2 ) 0.19 × 0.81 0.1 × 0.9
p1 −p2 ±z0.025 + = 0.19−0.1±1.96 + = (0.0468, 0.1332).(1)
n1 n2 400 900