Martin Luther University Halle-Wittenberg Statistics II

Chair of Econometrics

Exercise Set 12
Winter Semester 2023/24

1. A study is carried out to investigate whether unemployment duration differs across four
regions. Therefore, 40 unemployed individuals from four regions are randomly chosen. The
table below shows summary results:

region 1 region 2 region 3 region 4

mean 4.3 5.2 4.9 10.7
std. dev. 1.95 2.74 2.96 2.63
size 10 10 10 10

Assume that the populations are normally distributed with equal variances.

a) Test whether there are significant differences in the unemployment durations among
the regions. State the hypotheses, degrees of freedom, rejection rule, empirical value
of the test statistic, and conclusion (α = 0.05).

• hypotheses: H0 : µ1 = µ2 = µ3 = µ4 vs. H1 : Not all four µi (i = 1, 2, 3, 4)

are equal.
• degrees of freedom: numerator df r − 1 = 4 − 1 = 3 and denominator df n − r =
10 + 10 + 10 + 10 − 4 = 36
• rejection rule: reject H0 if F > Fr−1,n−r,α = F3,36,0.05 = 2.866
• empirical value of test statistic:
– grand mean:

1 r
x̄¯ = ∑ x̄i
4 i=1

4.3 + 5.2 + 4.9 + 10.7

=
4

= 6.275

– SSTR:
r
SSTR = ∑ ni (x̄i − x̄¯)2
i=1

= 10 · (4.3 − 6.275)2 + ·(5.2 − 6.275)2 + (4.9 − 6.275)2 + (10.7 − 6.275)2

= 265.275

1
 * SSE: with
1
s2i = (xi j − x̄i )2 ↔ ∑ (xi j − x̄i )2 = s2i · (ni − 1)
ni − 1 ∑

SSE = 9 · 1.952 + 2.742 + 2.962 + 2.632

= 242.897

* MSTR:
SSTR 265.275
MSTR = = = 88.425
r−1 3

* MSE:
SSE 242.897
MSE = = = 6.741
n−r 36
– test statistic:
MSTR 88.425
Fr−1,n−r = = = 13.116
MSE 6.741
• decision: Because the test statistic is greater than the critical value,
we reject the null hypothesis that the average unemployment duration is
equal among the four groups.

2. A study is carried out to investigate the effectiveness of four teaching methods. Therefore,
32 randomly chosen students are distributed among four groups, representing the teaching
method. The table below shows the results in the final exam for the students:

group 1 group 2 group 3 group 4

16 16 2 5
18 12 10 8
20 10 9 8
15 14 10 11
20 18 11 1
15 15 9 9
23 12 10 5
19 13 9 9

Test whether there are significant differences in the exam results among the groups. State
the corresponding hypotheses, degrees of freedom, rejection rule, empirical value of the test
statistic, and your conclusion (α = 0.05).

• hypotheses: H0 : µ1 = µ2 = µ3 = µ4 vs. H1 : Not all four µi (i = 1, 2, 3, 4) are

equal.
• degrees of freedom: numerator df r − 1 = 4 − 1 = 3 and denominator df n − r = (8 +
8 + 8 + 8) − 4 = 28
• rejection rule: reject H0 if F > Fr−1,n−r,α = F3,28,0.05 = 2.9467

2
• empirical value of test statistic:

– mean of sample i:
16 + 18 + ... + 19
x̄1 = = 18.25
8

16 + 12 + ... + 13
x̄2 = = 13.75
8

2 + 10 + ... + 9
x̄3 = = 8.75
8

5 + 8 + ... + 9
x̄4 = = 7.00
8
– grand mean:
n
∑ri=1 ∑ j=1
i
xi j
x̄¯ =
n

∑ri=1 ni x̄i
=
n

16 + 18 + ... + 19 + 16 + 12 + ...13 + 2 + 10 + ... + 9 + 5 + 8 + ... + 9

=
32

8 · 18.25 + 8 · 13.75 + 8 · 8.75 + 8 · 7

= = 11.9375
32
– SST:
r ni
∑ ∑ (xi j − x̄¯)2
i=1 j=1

= (16 − 11.9375)2 + ... + (19 − 11.9375)2 + (16 − 11.9375)2 + ... + (13 − 11.9375)2

+ (2 − 11.9375)2 + ... + (9 − 11.9375)2 + (5 − 11.9375)2 + ... + (9 − 11.9375)2

= 847.8750

– SSTR:
r
∑ ni (x̄i − x̄¯)2
i=1

= 8 · (18.25 − 11.9375)2 + 8 · (13.75 − 11.9375)2

+ 8 · (8.75 − 11.9375)2 + 8 · (7 − 11.9375)2

= 318.7813 + 26.2513 + 81.2813 + 195.0313

= 621.3750

3
 – SSE:
r ni
∑ ∑ (xi j − x̄i )2
i=1 j=1

= (16 − 18.25)2 + ... + (19 − 18.25)2

+ (16 − 13.75)2 + ... + (13 − 13.75)2

+ (2 − 8.75)2 + ... + (9 − 8.75)2

+ (5 − 7)2 + ... + (9 − 7)2

= 5.0625 + ... + 0.5625 + 5.0625 + ... + 0.5625 + 45.5625 + ... + 0.0625 + 4 + ... + 4

= 226.5

– MSTR:
SSTR 621.3750
MSTR = = = 207.1250
r−1 3
– MSE:
SSE 226.5
MSE = = = 8.0893
n−r 28
– test statistic:
MSTR 207.1250
Fr−1,n−r = = = 25.6049
MSE 8.0893
• decision: Because the test statistic is greater than the critical value, we
reject the null hypothesis that the average effectiveness is equal for among
the four groups.