Vous êtes sur la page 1sur 7

REVISION EQT271 SEM 1 2015/2016

Question 4
a)

[0.0547,0.1453]
Hence, we are 95% confidence that the difference in percentage supporters between the year
2000 and year 2014 is from 5.47% to 14.53%

b) =
z 4.1494 > =
z 1.96; Reject H 0
Therefore, at 95% confidence level, there is enough evidence to conclude that there is a
difference in percentage of supporters between the year 2000 and year 2014
c) Both statistical inference procedure show the same result in saying that the percentage of
supporters between year 2000 and 2014 is difference. However, by referring to a), it shows the
detail of the difference. The percentage of supporters in 2000 is slightly higher compare to 2014,
around 5.47% to 14.53%.
d)
i) P value
= 0.0192 <=
0.05; Reject H 0
Therefore, at 95% confidence level, there is enough evidence to conclude that the national
leaders popularity has decrease by more than 5%.
ii) The change will increase the p-value more than the

= 0.05 and H 0

will be failed to be

rejected
Question 5
a)

z = 0.9 < z = 2.5758; Fail to reject H 0


Therefore, at 99% confidence level, there is enough evidence to conclude that in 1990, 5% of the
household owned superbikes

b)

[ 0.0029,0.0084]

At 95% confidence interval the differences in the proportion of superbikes ownership between
year 1990 and 2010 is from -0.29% to 0.84%. Because of the existence of 0% difference
possibility between the two groups, it can concluded that there is actual no difference between
the two groups.

c) =
z 0.0412 < =
z 1.6449; Fail to reject H 0

Therefore, at 95% confidence level, there is not enough evidence to conclude that the
percentage of the superbike ownership increase in the year 2010
d) b) and c) shows a consistent result which is there is no difference between the proportion of the
two groups. However, the detail of the possible differences can be referred to the result of c).
Question 4 (b)
i) Its suitable to be used when the data is not normal
0
ii) H 0 : m A mB =

H1 : mA mB 0

=
=
iii) T min
( 36,116 ) 36
iv) T= 36 ( 54,98 ) ; Reject H 0
There is sufficient evidence to support the manufacturers claim

Question 5
a)
i) Free distribution The distribution can be in any shape
ii)
Sign test
Mann-Whitney test
Kruskal-Wallis test
b)
i) By looking at the mean for each program, Program A has a mean of 135.63 and Program
B has a mean of 180. Therefore, based on Table 5.1, on average, Program B is more
productive.
ii) Mann-Whitney test

39 ( 52,84 ) ; Reject H 0

Question 5
a)
b)

Hence, there is enough evidence to support the hypothesis that new program B is more
productive than program A

Parametric
Data normally distributed

Assumption
Parameter

c)

Nonparametric
Free distribution
M

i)
Group Memory Rank Group Memory Rank Group Memory Rank
1
5
1
2
7
6
3
8
8.5
1
6
3.5
2
9 12.5
3
9 12.5
1
6
3.5
2
9 12.5
3
10
16
1
6
3.5
2
10
16
3
11 18.5
1
6
3.5
2
10
16
3
11 18.5
1
8
8.5
2
12 20.5
3
12 20.5
1
1
1

8
8
9

8.5
8.5
12.5

83.5

94.5

53

ii)

H= 76.95 > = 5.9915; Reject H 0

iii)

There is significant difference in the number of word pairs recalled between the three groups
in at least one pair
Group 1 and 2 and group 1 and 3

Question 4

H= 44.05 > 2= 7.8147; Reject H 0

a)

Hence, there is difference in performance for at least one pair of machine


b) Question cancelled
Question 2
a)
b)

Independent, because

H 0 : =
=
TROYOTA
PRODA
PROTO
H1 : At least one pair is different from the other

c)

i)
Source
Df
SS
Between (Treatment)
2
1300.134
Within (Error)
12
1408.8
Total
14
ii) F = 5.537 > FCV = 3.885;Reject H 0

MS
650.067
117.4

F
5.537

Hence, there is enough evidence to support the researchers claim


d) The shape and position of the three boxplots shows at different length and place indicate that
they have different kind of variation, which was consistent with the answer in c(ii). The largest
variation in TROYOTA indicate that the number of user of TROYOTA between all 5 locations were
varies.
e)
i)

[ 37.05, 8.55]

R = -37.05 S = -8.55
Pair of PRODA and PROTA and pair of PROTO and TROYOTA because there is value 0 within
their intervals.
iii) Pair of PRODA and TROYOTA because there is no value of 0 within its interval
f) ..
g) ..
ii)

Question 2
a)
i)

No, there is no sufficient evidence to reject

H 0 because P value
= 0.352 > =
0.05

ii) Within groups variation, Between groups variation, small value of F will reduce the chance
for the
b)

H 0 to be rejected

iii) F-statistic should be more than the value of F critical, which in this case is 3.885
i)

Yes, there is sufficient evidence to reject

H 0 because P value
= 0.0104 <=
0.05

ii) Increasing the sample size will also increase the power of the test, hence giving a better
result.
iii) Yes, the conclusion will change because P value
= 0.0104 >=
0.01
c) I would suggest the company manager to follow the result obtained in b) because it has a better
result due to bigger sample size
d)
1) Paint A (shortest average drying time. Even though its variance was quite high - inconsistent,
the highest possible drying time of 137 can still be considered good)
2) Paint C (intermediate and consistent average drying time)
3) Paint B (Longest and inconsistent average drying time )
Question 3
a)

b)

B1
A
C
B

B2
B
C
A

B3
C
B
A

i)
Source
SS
Df
MS
F
Treatment (Sleep)
128.69
3
42.9
33.52
Block (Student)
28.19
3
9.4
7.34
Error
11.56
9
1.28
Total
168.44
15
ii) Between groups variation for treatment and Between group variation for block
iii) F = 33.52 > FCV = 6.992; Reject H 0
Hence, it can be concluded that at least one level of sleep is different from the other
iv) Yes, the variation will be different because the within group variation increase and affecting
the chance of rejecting

H0

Question 3
a) Factorial Experiment, because it has two main factors
b)
M
F
1
28
34
2
17
22.5
3
29.5
37

Interaction plot
40
35
30
25
20
15
10
5
0

F
1

No interaction, Programs have an effect, Gender also has an effect


c)

d)

Source
A
B
AB
Error
Total

SS
120.33
418.5
2.17
73
614

Df
1
2
2
6
11

MS
120.33
209.25
1.085
12.17

F
9.89
17.19
0.09

i) F = 0.09 < FCV = 10.925; Fail to reject H 0


Hence, it can be concluded that there is no interaction between program and gender

ii) F = 9.89 < FCV = 13.745; Fail to reject H 0

F = 17.19 > FCV = 10.925; Reject H 0


Hence it can be concluded that gender has no effect on training performance
Hence, it can be concluded that program has an effect on training performance
e) For the best program, I would advice the MNM Corporation to choose Program 3 because on
average, Program 3 produce better result of performance

Question 2 (Jun 2016 Kuratif)


a)

b)

i) Patients
ii) Drug A and Drug B
iii) Anxiety level
B1
70
54
56
45
56
77

A1
A2
c)
Source
A
B
AB
Error
Total
d)

SS
4.08
18.75
10.09
2070
2102.92

Df
1
1
1
8
11

B2
78
54
35
45
56
75
MS
4.08
18.75
10.09
258.75

F
0.02
0.07
0.04

F = 0.04 < FCV = 5.318; Fail to reject H 0


Hence, there is no significant interaction effect between Drug A and Drug B

Question 1
a) The line describes by the regression attempts to minimize the distance between points and
fitted line
2
2
b) rMachine A 0.8788
=
=
rMachine
0.9253
B

c)
d)

P value
= 6.31E 05 < =
0.1; Reject H 0
Hence, there is a significance linear relationship between hours of machine used and the size
of off-target

e) For the regression model of Machine A, r = 0.8788 indicate that 87.88% variation in offtarget production was explained by the hours of Machine A used. Therefore, it can
concluded that the model developed for Machine A is adequate. The same goes for Machine
2

B as the value of r = 0.9253 , which is even higher compared to Machine A


2

f) x A 43.91
=
=
hours xB 48.5hours

g) Machine B because based on e), it has more explained variation and based on f) its more
durable.

Question 4
a) From correlation we can only get a number describing the linear relationship between
two variables; in regression we can predict the relationship between variables.
b) Figure 4.1 Strong and positive relationship
Figure 4.2 Weak and positive relationship
c) t = 5.38 > tCV = 2.1604; Reject H 0
There is a significant linear relationship between Engineering and Statistics.
d) For 1 unit score change in statistics, the average change in the score of Engineering
subject is 0.731
69% variation of score in Engineering was explained by the score of Statistics.
e) It is expected that the value of F-statistics for calculus and statistics to be small because
it seems like they dont really has a linear relationship and in test,
to be rejected
f)
g) ..

H 0 should be failed