Estimation and Testing
Assignment 1
29-04-2015
Jeffrey Holtrop, s2242540
Aho Mirza, s2238462
29-04-2015 Estimation and Testing
Question 1
Question a
n = 72 since σ is unknown, we use s2 = 1.52749 x̄ = 6.0486
h0 : µ = 6.4
ha : µ 6= 6.4
α = 0.05
since n ≥ 30wecanusez
We reject the null hypothesis if z ≥ za /2, n − 1
where za /2, n − 1 = 1.96 or z ≤ -ta /2, n − 1 = -1.96
x̄−µ 6.0486−6.4
z= √σ = 1.52749
√
= -1.952
n 72
Since z≤ za /2, n − 1, we cannot reject the h0 . Also by the table in spss, we find p value = 0.055, larger
than 0.05 hence cannot reject. Note: 0.055 is 2x the pvalue of a 1 sided test.
Question b
since σ is unknown, we use s2 = 1.52749 x̄ = 6.0486
h0 : µ = 6.4
ha : µ < 6.4
since n ≥ 30wecanusez
We reject the null hypothesis if t ≥ za , n − 1
where za , n − 1 = 1.645 or z ≤ -ta , n − 1 = -1.645
x̄−µ 6.0486−6.4
z= √σ = 1.52749
√
= -1.952
n 72
Since -1.952 is smaller than -1.645 we can reject the h0 . Hence there average exam grade is less than 6.4.
Note of a, 0.055 /2 = 0.0275 hence it is below 5%, similarly reject.
Question c
n = 72 since σ is unknown, we use s2 = 1.52749 x̄ = 6.0486
h0 : µ = 6.4
ha : µ 6= 6.4
α = 0.05
We use Wilcoxon test. Since n≥ 15 we can use a normal approximation. Meaning we will reject the null
if z ≤ z0 .05 = 1.96 or z ≤ -z0 .05 = 1.96
T −µ
with z = σ
The Wilcoxon p value is 0.108, hence we cannot reject h0 .
Question d
x¯1 = 6.6314
x¯2 = 5.4973
s21 = 1.30040
s22 = 1.53686
n1 = 35
n2 = 37
1
29-04-2015 Estimation and Testing
h0 : µ1 = µ2
ha : µ1 6= µ2
α = 0.01
Reject the h0 if 2.576= za /2 ≤ z or -2.576 =−za /2 ≥ z
µ1 −µ2 −δ
z= q
s2 s2
1 + 2
n n
6.6314−5.4973−0
= p 1.30042 1.536862
35 + 37
= 3.387
Since we obtain 3.387 ≥ 2.576, we reject the null. There is a difference in the exam grades of students
between the tutorial groups on 1%.
Question e
x¯1 = 6.6314
x¯2 = 5.4973
s21 = 1.30040
s22 = 1.53686
n1 = 35
n2 = 37
h0 : µ1 = µ2
ha : µ1 > µ2
α = 0.01
Reject the h0 if 2.576= za /2 ≤ z or -2.576 =−za /2 ≥ z
µ1 −µ2 −δ
z= q 2 2
s s
1 + 2
n n
6.6314−5.4973−0
= p 1.30042 1.536862
35 + 37
= 3.387
Since we obtain 3.387 ≥ 2.326, we reject the null. The exam grades of the students in group 1 is larger
than the grades in group 2 on 1% significance. Note: pvalue is 0.001 on 2 sides, multiply by 2 gives 0.002.
This is less than 0.05. Hence reject null.
Question f
h0 : µ1 = µ2
ha : µ1 6= µ2
α = 0.01
We reject the null if U ≤ Ua without a normal approximation.
Reject the h0 if 2.326 = za /2 ≤ z or -2.326 =−za /2 ≥ z
U = 370 from the output.
We can calculate the mean which is n1∗n22 = 647.5
q
the standard deviation is n1∗n2∗(n1+n2+1)
12 = 88.757
Since n1 and n2 ≥ 20 we can use a normal approximation Hence we use
z = U −µ
σm
m
= 370−647.5
88.737 = −3.127
Since -3.127 ≤ -2.236, we can reject the null.
The exam grades differ between the groups.
2
29-04-2015 Estimation and Testing
Question g
h0 : µ = 0.5
ha : µ 6= 0.5
α = 0.05
The pvalue is 0.018.
Hence we can reject the null. There percentage of passing is unequal to 50%.
Question h
h0 : µ1 = µ2
ha : µ1 6= µ2
α = 0.05
We reject if Xa2 , n − 1 ≤ X 2 where n = 2 groups.
We obtain a Xa2 , n − 1 = 3.841.
We obtain a X 2 = 0.647.
Meaning we cannot reject the null. The passing rates do not differ between the groups.
Question 2
Question a
h0 : holidaytypesareindependent
ha : holidaytypesarenotindependent
a = 0.05
We reject Xa2 , n − 1 ≤ X 2
If X02 .05, 2 ≤ X 2
Pr Pc 2
We use the i=1 j=1 (f ij−eij)
eij = X2
2
We obtain X = 11.661.
Hence we can reject the null since we have 5.991 ≤ 11.661.
Meaning there is a 5% significance in the relationship between holiday destinations and the types of holiday.
Note: table gives a 0.003 pvalue. Hence reject.
Question b
h0 : ai = 0f ori = 1, 2, 3
ha : ai 6= 0 for at least one i = 1,2,3
a = 0.05
we use f = MSTR / MSE as test statistics and
We reject if f ≥ fa , sstr − 1, sse − 1 = f ≥ f0 .05, 2, 151
We use the F from table, this gives f = 53.710.
P value is 0.000.
Hence we can reject the null at 5%. There is a differenvce in cost of holiday between different destinations.
Question c
h0 : ui = u2 = u3 f ori = 1, 2, 3
ha : notallui areequalf ori = 1, 2, 3
a = 0.01
3
29-04-2015 Estimation and Testing
H ≥ X02 .01, 2 = 9.21 We reject if X 2 ≥ 9.21
H from the table gives 63.966. This exceeds ≥ X02 .01, 2 = 9.21. Hence we can reject the null. There is a
difference in costs of holiday between different destinations.
Question d
h0 : ai = 0 for the holiday destinations
h0 : bi = 0 for the types of holidays
h0 : abi = 0 for the interaction effect
ha : ai 6= 0
ha : bi 6= 0
ha : abi 6= 0
a = 0.01 MSTholiday / MSE as test statistics for holiday destinations MSTtype / MSE as test statistics for
types of holidays MSTinteraction / MSE as test statistics for interaction we use f0 .01, 2, 148 = 4.61
we use f0 .01, 1, 148 = 6.63
we use f0 .01, 2, 148 =4.61
for the respective one MSTholiday / MSE as test statistics for holiday destinations = 40.629 MSTtype /
MSE as test statistics for types of holidays = 5.983 MSTinteraction / MSE as test statistics for interaction
= 1.88 hence we can reject null for destination and holiday. There are differences in holiday and destination.
