Bethlehem University Faculty of Business Project
Bethlehem University Faculty of Business Project
Bethlehem University Faculty of Business Project
Faculty of Business
Project
Econ 234
By
Saif Jaber
&
Ahmad Abdallah
Fall 2013
Content:
No. Content
1
Testing the difference between two means of two dependent normal populations
Testing the differences between two means of two independent normal populations
10
11
Cranachs alpha
1.
Example:
To discover the result of training program on the chairs production
department employees, we took a 12 employees randomly, and their
outcomes of chairs during the month after and before the training are
given in the following table:
Employees No.
1
2
3
4
5
6
7
8
9
10 11 12
Outcomes before training 110 116 125 103 109 102 113 118 105 112 109 111
Y
Outcomes after training X 115 118 129 116 114 105 119 120 115 116 118 113
Pair 1
after
before
Mean
116.5000
111.0833
N
12
12
Std. Deviation
5.51856
6.50117
Std. Error
Mean
1.59307
1.87672
12
Correlation
.840
Sig.
.001
c.
d.
2.
Std. Error
Mean
1.01845
Lower
3.17507
Upper
7.65827
t
5.319
df
11
Sig. (2-tailed)
.000
Example:
Claim: md = 0
Hypotheses: H0: md = 0 , H1: md 0
Group Statistics
company
1.00
2.00
income
N
20
22
Mean
Std. Deviation Std. Error Mean
2275.0000
376.79395
84.25369
2386.3636
397.96561
84.84655
Sig.
df
Std. Error
Difference
Lower
Upper
income
Equal variances
assumed
.147
c.
d.
.704
-.929
40
.359
-111.36364
119.89180
-365.88312
143.15585
-.931 39.926
.357
-111.36364
119.57266
-365.22143
142.49416
Branch
Ramallah
Type of car
X-5
I-535
I-318
I-320
Total
57
63
34
17
171
Amman
120
86
97
81
384
Dubai
345
408
274
312
1339
Cairo
100
104
202
300
706
Total
622
661
607
710
2600
Study the claim that the type of car (Y) and the Branch (X) are
independent under 0.05 as a level of Significance?
Solution:
a. Claim: The two variables car types and Branches are independent.
b. Hypotheses:
Case Processing Summary
Cases
Valid
N
Branch * CarType
2600
Missing
Percent
100.0%
Total
Percent
0
0.0%
N
2600
Percent
100.0%
H0: The two variables car types and Branches are independent.
H1: The two variables car types and Branches are dependent.
Procedure: Data weight cases()weight case by frequency
variable (observed value)continue ok. Analyzedescriptive
statisticscrosstabs rows(branch position)columns(student living
place)cells ()observed ()expected continue statistics()chisquarecontinueok.
CarType
X-5
Ramallah
Amman
Branch
Dubai
Cairo
d.
I-318
I-320
57
63
34
17
171
Expected Count
40.9
43.5
39.9
46.7
171.0
Count
120
86
97
81
384
Expected Count
91.9
97.6
89.6
104.9
384.0
Count
345
408
274
312
1339
320.3
340.4
312.6
365.7
1339.0
100
104
202
300
706
168.9
179.5
164.8
192.8
706.0
622
661
607
710
2600
622.0
661.0
607.0
710.0
2600.0
Expected Count
Count
Expected Count
Count
Total
c.
Count
I-535
Total
Expected Count
df
206.708a
.000
Likelihood Ratio
211.929
.000
Linear-by-Linear Association
134.794
.000
Pearson Chi-Square
N of Valid Cases
2600
Department Week1 Week2 Week3 Week4 Week5 Week6 Week7 Week8 Week9
First
12
16
13
10
11
Second
14
14
11
16
12
10
16
Third
13
16
10
15
13
11
11
11
Fourth
11
12
15
12
15
14
13
10
Fifth
15
10
11
14
14
16
Solution:
a. Claim: u1 = u2 = u3 = u4 = u5.
b. Hypotheses: H0: u1 = u2 = u3 = u4 = u5.
H1: Not all means are equal.
10
df
Mean Square
7.154
1.788
Within Groups
143.017
30
4.767
Total
150.171
34
Sig.
.375
.824
Constant 59 51 56 48 57 55 44 51 53
High
52 43 54 46 52 53 38 50 51
Low
56 57 52 45 54 49 40 53 38
a.
b.
H0: u1 = u2 = u3.
H1: Not all means are equal.
Procedure: Analyzecompare mean One-Way
ANOVAdependent list(purchases number)
factor(purchase method) post-hoc ()
Benferroni ()Tuky ()Scheffesignificant level
()continue Ok.
ANOVA
Score
Sum of Squares
Between Groups
df
Mean Square
79.630
39.815
Within Groups
787.556
24
32.815
Total
867.185
26
c.
Sig.
1.213
.315
(J) Tempreature
Mean Difference
Std. Error
Sig.
(I-J)
Constant
Tukey HSD
High
Low
Scheffe
Constant
High
Low
Upper Boun
High
3.88889
2.70040
.337
-2.8548
10.63
Low
3.33333
2.70040
.445
-3.4104
10.07
-3.88889
2.70040
.337
-10.6326
2.85
-.55556
2.70040
.977
-7.2992
6.18
-3.33333
2.70040
.445
-10.0770
3.41
High
.55556
2.70040
.977
-6.1881
7.29
High
3.88889
2.70040
.370
-3.1558
10.93
Low
3.33333
2.70040
.478
-3.7114
10.37
-3.88889
2.70040
.370
-10.9336
3.15
-.55556
2.70040
.979
-7.6003
6.48
-3.33333
2.70040
.478
-10.3781
3.71
Constant
Low
Constant
Constant
Low
Constant
Constant
Bonferroni
High
.55556
2.70040
.979
-6.4892
7.60
High
3.88889
2.70040
.488
-3.0610
10.83
Low
3.33333
2.70040
.687
-3.6165
10.28
-3.88889
2.70040
.488
-10.8388
3.06
-.55556
2.70040
1.000
-7.5054
6.39
-3.33333
2.70040
.687
-10.2832
3.61
.55556
2.70040
1.000
-6.3943
7.50
Constant
High
Low
Constant
Low
High
Tuckey ost hoc, Sheffe and Benferroni tests indicated that working with
a constant temperature and working with low temperature or working
with constant temperature and working with high temperature or
working with high temperature and working with low temperature are
the same (P-values are 0.445, 0.478, 0.687 / 0.337, 0.370, 0.488 / 0.977,
0.979, 1 respectively).
One Way Repeated Measures ANOVA
Example:
We made seven machines annual maintenance we check number of
product produced every three months:
6.
Machines
14
13
12
13
19
18
17
18
12
11
10
11
Level means 10
Solution:
a. Claim: At least two means are significantly different.
b. Hypotheses: H0: u1 = u2 = u3
Mauchly's W
Approx. Chi-
df
Epsilonb
Sig.
Square
Greenhouse-
Huynh-Feldt
Lower-
Geisser
factor1
.000
.500
.500
Tests the null hypothesis that the error covariance matrix of the orthonormalized transformed dependent variables is proportional to an i
matrix.
a. Design: Intercept
Within Subjects Design: factor1
b. May be used to adjust the degrees of freedom for the averaged tests of significance. Corrected tests are displayed in the Tests of Wit
Subjects Effects table.
theMauchly Sphericity test is significant (P=0.0 ,which is smaller than 0.05 so we take the third
line [Huynh-Feldt]).
Tests of Within-Subjects Effects
Measure: MEASURE_1
Source
df
Mean Square
Sig.
Squares
factor1
Error(factor1)
Sphericity Assumed
10.381
5.190
27.250
.000
Greenhouse-Geisser
10.381
1.000
10.381
27.250
.002
Huynh-Feldt
10.381
1.000
10.381
27.250
.002
Lower-bound
10.381
1.000
10.381
27.250
.002
Sphericity Assumed
2.286
12
.190
Greenhouse-Geisser
2.286
6.000
.381
Huynh-Feldt
2.286
6.000
.381
Lower-bound
2.286
6.000
.381
In this case we take Huynh-Feldt and as you can see there is a high significant effect of the
level Variable since the P-value is 0.005 < 0.05. The statistic value is F = 27.250.
Pairwise Comparisons
Measure: MEASURE_1
(I) factor1
(J) factor1
Mean Difference
Std. Error
Sig.b
(I-J)
Lower Bound
1
Upper Bound
1.000
.000
1.000
1.000
1.714*
.286
.003
.775
2.654
-1.000
.000
-1.000
-1.000
.714
.286
.140
-.225
1.654
-1.714
.286
.003
-2.654
-.775
-.714
.286
.140
-1.654
.225
There is a significant difference between 3 months and 6 months (P =0) and between 3months and 9
months (P = 0.003), but no significant difference between 6 months and 9 months ( P = 0.140).
Asia
Africa
America
Europe
Children
Young
Old
age
Target Market
1.00
Children
28
2.00
Young
28
3.00
Old
28
1.00
Asia
21
2.00
Africa
21
3.00
America
21
4.00
Europe
21
Descriptive Statistics
Dependent Variable: test
age
Children
Young
Old
Total
Target Market
Mean
Std. Deviation
Asia
6.0000
3.46410
Africa
4.5714
2.63674
America
10.0000
1.41421
Europe
9.5714
1.81265
Total
7.5357
3.30524
28
Asia
7.5714
1.98806
Africa
6.5714
2.37045
America
11.5714
2.63674
Europe
11.4286
2.63674
Total
9.2857
3.23015
28
Asia
6.5714
1.90238
Africa
6.7143
3.03942
America
10.8571
1.95180
Europe
10.7143
2.42997
Total
8.7143
3.07748
28
Asia
6.7143
2.51282
21
Africa
5.9524
2.74729
21
10.8095
2.06444
21
America
Europe
10.5714
2.33605
21
8.5119
3.25056
84
Total
df1
df2
11
Sig.
72
.386
we have homogeneity of variance for the dependent variable across groups, since p-value =
0.486 > =0.04. Statistic value: F = 1.085.
Tests of Between-Subjects Effects
Dependent Variable: test
Source
df
Mean Square
Sig.
Squares
Corrected Model
455.274a
11
41.389
7.066
.000
Intercept
6086.012
6086.012
1039.075
.000
44.595
22.298
3.807
.027
405.369
135.123
23.070
.000
5.310
.885
.151
.988
Error
421.714
72
5.857
Total
6963.000
84
876.988
83
Age
targetM
age * targetM
Corrected Total
8.
First (theoretical)
Second (Practical)
No.
Xi1
Yi1
No.
Xi2
Yi2
No.
Xi3
Yi3
10
15
16
25
12
30
10
18
10
33
12
20
14
35
12
16
11
16
15
40
10
14
15
22
11
11
20
17
11
20
11
12
11
20
10
22
10
13
17
10
15
22
10
16
a.
b.
11
11
14
11
12
20
11
12
33
12
10
15
12
10
17
12
15
37
13
13
13
18
13
29
14
11
14
14
21
14
30
Procedure:
AnalyzegenerallinearmodelUnivariatedependentvariable{
depenent variable( y) fixd
factore{code}coveriate{independent variables(x)
contrasts {simple instead of none}{}fiest or last change
continue options display means for{code} {}compare
main effects {}homogeneity {}descriptive statistics
significant level {} continue ok.
Levene's Test of Equality of Error Variancesa
Dependent Variable: LearndMeasure
F
df1
10.031
df2
2
Sig.
39
.000
c.
Report: since the p-value is 0.000 which is less than 0.04 then the
variance of dependent variables are not equal across groups).
Test Results
Dependent Variable: LearndMeasure
Source
Sum of Squares
Contrast
df
Mean Square
1542.209
771.104
320.337
38
8.430
Error
Sig.
91.472
.000
df
Mean Square
Sig.
Squares
2422.734a
807.578
95.799
.000
Intercept
272.844
272.844
32.366
.000
PreMeasure
650.877
650.877
77.210
.000
1542.209
771.104
91.472
.000
Error
320.337
38
8.430
Total
18249.000
42
2743.071
41
Corrected Model
Method
Corrected Total
d. Statistic value: F= 91.472 and p-value = 0.000 < 0.04. upon this result we reject the null
hypotheses that the dependant has equal means across the groups.
Contrast Results (K Matrix)
Method Simple Contrast
Dependent
Variable
LearndMeasure
Contrast Estimate
Hypothesized Value
3.678
0
3.678
Std. Error
1.103
Sig.
.002
Lower Bound
1.445
Difference
Upper Bound
5.911
Contrast Estimate
14.360
Hypothesized Value
14.360
Std. Error
1.105
Sig.
.000
Lower Bound
12.123
Difference
Upper Bound
16.597
a. Reference category = 1
This table to show that group 2and 1 / 3and 1 are significantly different (0.002 <0.04 and
0.000 <0.04) respectively.
9.
Example:
If we check the performance for 10 employees about how to work in three different
companies ( X, Y, Z) in the morning and afternoon. Each employee performance is
tested six times. And the scores of the test was as follows:
Morning
Subject
1
2
3
4
5
6
7
8
9
10
Solution:
X
9
8
8
9
10
7
9
10
8
9
Y
12
11
10
12
12
10
14
13
12
13
Z
15
14
14
13
15
14
15
16
15
16
X
7
9
10
8
9
10
7
8
10
9
Afternoon
Y
9
10
12
10
11
13
10
11
14
10
Z
12
12
13
12
13
15
13
14
15
13
Within-Subjects Factors
Measure: MEASURE_1
day
1
com
1
2
Dependent
Variable
Morningx
Morningy
Morningz
Afternoonx
Afternoony
Afternoonz
Descriptive Statistics
Morningx
Morningy
Morningz
14.70
.949
Afternoonx
8.70
1.160
Afternoony
11.00
1.563
Afternoonz
13.20
1.135
Mean
Std. Deviation
8.70
.949
11.90
1.287
Epsilon(a)
Within Subjects Effect Mauchly's W
day
1.000
com
.807
day * com
.868
Approx. ChiSquare
.000
1.711
df
GreenhouseGeisser
Huynh-Feldt Lower-bound
.
1.000
1.000
1.000
.425
.839
1.000
.500
Sig.
0
2
1.129
2
.569
.884
1.000
.500
Tests the null hypothesis that the error covariance matrix of the orthonormalized transformed dependent variables is
proportional to an identity matrix.
a May be used to adjust the degrees of freedom for the averaged tests of significance. Corrected tests are displayed
in the Tests of Within-Subjects Effects table.
b Design: Intercept
Within Subjects Design: day+com+day*com
Report:
shows that for each of three effects ( day, company, and interaction ) pvalue > 0.05 so there is no effect.
Mauchly's Test
10. Multi-way
Example:
If we want to determine the effect between four printers ( so3, so5, MH, HK) and the
ink that we used ( cartridge, toner) on 24 papers , and we check the quality from
the lowest quality(1) to the highest (5) . the data are given as follows:
Printer
Paper
Number
cartridge
toner
p. No
t1
t2
t3
t4
t5
t1
t2
t3
t4
t5
So3
p1
p4
So5
p2
p3
p7
2
1
2
1
0
1
1
2
2
2
1
2
1
1
2
p5
p6
p10
0
1
3
1
1
2
1
2
2
2
2
4
2
3
1
p8
p9
2
1
1
1
2
0
1
1
3
3
p11
p12
1
2
1
2
2
3
2
2
2
3
MH
p13
p16
HK
p14
p15
p19
2
2
5
1
1
4
4
2
4
2
4
3
3
1
4
p17
p18
p22
2
2
3
2
4
4
4
1
4
2
3
1
0
2
1
p20
p21
3
3
5
5
3
3
1
0
2
1
p23
p24
0
1
5
2
1
4
2
0
3
1
Clime:.
H0: significant effect
H1: no significant effect
Spss solution:
Analyze general linear model repeated measures within- subject factor
name{weeks} number of levels {5} add define within- subject variables trial(1)
{W1, W2, W3, W4, W5} between subject factors {printer and ink} ok.
Sphericity Assumed
df
4
Mean Square
2.363
Sig.
3.489
.012
test * ink
test * printer
Error(test)
Greenhouse-Geisser
9.450
3.395
2.784
3.489
.018
Huynh-Feldt
9.450
4.000
2.363
3.489
.012
Lower-bound
9.450
1.000
9.450
3.489
.080
Sphericity Assumed
1.917
.479
.708
.590
Greenhouse-Geisser
1.917
3.395
.565
.708
.568
Huynh-Feldt
1.917
4.000
.479
.708
.590
Lower-bound
1.917
1.000
1.917
.708
.413
Sphericity Assumed
6.283
12
.524
.773
.675
Greenhouse-Geisser
6.283
10.185
.617
.773
.656
Huynh-Feldt
6.283
12.000
.524
.773
.675
Lower-bound
6.283
3.000
2.094
.773
.526
Sphericity Assumed
31.417
12
2.618
3.867
.000
Greenhouse-Geisser
31.417
10.185
3.085
3.867
.001
Huynh-Feldt
31.417
12.000
2.618
3.867
.000
Lower-bound
31.417
3.000
10.472
3.867
.030
Sphericity Assumed
43.333
64
.677
Greenhouse-Geisser
43.333
54.319
.798
Huynh-Feldt
43.333
64.000
.677
Lower-bound
43.333
16.000
2.708
Report:
There is no significant effect between test and ink p-value 0.812 > 0.05.
11. Cranachs
alpha:
Example:
Al- Watanya Company want to evaluate the performance of the first level
employees, for 2012 year, so it made a ten question questionnaire from five scale
from ( high) to ( low) as in the following table, and the company test it on 16
supervisor:
Question
NO.
1
2
3
4
5
6
7
8
9
Degre of performe
High
low
Degree of customer satisfaction5 4 3 2 1
Degree of committiment in the job days5 4 3 2 1
How quickly the jobs done5 4 3 2 1
Degree of respect between employees
5 4 3 2 1
Degree of sharing information between employees
5 4 3 2 1
Are the customer satisfied
5 4 3 2 1
Degree of how important the time for employees
5 4 3 2 1
Degree of jobs efficiency
5 4 3 2 1
Degree of love between employees
5 4 3 2 1
Questions
10
Q9
Q8
Q7
Q6
Q5
Q4
Q3
Q2
Q1
NO
10
11
12
13
14
15
16
Reliability Statistics
Cronbach's Alpha
Based on
Cronbach's
Standardized
Alpha
Items
N of Items
.829
.835
10
Q1
Q2
Q1
1.000
.119
Q3
Q4
Q2
Q3
Q4
.119
1.000
.016
.135
.016
.135
.276
.000
Q5
.069
Q6
.061
Q7
Q8
Q9
Q5
Q6
Q7
Q8
Q9
.276
.000
.069
.000
.061
.308
.064
.180
.460
.258
.437
.160
1.000
.244
.545
.484
.509
.313
.367
.244
1.000
.149
.715
.232
.600
.372
.000
.545
.149
1.000
.415
.726
.149
.462
.308
.484
.715
.415
1.000
.461
.556
.345
.064
.180
.509
.232
.726
.461
1.000
.325
.546
.460
.258
.313
.600
.149
.556
.325
1.000
.620
.437
.160
.367
.372
.462
.345
.546
.620
1.000
.164
-.126
.221
.683
The covariance matrix is calculated and used in the analysis.
.364
.633
.339
.618
.545
Q10