UNS/CASS DBA Program
Subject: Assignment of the Quantitative and
Qualitative Research Methodology
Lecturer: Dr Alice Chung
Name: D Yuen
Submission Date: 25 July 2012
SPSS Assignment
Suppose we have captured 1,500 records from Century 21 Real Estate
Company. They have recorded the properties market values at Dec 31, 2010 &
Dec 31, 2011.
1. Draw a histogram with properties building footages, and their market value
in 2011. Discuss the characteristics of those 2 distributions.
2. Test whether the average property market value in 2011 is significantly
different from $4,750,000 at 1% level of significance.
3. Suppose the 1,500 properties are separated into 2 districts: T.S.T. (Group 1)
& Jordan (Group2). Test whether there is any difference between the 2
districts in year of 2010 at 1% level of significance (Assume Equal
Variances).
4. Test whether the average property market value in 2010 is significantly
different from those in 2011 at 3% level of significance.
Answers:
1. Draw a histogram with properties building footages, and their market
value in 2011. Discuss the characteristics of those 2 distributions.
Answer:
The histograms, one shows the properties by their market values while another
by their sizes in 2011 are showed below. The values of skewness and kurtosis
of the first histogram in Figure 1.1 (that is, by the property market values) are
0.201 and -1.112 respectively, while the values for a normal distribution should
be both zero (Table 1.1). The skewness value of 0.201 shows a slight pile-up
of scores on the left of the distribution (the mean of $4.62 million is still close
to the median of $4.36 million). The kurtosis value of -1.112 (a negative value)
indicates the tails of the distributions are lighter than that of normal
distribution.
Figure 1.1
On the other hand, the values of skewness and kurtosis of the second histogram
in Figure 1.2 (that is, by the property sizes) are 1.411 and 3.094 respectively in
Table 1.1. The positive skewness value of 1.411 indicates that the scores are
piled up on the left of the distribution. Similarly we may infer such pattern of
distribution from the mean of 841.38 ft 2 which is larger than the median of 730
ft2. A positive value of kurtosis value of 3.094 (a positive value) shows that the
distribution is pointy and heavy-tailed.
Figure 1.2
Table 1.1
Statistics
Building ft2
N
Valid
2011 Price ($M)
1500
1500
810
810
Mean
841.38
4.620
Std. Error of Mean
13.715
.0690
Median
730.00
4.360
980
6.9
531.170
2.6710
282141.822
7.134
1.411
.201
.063
.063
3.094
-1.112
Std. Error of Kurtosis
.126
.126
Range
4297
9.7
Minimum
123
.2
Maximum
4420
9.8
1262071
6930.0
Missing
Mode
Std. Deviation
Variance
Skewness
Std. Error of Skewness
Kurtosis
Sum
We may convert the values of skewness and kurtosis to z scores to test their
significance. The z score can be derived from skewness or kurtosis values
minus the mean of distribution (which is zero in a normal distribution), and
then divided by their respective standard error in Table 1.2 below:
2011 Property market Property sizes
values
Skewness
0.201
Standard
error
of 0.063
skewness
Z score of skewness
3.19
1.411
0.063
Kurtosis
Standard
kurtosis
3.094
0.126
error
-1.112
of 0.126
22.397
Z score of kurtosis
-8.825
24.556
At the level of significance of 0.05, the scores are significant if the absolute
value of z scores are larger than 1.96. From the above table, it is evident that
the z scores of skewness and kurtosis of both 2011 property market values and
property sizes are significant. In other words, they are not normally distributed.
To test the normality of distribution, we may also use the Kolmogorov-Smirnov
test, supplemented by the Q-Q plot. A significance value of less than 0.05
indicates a deviation from normality. On a Q-Q plot, if the data are normally
distributed, then the observed values would be distributed along the diagonal
straight line. When the plotted dots are consistently below or above the
diagonal straight line, then the kurtosis differs from a normal distribution.
When the plotted dots are s-shaped, there is a problem of skewness.
Table 1.3
Tests of Normality
Kolmogorov-Smirnova
Statistic
df
Shapiro-Wilk
Sig.
Statistic
df
Sig.
Building ft2
.141
1500
.000
.895
1500
.000
2011 Price ($M)
.093
1500
.000
.957
1500
.000
a. Lilliefors Significance Correction
The significance values of both 2011 property market values and property sizes
under the Kolmogorov-Smirnov test are zero in Table 1.3 above. Therefore,
the distributions of both variables are non-normal.
As regards the Q-Q plots, the plotted dots of both variables are s-shaped in
Figures 1.3 and 1.4 and so there is a problem of skewness.
Figures 1.3
Figures 1.4
2. Test whether the average property market value in 2011 is significantly
different from $4,750,000 at 1% level of significance.
Answer:
The hypotheses are:
H0: the average property market value in 2011 = $4,750,000
H1: the average property market value in 2011 $4,750,000
Reject H0 if the p-value < 0.01 (two-tailed test)
Table 2.1
One-Sample Statistics
N
2011 Price ($M)
Mean
1500
Std. Deviation
4.620
2.6710
Std. Error Mean
.0690
One-Sample Test
Test Value = 4.75
t
2011 Price ($M)
df
-1.885
Sig. (2-tailed)
1499
.060
Mean Difference
-.1300
One-Sample Test
Test Value = 4.75
99% Confidence Interval of the
Difference
Lower
2011 Price ($M)
Upper
-.308
.048
After running the one-sample t-test in Table 2.1, the significance value is 0.06.
In other words, the null hypothesis cannot be rejected. There is no statistical
evidence that the average property market value in 2011 is not equal to
$4,750,000 at the 1 % level of significance.
3. Suppose the 1,500 properties are separated into 2 districts: T.S.T.
(Group 1) & Jordan (Group2). Test whether there is any difference
between the 2 districts in year of 2010 at 1% level of significance (Assume
Equal Variances).
Answer:
The hypotheses are:
H0: the average property market value in 2010 at T.S.T. = the average property
market value in 2010 at Jordan
H1: the average property market value in 2010 at T.S.T. the average property
market value in 2010 at Jordan
Reject H0 if the p-value < 0.01 (two-tailed test)
Table 3.1
Group Statistics
Location
2010 Price ($M)
Mean
Std. Deviation
Std. Error Mean
T.S.T.
615
5.160
2.6171
.1055
Jordan
885
5.541
2.7277
.0917
Independent Samples Test
Levene's Test for Equality of
Variances
F
2010 Price ($M)
Equal variances assumed
Equal variances not
assumed
t-test for Equality of Means
Sig.
1.999
t
.158
df
-2.706
1498
-2.727
1354.703
Independent Samples Test
t-test for Equality of Means
Std. Error
Sig. (2-tailed)
2010 Price ($M)
Mean Difference
Difference
Equal variances assumed
.007
-.3812
.1408
Equal variances not assumed
.006
-.3812
.1398
Independent Samples Test
t-test for Equality of Means
99% Confidence Interval of the
Difference
Lower
2010 Price ($M)
Upper
Equal variances assumed
-.7444
-.0179
Equal variances not assumed
-.7418
-.0206
After running the two-sample t-test (independent samples), the significance
value is 0.007 (for equal variances assumed) in Table 3.1. In other words, the
null hypothesis can be rejected. There is no statistical evidence that the
average property market value in 2010 at T.S.T. is equal to that of Jordan at the
1 % level of significance.
The question assumes Equal Variances, and the SPSS output also indicates that
they are (F(1, 1498)=1.999). If the Levenes Test for equality of variances
shows a p-value < 0.05, the variances are significantly different and the
assumption of homogeneity of variances has been violated. As the SPSS output
states a significance value of 0.158, we do not have sufficient evidence to reject
the null hypothesis that the difference between the variances is zero.
4.
Test whether the average property market value in 2010 is
significantly different from those in 2011 at 3% level of significance.
Answer:
The hypotheses are:
H0: the average property market value in 2010 = the average property market
value in 2011
H1: the average property market value in 2010 the average property market
value in 2011
Reject H0 if the p-value < 0.03 (two-tailed test)
Table 4.1
Paired Samples Statistics
Mean
Pair 1
Std. Deviation
Std. Error Mean
2010 Price ($M)
5.385
1500
2.6886
.0694
2011 Price ($M)
4.620
1500
2.6710
.0690
Paired Samples Correlations
N
Pair 1
2010 Price ($M) & 2011 Price
Correlation
1500
Sig.
.651
.000
($M)
Paired Samples Test
Paired Differences
Mean
Pair 1
2010 Price ($M) - 2011 Price
($M)
.7647
Std. Deviation
2.2392
Std. Error Mean
.0578
Paired Samples Test
Paired Differences
97% Confidence Interval of the
Difference
Lower
Pair 1
2010 Price ($M) - 2011 Price
Upper
.6391
.8903
($M)
Paired Samples Test
t
Pair 1
2010 Price ($M) - 2011 Price
13.226
df
Sig. (2-tailed)
1499
.000
($M)
After running the two-sample t-test (related samples), the significance value is
0 in Table 4.1. In other words, the null hypothesis can be rejected. There is no
statistical evidence that the average property market value in 2010 is equal to
that of 2011 at the 3 % level of significance.
Reference List
Field, A. (2009). Discovering Statistics Using SPSS. Third Edition. SAGE
Publications Ltd.
Gujarati, D. N. (2003). Basic Econometrics. Fourth Edition. McGraw-Hill
Higher Education.
Salvatore, D. and Reagle, D. (2002). Schaums Outline of Theory and Problems
of Statistics and Econometrics.
Second Edition.
The McGraw-Hill
Companies, Inc., Kindle Edition. Retrieved from www.amazon.com
