Empirical Research Methods for Real Estate Market Analysis

This article discusses the empirical research methods used for real estate market analysis including variable exploration, normality check, correlation analysis and regression model construction. The article also includes descriptive statistics, confidence intervals and plots for the variables analyzed. The linear regression model constructed to assess the reliance of price on distance from bus and railway station was found to be insignificant. Desklib provides solved assignments, essays, dissertations and more for students.

11
Empirical Research Methods for Business

11
Answer 1
The real estate market conditions were analyzed by the researcher in the data analysis
segment. The dependence of selling price of properties on the distance of the house to the nearest
train station and the nearest bus stop was investigated. The report has been presented as follows.
Variable Exploration
a) Price
The selling price of the real estate properties (M = $ 886580. SD =$ 324950) was noted
to be approximately normally distributed (SKEW = 0.43). The median of selling price was
identified to be at $ 852 and was noted to be less than the mean value. With 95% confidence, the
average selling price for real estate properties in Australia was estimated to be within [$ 824200,
$ 942530].
Table 1: Descriptive Statistics for Price (in $’000)
886.58
29.66
852.00
811.00
324.95
0.43
0.22
-0.15
0.44
1569.00
192.00
1761.00
25 633.75
50 852.00
75 1087.25
Mean
Std. Error of Mean
Median
Mode
Std. Deviation
Skewness
Std. Error of Skewness
Kurtosis
Std. Error of Kurtosis
Range
Minimum
Maximum
Percentiles
Table 2: Confidence Interval of Average Selling Price of Properties
Lower Upper
886.58 0.06 29.32 824.20 942.53
Bias Std. Error
95% Confidence
Mean
Statistic
Bootstrap a

11
The variable (Price) was plotted in a histogram and has been presented in Figure 1. The
normal line has also fitted the histogram. The variable was noted to follow Gaussian nature with
almost near zero skewness. The confirmatory tests was conducted using Shapiro-Wilk (W =
0.98, p = 0.12) and Kolmogorov-Smirnoff (S = 0.06, p = 0.20) tests. The null hypothesis failed to
get rejected at 5% level of significance, and it was concluded that the price of the real estate
properties was normally distributed (Corder, & Foreman, 2014).
Figure 1: Histogram for the Selling Price of Real Estate Properties
Table 3: Normality Check with Shapiro-Wilk and Kolmogorov-Smirnoff Tests
Statistic df Sig. Statistic df Sig.
Price 0.06 120.00 .200* 0.98 120.00 0.12
LotsizeSQ 0.23 120.00 0.00 0.86 120.00 0.00
Material 0.25 120.00 0.00 0.78 120.00 0.00
Condition 0.20 120.00 0.00 0.88 120.00 0.00
Kolmogorov-Smirnova Shapiro-Wilk
b) Lot Size
Lot size of the real estate properties (M = 1175.23 Sq.M, SD = 372.90 Sq.M) was noted
to be have a positive skewness (SKEW = 0.84). The median of lot sizes was identified to be at
980 square meters and was noted to be highly less than the mean value. With 95% confidence,
the average lot size for real estate properties in Australia was estimated to be within [1107.26,
1242.16] square meters.

11
Table 4: Descriptive Statistics for Lot Size (Sq .M)
N Valid 120.00
1175.23
34.04
980.00
890.00
372.90
0.84
0.22
-0.56
0.44
1318.00
632.00
1950.00
25 910.00
50 980.00
75 1438.00
Minimum
Mean
Std. Error of Mean
Median
Mode
Std. Deviation
Maximum
Percentiles
Skewness
Std. Error of Skewness
Kurtosis
Std. Error of Kurtosis
Range
Table 5: Confidence Interval of Average Lot Size of Properties
Lower Upper
1175.23 -1.97 34.75 1107.26 1242.16
95% Confidence
Mean
Statistic
Bootstrap a
Bias Std. Error
The lot size was plotted in a box-plot and has been presented in Figure 2. The variable
was noted to be significantly positively skewed. The confirmatory tests was conducted using
Shapiro-Wilk (W = 0.86, p < 0.05) and Kolmogorov-Smirnoff (S = 0.23, p < 0.05) tests. The null
hypothesis got rejected at 5% level of significance, and it was concluded that lot sizes of the real
estate properties were not normally distributed.
Figure 2: Box Plot for Lot Size of the Properties

11
c) Material
The most used material for the real estate properties was noted to be Veneer (Median = 2)
and was also prominent to have a positive low skewness (SKEW = 0.10). The variable was
categorical in nature, and the median was considered as the proper descriptive statistic. With
95% confidence, the estimated interval was found to be [1.81, 2.09] which contained 2 (Veneer)
as the estimated material for real estate construction.
Table 6: Descriptive Statistics for Materials Used
N Valid 120.00
1.95
0.08
2.00
1.00
0.84
0.10
0.22
-1.58
0.44
2.00
1.00
3.00
25 1.00
50 2.00
75 3.00
Mean
Std. Error of Mean
Median
Mode
Std. Deviation
Skewness
Std. Error of Skewness
Kurtosis
Std. Error of Kurtosis
Range
Minimum
Maximum
Percentiles
Table 7: Confidence Interval of Average Material Used
Lower Upper
1.95 0.00 0.07 1.81 2.09
95% Confidence
Mean
Statistic
Bootstrap a
Bias Std. Error
The material was plotted in a histogram and has been presented in Figure 3. The
distribution of this categorical variable was noted to be not normal. The confirmatory tests was
conducted using Shapiro-Wilk (W = 0.86, p < 0.05) and Kolmogorov-Smirnoff (S = 0.25, p <
0.05) tests. The null hypothesis got rejected at 5% level of significance, and it was concluded
that material of the real estate properties was not normally distributed.

11
Figure 3: Histogram for Material Used for Real Estate Properties
d) Condition
The most observed condition for the real estate properties was noted to be good (Median
= 3) and was also prominent to have a low negative skewness (SKEW = - 0.07). The variable
was categorical in nature, and the median was considered as the proper descriptive statistic. With
95% confidence, the estimated interval for the average condition was found to be [2.43, 2.77]
which contained 3 (Good condition) as the estimated condition for real estate structures.

11
Table 8: Descriptive Statistics for Condition of the Properties
N Valid 120.00
2.61
0.09
3.00
3.00
0.94
-0.07
0.22
-0.87
0.44
3.00
1.00
4.00
25 2.00
50 3.00
75 3.00
Minimum
Mean
Std. Error of Mean
Median
Mode
Std. Deviation
Maximum
Percentiles
Skewness
Std. Error of Skewness
Kurtosis
Std. Error of Kurtosis
Range
Table 9: Confidence Interval of Average Condition of the Real Estate Structures
Lower Upper
2.61 0.00 0.09 2.43 2.77Mean
Statistic
Bootstrap a
Bias Std. Error
95% Confidence
The condition was plotted in a histogram and has been presented in Figure 4. The
distribution of this categorical variable was noted to be not normal. The confirmatory tests was
conducted using Shapiro-Wilk (W = 0.88, p < 0.05) and Kolmogorov-Smirnoff (S = 0.20, p <
0.05) tests. The null hypothesis got rejected at 5% level of significance, and it was concluded
that condition of the real estate properties was not normally distributed.

11
Figure 4: Histogram for Condition of the Real Estate Properties
Answer 2
The price of the real estate properties was observed earlier to be normal in nature,
whereas the distance from the train (To train) (W = 0.96, p < 0.05) and bus (To bus) stations (W
= 0.93, p < 0.05) were found to be not normally distributed.
Table 10: Normality Check with Shapiro-Wilk and Kolmogorov-Smirnoff Tests
Statistic df Sig. Statistic df Sig.
Price 0.055 120.000 .200* 0.982 120.000 0.116
To Train 0.095 120.000 0.010 0.961 120.000 0.002
ToBus 0.140 120.000 0.000 0.926 120.000 0.000
Kolmogorov-Smirnova Shapiro-Wilk
Price of the properties was found to have a low negative and insignificant (r = -0.024, p =
0.796) correlation with a distance of the properties from the nearest railway station. Price of the
properties was found to have a low negative and insignificant (r = 0.003, p = 0.974) correlation
with the distance of the properties from the nearest bus station.

11
Table 11: Correlation of Price with Distance of Train and Bus stations
Price To Train ToBus
Pearson
Correlation
1.000 0.003 -0.024
Sig. (2-tailed) 0.974 0.796
Pearson
Correlation
0.003 1.000 -0.002
Sig. (2-tailed) 0.974 0.980
Pearson
Correlation
-0.024 -0.002 1.000
Sig. (2-tailed) 0.796 0.980
Price
To Train
ToBus
To assess the reliance of price on distance from bus and railway station, a backward
regression model was constructed. The assumptions were also cross-checked for the validity of
the model. From Table 12 it was evident that none of the models were able to explain the
dependent variable, the price of the real estate properties significantly. The coefficient of
determinations was almost zero for the three models (Landers, 2015).
Table 12: Changes in R Square of the Regression Models
R Square
Change F Change df1 df2
Sig. F
Change
1 .024a 0.001 -0.017 327.618 0.001 0.034 2.000 117.000 0.967
2 .024b 0.001 -0.008 326.228 0.000 0.001 1.000 117.000 0.975
3 .000c 0.000 0.000 324.947 -0.001 0.067 1.000 118.000 0.796
Model R R Square
Adjusted R
Square
Std. Error of
the Estimate
Change Statistics
a. Predictors: (Constant), ToBus, To Train
b. Predictors: (Constant), ToBus
c. Predictor: (constant)
From Table 13 it was noted that all the three regression models were insignificant in
nature. Hence, none of the independent variables were the valid predictor of the models.

11
Table 13: ANOVA for the Three Regression Models
Sum of Squares df Mean Square F Sig.
Regression 7246.590 2.000 3623.295 0.034 .967b
Residual 12558002.735 117.000 107333.357
Total 12565249.325 119.000
Regression 7136.788 1.000 7136.788 0.067 .796c
Residual 12558112.537 118.000 106424.683
Total 12565249.325 119.000
Regression 0.000 0.000 0.000 .d
Residual 12565249.325 119.000 105590.330
Total 12565249.325 119.000
Model
1
2
3
a. Dependent Variable: Price
b. Predictors: (Constant), ToBus, To Train
c. Predictors: (Constant), ToBus
d. Predictor: (constant)
From the linear model it was found that the coefficients of the predictors were
statistically insignificant, and therefore linear relation between the variables was not feasible.
The co-linearity statistics were measured by VIF, where the VIF values were found to be equal
to one. Hence, the multi-co-linearity was not present in the regression models (Park, 2015).
Table 14: Regression Models
Standardized
Coefficients
B Std. Error Beta Lower Bound Upper Bound Tolerance VIF
(Constant) 896.395 77.863 11.512 0.000 742.190 1050.599
To Train 1.968 61.520 0.003 0.032 0.975 -119.869 123.805 1.000 1.000
ToBus -12.863 49.898 -0.024 -0.258 0.797 -111.683 85.958 1.000 1.000
(Constant) 898.187 53.830 16.686 0.000 791.589 1004.785
ToBus -12.867 49.686 -0.024 -0.259 0.796 -111.259 85.526 1.000 1.000
3 (Constant) 886.575 29.663 29.888 0.000 827.838 945.312
Model
Unstandardized Coefficients
t Sig.
95.0% Confidence Interval
for B Collinearity Statistics
1
2
a. Dependent Variable: Price

11
Table 15: Co-linearity Diagnostic for the Models
(Constant) To Train ToBus
1 2.634 1.000 0.020 0.028 0.037
2 0.270 3.121 0.008 0.322 0.672
3 0.096 5.250 0.972 0.651 0.291
1 1.833 1.000 0.083 0.083
2 0.167 3.313 0.917 0.917
3 1 1.000 1.000 1.000
Condition
Index
Variance Proportions
Model Eigenvalue
1
2
a. Dependent Variable: Price
Table 16: Partial Correlation of the Excluded Variables for the Tree Regression Models
Collinearity
Statistics
Tolerance
2 To Train .003b .032 .975 .003 1.000
To Train .003c .033 .974 .003 1.000
ToBus -.024c -.259 .796 -.024 1.000
3
a. Dependent Variable: Price
b. Predictors in the Model: (Constant), ToBus
c. Predictor: (constant)
Model Beta In t Sig.
Partial
Correlation
For the first regression model, the residual plot in Figure 5 was analyzed. It was found
that the horizontal spread of the plot was between -2 and 2, whereas the vertical spread was
between -3 and 3. Hence, variances of the error terms were spread across the independent
variables in a similarity. Homoscedasticity assumption of the regression model was satisfied
(Bolin, 2014).

11
Figure 5: Residual Plot for Homoscedasticity Test
Answer 3
The satisfaction levels of the employees were verified for the reliability of the thirteen
statements of the dataset. Cronbach's alpha was evaluated for the purpose, and the item deleted
matrix was also analyzed (Yockey, 2017). The value of Cronbach's alpha was evaluated as 0.92
with 13 items, which indicated that the responses were highly and significantly reliable for the
thirteen questions. The inter-item correlation matrix was constructed and provided in Table 17.
From the particular matrix question number 8 was identified separately for possessing low
correlations. The correlations of Q8 with other items were considerably low for reliability
purpose. Probably, Q8 was denoting or representing some other aspects or views other than the
rest of the items. Q11 was had low correlations with other items, especially with Q3 and Q6.
Other than Q8, rest of the items was noted have considerable correlations with other items.

11
Table 17: Inter-Item Correlation Matrix
From the item deleted reliability matrix significance of the items was clear. From Table
18 it was noted that Q2, Q4, Q5, Q7, Q12, and Q13 were discovered as the most significant
impact factors in the reliability of the dataset. Deletion any one of these items would have
reduced the reliability of the dataset. Q1, Q3, Q6, Q9, Q10, and Q11 were those factors, where
removing them one at a time did not affect the overall reliability of the dataset. But, the item
having a different orientation than other items was observed to be Q8. Deletion of Q8 obtained
Chronbach’s alpha = 0.93. Hence, the observation of Q8 as a differently oriented factor other
than the rest of the items was justified from the inter-item correlation matrix (Bonett, & Wright,
2015).

11
Table 18: Item-Total Statistics for Reliability
Scale Mean if
Item Deleted
Scale
Variance if
Item Deleted
Corrected
Item-Total
Correlation
Squared
Multiple
Correlation
Cronbach's
Alpha if Item
Deleted
Q1 19.77 55.96 0.69 0.57 0.92
Q2 20.14 54.94 0.74 0.66 0.91
Q3 19.97 55.62 0.70 0.67 0.92
Q4 20.08 53.71 0.79 0.76 0.91
Q5 20.23 52.74 0.79 0.76 0.91
Q6 20.76 56.61 0.56 0.41 0.92
Q7 20.24 54.49 0.76 0.68 0.91
Q8 20.39 59.53 0.33 0.27 0.93
Q9 20.39 56.34 0.53 0.41 0.92
Q10 20.11 54.06 0.70 0.61 0.92
Q11 20.18 56.64 0.46 0.33 0.92
Q12 20.31 51.69 0.80 0.73 0.91
Q13 20.44 51.65 0.79 0.76 0.91

11
References
Bolin, J. H. (2014). Hayes, Andrew F.(2013). Introduction to Mediation, Moderation, and
Conditional Process Analysis: A Regression‐Based Approach. New York, NY: The
Guilford Press. Journal of Educational Measurement, 51(3), 335-337.
Bonett, D. G., & Wright, T. A. (2015). Cronbach's alpha reliability: Interval estimation,
hypothesis testing, and sample size planning. Journal of Organizational Behavior, 36(1),
3-15.
Corder, G. W., & Foreman, D. I. (2014). Nonparametric statistics: A step-by-step approach.
John Wiley & Sons.
Landers, R. (2015). Computing Intraclass Correlations (ICC) as Estimates of Interrater
Reliability in SPSS, The Winnower 2: e143518. 81744, 2015, DOI: 10.15200/winn.
143518.81744 Landers This article is distributed under the terms of the Creative
Commons Attribution 4.0 International License, which permits unrestricted use,
distribution, and redistribution in any medium, provided that the original author and
source are credited. Recently, a colleague of mine asked for some advice on how to
compute interrater reliability for a coding task, and I discovered that there aren’t many
resources online written in an easy-to-understand.
Park, H. M. (2015). Linear regression models for panel data using SAS, Stata, LIMDEP, and
SPSS.
Yockey, R. D. (2017). SPSS demystified. Taylor & Francis.