Professional Documents
Culture Documents
FOR THE DEGREE OF MASTER OF PHILOSOPHY
Monday 04 January 2010 09.0011.00
RM01 Quantitative Research Methods I
Two Hour Examination.
This paper is worth 50% of the total mark for this module.
All answers should be given to two decimal places if necessary.
All questions carry equal weight. Where a question has several parts, each
part carries equal weight.
Stationery Requirements Special Requirements
1 x 20 page answer booklet. An approved calculator may be used.
Rough Notes pad.
You may not start to read the questions
printed on the subsequent pages of this
question paper until instructed that you
may do so by the Invigilator
2
SECTION A
1. A linear regression model is estimated using a sample of 4,682 residential
property transactions from January 2001 to December 2006. The data set
contains the following variables.
LNPRICE: natural log transformed house price
PA: age of the property in years divided by 10
PA2: PA squared
SQFT: size of the property in squared feet
SQFT2: SQFT squared
Y2002Y2006: annual time dummy variables
Descriptive statistics on some of the variables are given as follows.
Mean Median Maximum Minimum Std. Dev.
LNPRICE 11.55 11.57 12.83 9.90 0.33
PA 2.50 1.80 9.60 0.00 2.01
SQFT 14.49 14.36 22.00 6.00 3.41
The computer output of the regression model is given below.
ANOVA
Significance
df SS MS F F
Regression 9 317.63 35.29 828.10 <0.01
Residual 4672 199.11 0.04
Total 4681 516.74
Standard
Coefficients Error t Stat
Intercept 10.3185 0.0514
PA 0.0001 0.0045
PA2 0.0032 0.0006
SQFT 0.1248 0.0069
SQFT2 0.0019 0.0002
Y2002 0.0312 0.0116
Y2003 0.0834 0.0113
Y2004 0.1097 0.0111
Y2005 0.1718 0.0113
Y2006 0.2211 0.0109
3
a) Calculate R 2 of the regression model and t test statistics on all independent
variables.
b) Discuss the effect of age and size on house prices.
c) What is the estimated change of property prices from 2001 to 2003? Test
the claim that property price dropped by more than 5% from 2001 to 2003.
d) Construct a property price index using year 2001 as the base year.
SECTION B
LNWAGE: natural log transformed earnings per hour
EDUC: years of education
EXPER: experience in years
FEMALE: = 1 if female and 0 otherwise
FULLTIME: = 1 if fulltime worker and 0 otherwise
MARRIED: = 1 if married and 0 otherwise
F_M: = FEMALE * MARRIED
Correlation Matrix
EDUC EXPER FEMALE FULLTIME MARRIED
EDUC 1.00 0.18 0.02 0.05 0.06
EXPER 0.18 1.00 0.01 0.03 0.29
FEMALE 0.02 0.01 1.00 0.22 0.07
FULLTIME 0.05 0.03 0.22 1.00 0.01
MARRIED 0.06 0.29 0.07 0.01 1.00
TURN OVER
4
Two regression models are estimated as follows:
Model 1
Variable Coefficient Std. Error tStatistic Prob.
Intercept 0.1670 0.0934 1.79 0.07
EDUC 0.1087 0.0060 18.22 <0.01
EXPER 0.0102 0.0014 7.55 <0.01
MARRIED 0.1429 0.0306 4.68 <0.01
FULLTIME 0.3229 0.0431 7.48 <0.01
Adjusted Rsquared 0.3600 Fstatistic 122.4549
Sum squared residuals 204.5787
Model 2
Variable Coefficient Std. Error tStatistic Prob.
Intercept 0.3163 0.0943 3.36 <0.01
EDUC 0.1083 0.0058 18.65 <0.01
EXPER 0.0103 0.0013 7.74 <0.01
FEMALE 0.1388 0.0433 3.21 <0.01
MARRIED 0.1890 0.0423 4.47 <0.01
FULLTIME 0.2413 0.0434 5.55 <0.01
F_M 0.1221 0.0575 2.13 0.03
Adjusted Rsquared 0.3622 Fstatistic 95.5626
Sum squared residuals 193.5372
a) Based on Model 2, identify the sample regression equation for married
female workers and single male workers respectively.
b) Test the claim that gender has no influence on wage (use a 1%
significance level).
c) Compare the coefficient estimates on EDUC and FULLTIME in the two
models. Explain why omitting variable FEMALE and F_M has a different
effect on these two coefficient estimates.
5
3. Using a sample of 392 observations, two regression models are estimated.
Model 1
Variable Coefficient Std. Error tStatistic Prob.
Intercept 46.22 0.80 57.87 <0.01
WGT 0.01 0.00 29.65 <0.01
Adjusted Rsquared 0.69 Sum squared residuals 7321
Model 2
Variable Coefficient Std. Error tStatistic Prob.
Intercept 44.37 1.48 29.97 <0.01
CYL 0.27 0.41 0.65 0.52
ENG 0.01 0.01 1.54 0.13
WGT 0.01 0.00 8.00 <0.01
Adjusted Rsquared 0.70 Sum squared residuals 7163
Correlation matrix of independent variables
ENG CYL WGT
ENG 1.00 0.95 0.93
CYL 0.95 1.00 0.90
WGT 0.93 0.90 1.00
a) Test if ENG and CYL are individually significant at the 5% level.
b) Test if ENG and CYL are jointly significant at the 5% level.
c) Contrast and discuss your answers to part a) and b).
TURN OVER
6
5. Illustrate how to interpret coefficient estimates of
a) interaction terms between two dummy variables
b) interaction terms between one dummy variable and one continuous
variable
c) interaction terms between two continuous variables.
Note: Useful formulas and critical values
END OF PAPER