Scribd
Upload
26 views3 pages

Mock Test Econ

The document discusses modeling variable Y based on predictors X2 and X3, detailing linear regression for continuous Y and logistic regression for binary Y. It addresses heteroscedasticity, its definition, consequences, and the White test procedure, as well as estimating a model for yearly food expenditure with various predictors and interactions. Additionally, it covers significance testing for the model, the implications of multicollinearity, and logistic regression analysis for student grades based on GPA and study system, including coefficient interpretations and probability calculations.

Uploaded by

QP2164 Le Nguyen Phuong An
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
26 views3 pages

Mock Test Econ

The document discusses modeling variable Y based on predictors X2 and X3, detailing linear regression for continuous Y and logistic regression for binary Y. It addresses heteroscedasticity, its definition, consequences, and the White test procedure, as well as estimating a model for yearly food expenditure with various predictors and interactions. Additionally, it covers significance testing for the model, the implications of multicollinearity, and logistic regression analysis for student grades based on GPA and study system, including coefficient interpretations and probability calculations.

Uploaded by

QP2164 Le Nguyen Phuong An
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Q1: Assume we want to model variable Y on X2 and X3, data is cross sectional.

Propose the model


when: a) Y is continuous; b) Y is binary (no order)

Y is continuous: WE use the linear model


Y= B1+ B2*X2+B3*X3 +u
Y is binary: Binary logistic Regression model:

p=P(Y=1) = eB1+B2*X2+B3*X3 / (1+ eB1+B2*X2+B3*X3)

Assume we consider the linear model (Y on X2 and X3). We want to check for the problem of
heteroscedasticity. Discuss this problem on the following aspects: i) The defintion; ii) The
consequences; iii) The procedure of White test to detect the problem.

Q2 Given the following model estimation in which the variables are:


Exp: Yearly Expenditure on food ($1000), Married = 1 if married, 0 otherwise, Age=Age in years,
Uni = 1 if having an University degree, 0 otherwise; (standard error of respective coefficient is within
brackets, n=35).

¿ 2
exp 34.9+ 7.4 Age−0.16 Age +7.63 Uni+8.2 Married−0.06 Married∗Age+3.74 Uni∗Age
(1.2) (0.015) (1.35) (1.82) (0.065) (0.42)
1. Write the sample estimated model for each combination of Married and Uni, explain the meaning
of the coefficients. Why we use the square of Age in the model, what is its economic meaning?
2. Given adjusted R2=0.92. Find R2, explain its meaning and test for the significance of the model.
Use the formula of adjusted R square and R2 : explain the meaning of R2
Using F test of overall significance by R2 for the model significance.
(look at the formula of F)

3. Does the rate of change of Exp with pespect to Age decrease when Age increases?
Test the hypothesis: B3 < 0 by t test: t = beta^3/se(beta^3) = -0.16/0.015
 Reject Hi if t < t(n-k)α

4. Check that the rate of change of Exp with respect to Age is different between Married and Not
Married, between people with and without University degree.
We test B5 = 0 for comparing rate of change of Exp by Age between Married and Non married
Similarly, test B6 =0 to compare the rate of change of Exp by Age between Uni and NonUni
Using t test

5. If you want to consider the interaction between Married and Uni, propose the model.
Add the interaction term: B8*(Married*Uni) + B9*(Married*Uni*Age)

6. Test if we need Qualitative factors (Married and Uni) in the model if we estimate the model of Exp
on Age and Age2 having R2=0.84.

This is the test of dropping all variables related to Married and Uni:

Ho: B4=B5=B6= B7, Using F test

7. Do we have perfect multicollinearity in the above model. What are the consequences of imperfect
multicollinearity and how to correct the problem?

No perfect multi in this case because if it has, we can not estimate the model.

Problem 3 (3p): Given the output below for the logistic regression in which GRADE is 1 if students
get A, 0 if not A. GPA is the grade point average before entering the Uni, PSI=1 if study in the new
system, 0 if not. Answer the following questions:
Logistic regression Number of obs = 32
LR chi2(2) = 14.93
Prob > chi2 = 0.0006
Log likelihood = -13.126574 Pseudo R2 = 0.3625

GRADE Coef. Std. Err. z P>|z| [95% Conf. Interval]

GPA 3.063366 1.22287 2.51 0.012 .6665848 5.460148


PSI 2.337775 1.040799 2.25 0.025 .2978461 4.377704
_cons -11.60156 4.212985 -2.75 0.006 -19.85886 -3.344263

1. Write the estimated logistic model and logit model


Logistic: p=P(get A)= e-11.6+3.06*GPA + 2.34*PSI / 1+ e…….)

Logit: ln(p/1-p) = -11.6+3.06*GPA + 2.34*PSI

2. Test for significance of Independent variables


P value in the z test for the significance of two variables, GPA and PSI are smaller than 0.05,

 Reject Ho the two variables are all significant.


3. Explain the meaning of each coefficient (using log odd form)
Logit model:

ln(p/1-p) =-11.6 +3.06*GPA +2.34*PSI

beta^2 = 3.06: If PSI is fixed, GPA increases 1 unit, odd of gettting GRADE A

will increase (e3.06 -1)*100 %

beta^3 = 2.34: GPA fixed, if a student changes the study system from OLD to NEW,

The odd of getting A will increases (e2.34 -1)*100%

4. What is the probability of getting Grade A when a student studying in the new
system and having GPA of 3
Replace GPA=3 and PSI =1 into the logistic model

p=P(get A)= e-11.6+3.06*GPA + 2.34*PSI / 1+ e…….) to calculate p0

5. At the level in question 4, if PSI is fixed, how the probability of getting A changes if
GPA increases 1 point?

Change in Probability is: dp = beta^2* po*(1-p0) ; po is from question 4.

You might also like