Final Exam Fall

Final Exam Fall 2020 ECON 390

Prof. Gharehgozli

1

Instruction

• Thursday Dec 3rd to Wednesday Dec 9th. This is Final Exam and the due date is

Wednesday. Late submissions (even for 1 seconds) receive 0.

• Your submission must include a rscript that is error-less, with your name on top of

the script.

• All questions must have numbers in the script. EXAMPLE: “#Questions 1 Part a:”.

Note that without the correct rscript I will not grade the submission.

• Please respond to each and every single question on the PAPER as well with numbers

and sub-numbers. Otherwise I would consider 0 for your answer if you do not mention

what question you are responding to exactly, even if you have the correct answer. For

example if you are responding to question 1 part a, in your response you should start

with 1.a. …Again if you are using RStudio I want the response on both the script as

well as on paper :).

• Important: if you work with classmates you should submit 1 submission and both

your final exam scores will be multiplied by 0.9. If you collaborate and do not inform

me your score will be 0. Forget about the group chats :).

The data description is attached along with data.

Questions

Follow the instruction above and answer questions below:

Question 1 : (15/100)

• Question1 part a (5/100): Please refer to the data description, and define variables

”alcohol”, ”birthweight”, ”smoker”, ”unmarried”, ”drinks”, ”nprevist”, and ”tripre3”;

2

meaning what they are and what is their unit of measurement.

• Question1 part b (5/100): Provides summary of the variables mentioned above: min,

max, mean, and variance of each variable.

• Question1 part b (5/100): How many observations do we have? What is the unit of

observations (is it individuals, households, states,…)? How many dummy variables do

we have?

Question 2 : (20/100)

• Question 2 part a (5/100): Run a regression of birthweight over age.

• Question 2 part b (5/100): What is the model and the predicted equations?

• Question 2 part c (5/100): What is the slope and what is the interpretation of this

number?

• Question2 part d (5/100): Is the slope significant? write down the null and alternative

of this hypothesis testing, write down the tstat/zscore formula and calculate it and

provide your conclusion with the 2 approaches, tstat/zscore and pvalue approaches.

Question 3 : (35/100)

• Question 3 part a (5/100): Run a regression of birthweight over age, educ, unmarried,

smoker, alcohol.

• Question 3 part b (5/100): What is the model?

• Question 3 part c (5/100): What is the predicted equation?

• Question 3 part d (5/100): Among all the coefficients, which ones are significant and

which ones are not. WHY?

• Question 3 part e (10/100): In this model , run a ”F” test checking for simultaneous

significance of mother’s unhealthy behaviors (smoker and alcohol variables) on the

birthweight of the infants. (Hint: the regression in Question3 part a is your unre-

stricted model. Create you restricted model and run the Ftest. Important: what is

the null and alternative of this test, and what is your conclusion? calculate the Ftest

3

on paper and use the table for the critical value. You can then double check your

result with the anova command)

• Question 3 part e (5/100): In this model (part 3.a ) what is the predicted value of

infant’s birth weight for a mother who is 27 years old, who has 13 years of education,

who is married, who is not a smoker and does not drink alcohol.

Question 4 : (10/100)

• Question 4 part a (5/100): Create log of birthweight and run a single ”log-linear”

model of log(birthweight) over age.

• Question 4 part c (5/100): What is the predicted slope and what is the interpretation

of the slope? (Note it is log-linear so the interpretation is not regular)

Question 5 : (15/100)

• Question 5 part a (5/100): Provide a scatterplot of birthweight over nprevist.

• Question 5 part b (5/100): Create quadratic and cubic terms for nprevist, then run

a regression of birthweight over nprevist, nprevist2, nprevist3, unmarried, and smoker.

• Question 5 part c (10/100): Is this a good model or do we have to go with a quadratic

or linear relationship between birthweight and nprevist? Why?