The final paper should be 10-15
pages of finished writing,
excluding the references and appendix. Paragraphs should be single-spaced with
a font size 12 pt. (Time New Roman or Arial) The final paper needs to
have the following sections in the order listed:
Title Page:
Should have:
–
your name
–
title
–
abstract (no more than 300 words):
Abstract should have your research question, which is the main relationship
that you are examining, and briefly, your main findings. You can include data
set, variables, and innovations in your paper.
Introduction:
Should include the following:
–
Research question which is the
relationship that you are examining in your paper
–
Why that question is relevant
(motivation)
–
A brief description of the
economic model(s) used in the analysis
–
Summary of the major findings
of the analysis.
–
Final paragraph should have
the outline of the different sections in the paper (section II is lit review,
section III is econometric model etc..)
Literature Review:
Should contain a brief summary of 3
or 4 similar studies conducted in the same area as your paper. When you
summarize a paper, explain the innovation in the paper relative to past
studies, data set, important variables and results that are relevant to your
paper. You should also mention how your paper is similar and different from these
studies that you are reviewing. Try to include regression-based papers, which will
make it easier to compare similarities and differences with your paper. Try to
devote at most 2 pages to this section.
Data and Descriptive Statistics:
–
Describe your data (type of
dataset, source description of data set, sample size etc) and the description
of the dependent and independent variables used in your analyses.
–
It is always important to
correctly identify and handle missing values. Explain how you took care of
missing observations. Non-numerical data also have to be handled with great
care.
–
Don’t devote too many pages in
describing the variables, since Table 1 in appendix will have the variable
definitions. Tables should go in the appendix section.
–
Explain tables 2 and 3
highlighting the interesting trends/findings. You can tie this to past
studies/news/ general observations. Write in paragraphs.
Econometric Model:
Write the equation for the regression
model and state the parameter/s of interest (which is the β associated with the
main x). Explain the econometric models that you are running – DinD, panel data methods (Fixed effects or
Random effects), logit/probit, multinomial logit etc. When using panel methods:
Explain what the unobserved individual specific effects stand for, and how they
are accounted for. In any model, discuss if you suspect omitted variable bias
and how you deal with it.
Also explain any log terms or interaction terms
used, and any specification tests that you are doing. Use Microsoft Equation editor
for writing equations. Explain the economic theory behind the econometric model
that you are estimating. (Why are you using the model you are using? How do you
know it’s the most appropriate? Are the estimators efficient?)
Also, do you expect any heteroskedasticity
or serial correlation? how do you deal with it?
Note: There is no need to write regression
equations for all the models that you are estimating. Only specify the
main one.
General Guidelines:
•
Carefully
choose functional form specifications (logs, squares etc.).
•
Make sure
the key assumptions are satisfied in your model.
•
Beginners
mistake: do not include variables that are listed as numerical values but have
no quantitative meaning (e.g., 3-digit occupations).Transform such variables to
dummy variables representing categories.
•
For binary/limited dependent
variables, use logit/probit
•
Sensitivity
analysis: look at variations of your specification/method. Run the appropriate specification
tests as needed (joint F-tests, Hausman test etc.)
Results:
Explain the relevant and interesting
results in your analysis (Tables 4 – 7). Tables should look professional with
proper titles and labels; they should be created using the “outreg2” command. Say how the main
estimate changes in different models. Refer Table #s and model #s in the tables
when explaining. Pay careful attention to the units when you interpret. You should
link your results to past studies. Results should be in paragraph form. You can
choose to focus more on one result than the other but try to mention all
results in your explanations. In other words, don’t just present tables
without mentioning them in the text of your paper.
Conclusion:
i) Reiterate the relationship that you are examining
iii) State your important result(s) again and the broad
implications of it
iv) What are the drawbacks/limitations of your study
v) Any possible extensions to your paper
References:
You should include a detailed reference
section citing all your work. You can use any of the acceptable formats for
academic work (but all references should follow the same format). Please see
the following link for APA format: APA Style Introduction
// Purdue Writing Lab
The references that you cite at the end,
should also appear in the main text of your paper at the appropriate places. You
can refer to them in the main text using first author last name and year
(Kafali, 2010). Write journal names in the references section, not just the
links.
Appendices:
Must include
–
Appendix A: Tables
–
Appendix B: Stata do-file (the do-file should be complete and include all Stata
commands that are used for the paper)
More information on the tables are as
follows:
Table 1: Description of ALL
variables in your model. You create this table manually and not in Stata. You should
explain all variables that you are using in your paper. For example:
Table
1: Description of Variables
Variable
Name
Description
Educ_yr
Number of years of
completed schooling
Female
Dummy variable
which is 1 if female and 0 if male
Married
Dummy variable
which is 1 if the respondent is currently married and 0 otherwise
(divorced/widowed/never married)
Table 2: Descriptive
statistics of ALL variables in your model. You should use the “outreg2” command to produce a
professional table with proper titles and labels. The summary statistics should
include the mean, standard deviation, minimum and maximum, number of observations
for all variables that you have in your paper. (Make sure n is the same for all
variables. If not, it means you have missing values you forgot to drop.)
Table 3: Descriptive
statistics according to certain characteristics. You should compute input for this
table using Stata. For example, you should show the differences in means of all
variables according to race (if race is your main x variable). For the first
row, you should write the mean education level for blacks and the mean
education level for whites and difference between the white and the black mean education
levels. You should also compare the means of blacks and whites using Stata and
report the p-values from Stata in the last column. This is a hypothesis test
comparing whether the two means are equal to each other (Ho: mblack=mwhite). You should use
the ttest command in Stata (two
sample t-test) to get the p-values. Refer to ppt slides “Hypothesis testing in
Stata” in Blackboard if you need help on using the command.
Table
3: Descriptive Statistics according to Race
Variable Name
Black
White
Difference
p-value
Educ_yr
Female
Married
Table 4: Regression
results. You should use the “outreg2”
command to produce a professional table with proper titles and labels. You must
show the appropriate variations of your model. For example, you can show different
functional forms (LPM/logit/probit) or (FE/RE) or different control variables. See
the table below where model 1 is the pooled OLS, model 2 is the Fixed Effects
model, model 3 is the Random Effects model.
For each model, make sure to report the estimated
coefficients and p-values.
Table
4: Regression Models
Variables
Model
1: Pooled OLS
Model
2: FE
Model
3:RE
Educ_yr
—
( )**
—
( )**
—
( )**
Educ_yr2
—
( )**
Female
—
( )**
—
( )*
—
( )**
Married
—
( )**
—
( )**
—
( )*
Year2020
—
( )**
—
( )**
—
( )**
Female*2020
—
( )**
R2
N
Table 4b: Additional
regression results if you have any that doesn’t fit in Table 4. The main
results of the paper should be in Table 4. If you are doing logit/probit this
table can present marginal effects.
Table 5: Additional hypotheses or specification tests. You must include any
joint F-test that are relevant; Hausman test if you are running FE/RE; comparison
of percent correctly predicted for if you are doing Logit/probit. For each
model, present the appropriate test results in a table along with any p-values.
Other guidelines on the paper:
* How do I start my paper?
· First check the dataset you are planning to use to see what kind
of variables you have access to.
· Think about any RELATIONSHIP that is interesting to you. You could
have read about it in the news or just wanting to think about it more. Many
students want to examine relationships that they have studied in other classes.
· Make sure that you read documentation about the main x and y
variables you are interested in. Some variables may be available for a limited
number of years.
· Think about what years or months of data you want to use. If you
want to do a diff-in-diff model using CPS data (repeated
cross-section) think about what years are relevant to compare for the
before and after. Once you decide on the years you want to compare, you can
download the same file for each year. (For instance, if you want to compare
2000 vs 2010 you can download the March file for each year.)
· If you want to use the panel data methods using World Bank data,
again think about what years you are interested in using (i.e. 2010 to 2020?)
and why.
· You can also use panel data methods with CPS data as well
but the panel is very short there (see data source file for more info). CPS has
panel data available to download on individuals following them for 2 years, so
that each individual has 2 observations recorded over time (for example: ASEC
2000 file you download will have 2 observations for each person, recorded in March
2000 and March 2001).
* When should I start my paper?
You should start your paper as soon as possible! A lot of
background work about finalizing your question can be done without having
knowledge of diff-in-diff or panel methods.
* Can I change my topic?
Some points will be deducted if the final paper is done on a topic
that is significantly different from the draft. In other words, start working
on your paper seriously from the beginning of the semester, so that you don’t
have to change your question repeatedly (changing your main x or y makes it a
completely new topic).
* Stata questions and help:
The basics you have learned on Stata in EC203-204 is enough for
you to be able to tackle the coding. If you need a refresher on those, the
Stata instructional videos are in the Econ department website: Stata Resources. If you have specific Stata questions on your code you need help
with, the TA is available to answer your questions during office hours. You can
of course come to my OH as well.
* Strictly abide by all the deadlines. No
late papers will be accepted.
* Writing guidelines:
–
You may create subheadings in
each section, if needed. For example, the results section can have two
subheadings: i) Returns to education on earnings and ii) Differences in returns
to education according to gender.
–
Include page numbers.
–
You can make use of footnotes.
For example, if you want to explain in detail what your hypotheses tests are,
do so as footnotes and not in the text. This will save you space and make the
paper more professional.
–
Important: every sentence that you write in your draft/paper
should be an original sentence written by you unless it is in quotes and
accompanied by a citation. No writing AI software such as ChatGPT allowed, and
it will be treated as plagiarism.
–
Do not include more than two or three direct quotes.
Changing a few words here and there isn’t paraphrasing, it is plagiarism.
–
An example of citation:
Much of the economic burden of mental health is due to reduced
wages (Marcotte, 2001)
–
An example of a quotation:
According to Kafali (2015) “presence of mental disorders reduces
the probability of being employed full-time by 10 percentage points” since this
is exactly what they said.
In both cases, you should include the associated papers in your
“References” section.
You should also include in your References the online sources such
as news articles.
* Final paper
grading: A rough rubric for
grading the final paper is as follows. Mostly, I am grading on the econometric
analyses you have done, accuracy of your explanations, and clarity in your
writing.
–
Econometric
analyses/Results (extensiveness, depth, and clarity in explanation of results):
~ 50 points
–
Introduction:
~10 points
–
Literature
Review: ~10 points
–
Data and
Descriptive Statistics: ~20 points
–
Format and
professionalism (general structure of the paper, reference list, formatting of
tables, page#s, etc): ~10 points
* Plagiarism:
I will be using a plagiarism detection software (Turnitin). If any paper is
found in violation, then you’ll receive grade of 0 and it will be reported.
