Introduction to Econometric

Empirical Project Proposal Instructions
The empirical project proposal is meant to help you synthesize what you have learned in class into a proposal
for a potential research project. You are not actually going to undertake this project (though with the right
question it could make a great prospectus for a senior thesis!). Be creative with your question. You may not
use questions we have talked about extensively in class – for example, you may not propose to look at the
impact of education or gender on wages, or the impact of infant mortality on international aid, or the impact
of smoking on baby birth weight. Have fun with this J I expect the length will be 2-4 pages, 12 pt font, double
spaced but shorter or longer proposals may be appropriate depending on your question.
It should be written up like a short paper – that is use full sentences and write in paragraphs.
Your empirical project proposal should be type-written, and include the following parts which cover the
following topics:

1. Research question
   Clear statement of a research question that estimates an association between two variables of interest (that
   can be measured!). Why is this question important and/or interesting?
   What is your dependent variable? What is your main independent variable of interest? What is your null and
   alternative hypothesis? What is the ideal experiment that would help you answer this research question
   causally?
2. Potential omitted variables
   Given that you cannot or will not be able to run this experiment, if you had observational data instead, what
   potential omitted variables might there be that could bias your estimate of the association between a change
   in the independent variable and a change in the dependent variable? Name at least one, if not more, potential
   omitted variables. Are these omitted variables potentially measurable? (i.e. IQ and height are measurable;
   work ethic may be harder to measure and you would have to describe how you can try)
3. Potential sources of data
   Given that you cannot or will not be able to run this experiment, what is the ideal observational data you
   would want to find to answer your question? Can you find actual data that includes both your independent
   variable and dependent variable, as well as at least one omitted variable, that you could use to answer this
   question?
   If so: describe it – What is the name of the dataset? Who collects it? how many people are in this dataset? Is
   the sample random? Who are they (i.e. a random sample of college students? US households? Likely voters?).
   You do not need to actually be able to access the data
   Here are some places to look for sources of data:

• http://libguides.rutgers.edu/c.php?g=336594&p=2271107
• http://dss.princeton.edu/cgi-bin/dataresources/guides.cgi
  If not: describe how you would undertake the ideal survey or data collection to answer your question of
  interest.

1. Estimating Equations
   Write the equation for the main (univariate) regression you would estimate (use insert->equation in Microsoft
   Word or google docs). Why did you choose that particular equation - Is the main relationship you want to
   estimate linear or non-linear? Why or why not? How would you interpret the estimate coefficient on your
   main independent variable of interest? (that, again, you aren’t actually estimating so you can use �" as a standin for the actual estimate).
   Given the potential OVB bias, write an equation that includes at least one of your easily measurable omitted
   variables. Now how would you interpret your estimated coefficient on the main independent variable (that,
   again, you aren’t actually estimating so you can use �" as a stand-in for the actual estimate). Are there still
   potential omitted variables that are hard to measure and thus will cause remaining OVB?
2. Dealing with remaining OVB?
   Are there any potential techniques you could use to deal with remaining OVB or other threats to
   identification? What are they? How might you implement them? If not, why do some of the techniques we
   have learned in class not apply in your case?
   Due Date: Wednesday May 1, 2019 – uploaded to Sakai by 1:00pm. Please save as .pdf and Name the File
   

Sample Solution