Spring 2019
Guidelines for Econometric Projects
Doing the work as a team
The work on these projects is to be done (in most cases) by teams of two students. The object of these projects is to learn to do econometrics. The reason that they are collaborative is to allow you to learn from (and with) your classmates.
Both partners are expected to contribute to all phases of the project: the econometric estimation tasks, the interpretation and analysis of results, and the writing of the report. Both partners are expected to contribute to each sub-part of the project: if there are two problems assigned, do not have one partner do one problem and the other partner do the other.
The best mode of collaboration is to meet at the beginning to decide on an econometric strategy for approaching each problem. Then perform the Stata tasks either together or with regular consultations to assess the results and determine the next steps that should be undertaken. Similarly, both partners should participate in the writing and editing of your reports and both partners must understand and be satisfied with the final result.
Contents of the report
The report that you submit should be in MS Word or pdf format and should be organized as a literate response to the questions, organized in sections and paragraphs that can be understood by someone who didn't just read the question. (In other words, don't just say "No, the coefficients are different.") Include your basic numerical results in your paragraphs along with the appropriate analysis and interpretation of them. You may put the key results into Word tables if you wish, but don't spend an excessive amount of time worrying about formatting as long as the content is clear.
The organization of the report on each problem/exercise should have the following form:
- Introduction. What problem/exercise are you working and what is it about?
- Data. What dataset are you using? Where did it come from? What are the descriptive sample statistics for each variable? (For problems from HGL, this information can be reproduced from the .def file.)
- Analysis sections. One or more topically organized sections that answer the questions posed in the multiple parts of the problem/exercise. You should identify which parts of the problem each section relates to, but the analysis should be self-contained as though it was part of a research paper.
- Conclusions and assessment of validity. A short summary of the main results of your analysis, followed by a validity assessment of the assumptions of the estimators that you used, a discussion of how potential internal validity problems would affect the estimates, suggestions for alternative estimators that might mitigate these problems, and a short discussion of external validity issues. If the problem asked you to use more than one model, discuss the advantages of the alternative models and which one you think is best.
These sections roughly mirror the organization of a typical empirical paper. There will be situations in which you do not have enough information about the dataset to make a valid assessment or where we have not yet studied alternative estimators. In these cases, use the information you have and explain what else you would need to know in order to improve your assessment.
Although you would never do this in a thesis or a published paper, please include the Stata output from which your results are generated before or after the results are discussed. This is helpful to me if your results don't agree with mine, so that I can attempt to determine what you did differently and which of us is correct. Stata text and table output can be copied and pasted into the report. The Stata tables are space aligned and will usually look best if you set them in a fixed-spaced typeface such as Courier, and you may need to use a small font to avoid wrapping lines. (I have created a Word style called "Stata" in my template that has 9-point bold Courier font and dark blue type. This makes it easier to format output by just setting the style to "Stata" before pasting the output into the file.)
Copying graphs from Stata to Word can be a little tricky. While you can use copy and paste, this is not reliable with some older versions of Mac Word due to a software bug that was never fixed. Although the graphs show up in Mac Word, they are not visible in other versions of Word, including mine. If you are using an older version of Mac Word (like, 10 years old) to prepare your document, please save the graph from Stata onto your hard drive, then import it into Word using Insert --> Picture. This takes a few seconds longer, but it will allow me to see your work. Alternatively, you can send your project report as a pdf, in which there are no such difficulties.
I have prepared a sample report covering Hill, Griffiths, and Lim's Exercise 2.12 on page 81. Use this sample as a guide for format and style in preparing your reports.
Submitting the report
Reports are to be sent electronically as an email attachment to the instructor at parker@reed.edu. (Note the absence of a first initial before or after the name. Please be careful because there are other Parkers at Reed who have a j before or after and your message will go to them if you get it wrong.) Project reports are generally due by midnight Monday nights.