The final project will be an individual exercise in which you will choose a narrow research question with which you can use your data set to answer. The papers will be short, no more than three typed pages of text and one table having at most 4 columns, where each column will present the results of a regression equation. In a normal paper, there would be a literature review and a data section. But you don’t have space for a literature review and we have spent a lot of time with the data so no need for that. You can look up some related literature and cite it in your text if you like. The log files for all the empirical work must be appended to the paper, therefore if you use some scatterplots to determine the appropriate specification, they will be displayed in this file.
Your paper will start with a short introduction that explains your research question, why it is important, and then a quick description of the dataset you are using to answer it.
Next you will identify and justify the introductory equation you will be estimating: the empirical technique, the treatment of errors, the dependent variable, what independent variables are included, and why. Here you will need to identify which coefficient is especially important for the research question. Ideally you might identify a null hypothesis related to a coefficient which relates to the research question. If you did specification tests to determine this equation, you can describe them; their graphs will be in your log file. You can then describe the results from that equation as observed in the first column of your table.
Then you will discuss any extensions from this initial regression. You may decide that a second, third, or fourth equation is warranted –maybe adding other variables to reduce omitted bias, or doing an instrumental variable technique, or adding fixed effects, or estimating separately across groups. Describe these extensions and why they are important, and how they change the initial regression. Then present and discuss the results. I expect a paragraph for each equation (column of the table) would be warranted. Again in these discussions, you need to identify what coefficients are relevant to answer your question.
A final paragraph will sum up the results relative to the research question.
This must be your own work, so I prefer you not speak to other students beyond telling them what you are researching. I will check in with each of you over Thanksgiving week to make sure you have a topic and know how to get going.
Possible Questions
You may pick another question –just run it by me
Laboratory—section A
Does empathy lead to lower wages and is this effect different for men and women?
Does willingness to compete impact salaries and is this characteristic a potential reason why women earn lower salaries than men?
How does confidence affect salaries and is its effect different for men and women?
Are differences in risk aversion a potential reason why men and women earn different salaries?
Do the effects of confidence and willingness to compete interact to impact salary?
Does participation in college sports impact salaries (I can get this data for you.)
Is GPA a predictor of salary and how are its effects on salary different for different majors?
GLBT—session B
Is there evidence of employment discrimination against non-heterosexual individuals?
Among non-heterosexual workers, do different labor market outcomes arise between gay, lesbian, and bisexual workers?
Are the returns to human capital (experience and/or education) different for heterosexual and non- heterosexual individuals?
Are the returns to human capital (experience and/or education) different for men and women?
What are the determinants of decisions to tell boss about sexual orientation among non-heterosexual individuals?
Among non-heterosexual respondents, are those who are out at the workplace penalized (in terms of salary) relative to their peers?
Do the effects of having children on labor market outcomes vary for men and women?
