Based on this chosen dataset, you will pick a research question and analyze the data to answer the research question.
- Think about your unit of analysis (UoA). (What are the observations in your dataset)? It might be people, or states, or countries. Defining the UoA can be tricky because it relates to the generalizability of the results from your analysis dataset.
a. For example, suppose you want to understand why people drop out of college. You may think that your UoA will be dropout students. But don’t forget that if you don’t also have data on non-dropouts you won’t know whether the characteristics you observe in dropouts are also found in non-dropouts.
b. The other problem is that you often can’t get individual-level data because of data availability and privacy issues etc. So, you may have to work with aggregated data; for example, dropout rates at the university level rather than the student level. So, in this case, your UoA would be universities, not individual students. Thus, the generalizability of your results should only really be applied to universities, not students. To generalize university level data to students is called an ecological fallacy.
c. You must have enough observations to ensure that the size of your sample (n) is large enough to generalize to the appropriate population. During the course, you will learn about why this is. For your sample size (number of observations), the following must be true: n-k-1>30. n=sample size, k=number of independent variables. The rule is that the larger your sample, the more it will reflect the population. - To get full credit, you need to do a multivariate test. This means you need to have two or more independent variables (IVs) and one dependent variable (DV). I encourage you not to have more than one DV. It makes the paper more difficult and forces you to do additional analyses (Read: More work.).
a. You can do a multivariate test with all categorical variables (a layered crosstab and chi-sq), multiple t-tests across samples and/or a multiple regression. - The complete final paper must be structured as follows:
a. Introduction: Present a short introduction to the issue you studied: What is the issue? Why is this issue important? What are the relationships you are specifically going to test?
b. Review of Relevant Literature: Conduct a brief literature review on the scholarly and applied research (i.e. GAO, CBO, academic journal etc.) that informs your question. Be succinct but informative. All the items that you cite should be relevant to the topic you are studying. (Do not go overboard please.)
c. Research Design: Identify the unit of analysis; the dependent and independent variables; and the hypotheses. Identify the source of the data and any weaknesses in the data. Then, identify the analytic technique that you will learn to analyze this relationship (i.e. t-test, ANOVA, chi-square, linear regression).
d. Results and Discussion: Present descriptive statistics (e.g., mean, standard deviation, minimum, and maximum for continuous variables and percentages for categorical variables, etc). Present and interpret the results of your statistical analyses (e.g., chi-square tests or regressions). You should interpret both the statistical significance of your results and their actual meaning. To get full credit you must do a multivariate analysis (i.e., with a dependent variable and at least two independent variables).
e. Conclusion: Discuss the real-world implications of your results (i.e. offer policy recommendations (if relevant). Also, you should address the limitations of your analysis. - The report should be no more than 10 pages long (double spaced), including tables and charts. You are essentially learning how to write for policymakers. Don’t get carried away. You must create your own tables and charts for descriptive statistics and test results rather than just using STATA output. Also be sure to look at the grading rubric for the assignment (available to view in toward the middle of the semester). Use APA citation style.
- A note about writing. Do not use 15 words when you can use 9 words. Your writing is expected to be coherent and organized with topic sentences and subheadings. If you are struggling with essay and paper writing, I encourage you to contact the writing center, GMU SPGIA Student Services or access other learning tools are available to GMU students (http://ctfe.gmu.edu/teaching/student-support-resources-on-campus/).
- In completing this paper please consult the Data Analysis Paper Rubric.
1 M.A. and Ph.D. students may choose to follow this format. Or, they may do a more academically oriented paper aimed at (eventually) publication in a scholarly journal. They should clear this with me before taking this route.