POLS 2140 Introduction to Political Analysis
Prof. Hansen
FINAL PROJECT
(My expectation is about 3 pages of text, plus tables, output, graphs, etc., from your analysis.)
You are going to use the 2016 ANES data set and codebook to write a mini research paper that incorporates all the pieces of the research process or scientific method that we have covered in this course. Refer back to Lab #4 for details on recoding variables, etc.
PLEASE USE POST ELECTION QUESTIONS; MAKE SURE TO RENAME, RECODE, AND LABEL YOUR VARIABLES!
Paper must include the following:
Develop a research question that you are interested in exploring or learning more about.
Look for literature in political science journals or other related disciplines that addresses the issue you are exploring and briefly discuss what others have written on the issue or how it informs your theory.
Propose an explanation or theory for what you are studying using political science or other disciplines. Why do you expect your independent variable(s) to cause your dependent variable? You want your theory to be focused on your MAIN independent variable(s) and its relationship to the dependent variable.
(Remember Poe and Tate. Their main argument was that the Type of Government causes human rights violations; particularly, they argued that democracies should engage in fewer human rights violations.)
Formulate specific hypotheses that you can test. For this paper, write out the null and alternative hypotheses for each of your independent variables. (e.g. Ho: How Democratic a government is has no effect on level of human rights violations vs. H1: The more democratic a country is the fewer human rights violations they are likely to commit. Ho: War has no effect of human rights violations. H1: Countries engaged in civil war are likely to commit more human rights violations. etc.)
Describe the data and specific variables that you will be using to test your hypotheses. You need one DV and at least three IVs for your paper. You can have more IVs. Poe and Tate have about 10!
In an appendix, provide basic descriptive statistics (frequencies, means, std dev, histograms, etc) for each of your variables. You must include frequencies for all nominal or ordinal measures and means/std dev/variance for all continuous measures for all of your variables. Put this in an appendix.
First, explore and provide analysis of bivariate relationships between your variables, in particularly between your DV and each IV, using a method appropriate to the variables (correlations, cross tabs, difference in means tests, etc).
Estimate a regression model. Describe the regression equation (model) that you will use to predict your DV using at least three IVs. This will be a multivariate model where all of your IVs will be in ONE model to predict your DV. You can write out the model in an equation like Poe and Tate do.
After your data analysis, explain all of your findings.
Turn in ALL of your R commands and output for the analysis—R commands should be in an appendix.
Offer some conclusions based on your findings.
Below I provide some general guidelines for structuring a research paper. Even though this is just a mini paper, I thought these guidelines might be helpful here or in other classes that you take.
General Paper Structure and Tips
The paper has 5 major parts (but also subparts/subheadings within them, organization matters):
a) Title and Abstract
b) Introduction & literature review & Theory
c) Data and Methods
d) Results
e) Discussion/Conclusion
Title. Give your paper a title that hints at what it is about.
Abstract. Always include an abstract of 150 words or less. In the abstract, answer these questions and discuss results:
What question/puzzle/problem are you trying to solve?
How did you go about achieving your objective?
Discuss main results.
What are the implications?
Introduction/Literature/Theory. The introduction should explain your hypothesis and introduce your theory and discuss the literature. Your theory should justify why you think that you will find an association between your main independent and dependent variable. You should define your dependent and independent variables and explain why your causal relationship is theoretically justified. Your hypotheses should go here as well.
Data and Methods. This section should describe the data and the method(s) you are using to answer your hypotheses. You also want to include a description of your variables at this point.
Results. The results section will, of course, present your results. You want to start simple with some frequencies, crosstabs and t-tests, correlations. These things depend on your hypothesis and the variables you have chosen. Your results do not have to lead to a confirmation of your hypothesis (actually a rejection of the null hypothesis). They may lead to a rejection of your hypothesis (an acceptance or failure to reject the null hypothesis). Finally, the main part of your paper is your regression model. Be sure to explain the goodness of fit and the coefficients of your IVs. What is the size and direction of the impact of each of your independent variables on the dependent variable.
Discussion/Conclusion. Summarize your findings or non- findings. Possibly you may want to suggest future survey questions or future research that will address your theory better.
Other tips
Put your name on the paper!
Always include page numbers.
Always include a reference section.
Section your manuscript with headings and sub-headings where appropriate. These help readers keep track of where you are headed.
Be sure to include a careful discussion of your theory and hypotheses and a careful discussion of your data and methods. Of course, the other parts as well!
Always include Tables that are self-explanatory and can stand alone. Tables should be numbered and have titles and anyone should be able to read the table without reading the paper. Look at any political science article for examples, like Poe and Tate article.
Graphs are good too, if they show something interesting. These won’t work with ordinal and nominal data.
Never use abbreviated variable names from your data set; use names that anyone can understand.
For example, suppose I am interested in studying people’s attitudes towards international trade.
I choose this variable as my DV:
table V162176x
POST: SUMMARY- Favor/oppose free trade |
agreements | Freq. Percent Cum.
—————————————-+———————————–
-9. Refused | 21 0.49 0.49
-8. Don’t know | 33 0.77 1.26
-7. No post data, deleted due to incomp | 86 2.01 3.28
-6. No post-election interview | 536 12.55 15.83
1. Favor a great deal | 354 8.29 24.12
2. Favor moderately | 821 19.22 43.34
3. Favor a little | 245 5.74 49.08
4. Neither favor nor oppose | 1,449 33.93 83.00
5. Oppose a little | 138 3.23 86.23
6. Oppose moderately | 340 7.96 94.19
7. Oppose a great deal | 248 5.81 100.00
—————————————-+———————————–
Total | 4,271 100.00
I would have to create a new variable from this one, which I will call Free Trade Agreement, excluding options like “Don’t know” or “Refused”, etc. For this variable, I would only want to keep categories 1 through 7 in the new variable.
Next, I would want to choose a focal variable(s) that I am most interested in using to explain attitudes towards free trade. For example, it could be an IV that measures how people feel about the economy. How the economy is doing might affect how people feel about trade agreements because if they believe the economy is doing poorly, they might think we should stop allowing free trade agreements and focus on our domestic economy, not the world. Alternatively, maybe I believe people’s attitudes about free trade are shaped by how they feel about foreign policy or our relationships with other countries.
Once you choose a focal IV, you need to think about what other factors you need to control for in order to be more certain that your focal variable is really causing the relationship that you are observing. For example, there is a load of political science literature that says that someone’s party ID affects attitudes towards trade. So I might want to include party ID (Dem, Rep, Indep) as one of my independent factors; I might incorporate this with TWO dummy variables (remember the number of dummies is the number of categories -1), choosing one of the categories as my base category. So I could have a dummy variable for Dems and a dummy variable for Indep, and leave Reps out as the base of comparison. How do Dems feel relative to Reps. How do Independents feel relative to Reps. Maybe instead of partyID, I use a measure of ideology. What else might you have to control for in predicting attitudes towards trade? For example, maybe level of education matters, etc. You need to choose at least 3 IVs for your regression model. Once selected, I might start with a cross tabulation table to look at the relationship between party ID and trade using the chi-squared test of significance. Finally, I would use a regression model for the following equation:
Attitude towards trade agreements
= b1 + b2*Democrat + b3*Independent + b4*education + b5*economy + e
And look at the impact of each IVs on my DV.
Questions? You can email me anytime. I will try to check my email regularly. As most of you know, I usually get back to you pretty quickly. If you don’t hear back from me, you may try Bibek. He will have access to this assignment as well, and is familiar with the ANES data and can help with R questions, probably better than I can!
If you want feedback on the writeup, you can put the paper on google docs and share it with me; I will try to give you feedback, time permitting. I can also zoom if you want.