Statistical Modeling Final Paper/Project

Instructions for the second draft ...

Now that you have a topic and a data set, we need to continue to build the paper. Writing an empirical paper is a lot like baking a cake. You have to add all the ingredients, and there is a clear list of necessary ingredients for you to include. At this point, your paper should have a title page, an abstract (that you will amend as your paper matures), an introduction that states the basic subject matter and the problem you want to address, and a brief review of the literature. You should also now be in possession of a version of this that your instructor has corrected and on which comments have been added. The first thing you want to do is to address all of the corrections and comments. Then add new stuff.

Hypotheses:
Now you need to begin working with hypotheses. You need to begin thinking about how you think the world works, and why it works the way you think it does. Hypotheses are statements that explain why you think something in the world works in a certain way. Hypotheses explain phenomena. Hypotheses always come in pairs. Thus, there is always a null hypothesis that states that there is insufficient support for your ideas, and we should leave everything well enough alone. And then there is an alternate hypothesis that is your idea that explains why the world works the way it does. Some people like to list a bunch of hypotheses (always in pairs). They call these hypothesis #1, hypothesis #2, and so on, often abbreviating them as H1, H2, etc. But people who are more experienced with scientific writing tend not to list hypotheses in such a primitive fashion. They just explain their point of view, and then discuss their empirical results that either support or demolish this view. Quite honestly, listing the hypotheses is terribly boring to do, and such lists are awful to read. Moreover, the lists of hypotheses tend to reduce complicated scientific questions down to simple explanatory statements that often border on being trivial. Nonetheless, it is common for graduate students to list their hypotheses, and many faculty members like to see such lists in their students' work. Thus, we will write some of our hypotheses for this paper/project so that you can learn this aspect of the culture of scientific writing. But we will also try to talk about the subject intelligently so that the hypotheses do not appear either obvious or boring. There is one particularly useful rule of thumb: Include only high-impact ideas in your hypotheses. Leaving trivial stuff to text buried in paragraphs.

There are two approaches to hypothesis writing. The first is that one should state one's hypotheses before doing any research, and then do the research to confirm or reject these hypotheses. Frankly, while this is sometimes seen as the dominant view of how to do hypotheses, it really makes no sense. No scientist in his or her right mind will commit to hypotheses based on a level of knowledge of a subject matter that exists prior to doing any exploratory research. Nearly all initial ideas are thrown out after a scientist gets deeper into a subject. Thus, we have a second approach to hypothesis writing that is more realistic, in the sense that it more closely matches the way scientists really work. Make an initial stab at thinking about your subject. Pick a problem. Try to explain it with some initial ideas. Use your initial explanatory ideas to frame an initial set of hypotheses. Try to make your hypotheses nonobvious. That is, try to look at a commonly held view of why something works, and then question this with a new idea of your own. Your null hypothesis will then be the commonly held view, and your alternative hypothesis will be your competing idea. Then do your research. You will likely find out that your idea needs to be modified. You may discover that you were onto something, but that you forgot to include a crucial component. Thus, you will want to modify your alternate hypothesis, or perhaps develop a additional hypothesis. In each case, your null hypothesis will capture a view of the world in which your idea does not apply. This is the status quo in terms of existing knowledge. Your alternative hypothesis (if accepted) will throw out this null hypothesis, assuming your evidence is persuasive.

Thus, in your second draft, list some hypotheses that you want to test. Be sure to discuss the variables that you will use to test these hypotheses. You will need at least one dependent variable. (Try not to have more than two. If you have more than one, make sure that they are related in some way so that it is clear that you are addressing the same general subject with each variable.) You will also need a list of independent variables, and you can have many of these. Include at least one table that contains descriptive statistics of your dependent variable(s). Since you are trying to explain variation in a dependent variable, you need to show what variation exists. (Be sure that there is variation in your dependent variable or you will not be explaining anything!) You may also want to include some descriptive statistics for one or more of your most important independent variables. Again, be sure that you have identified the unit of analysis.

Research Design:
In this draft, you need to describe your research design. There should be a section of your paper titled "Research Design." Here is where you will describe your data set, your dependent variable, your independent variables, and your unit of analysis. You will also describe how you plan to test your hypotheses. Be sure to state how the various variables are coded so that readers know how to interpret your results. Also, here is where you put the descriptive information regarding your dependent and independent variables. All this takes 2-4 pages, and clarity is essential here. Explain whatever needs to be explained. Readers should be able to read this section and understand right away exactly what you are doing in terms of your research. You do not need to include any empirical results in this draft since you will do that in the next draft. This second draft is designed to get you set-up to dive into the results section as a next step.

The Most Important Two Ingredients:
In any presentation of empirical data, the two most important points that need to be addressed are (1) the unit of analysis, and (2) the dependent variable. Read this last sentence five times, and commit it to memory. When you write your paper, be sure that this information is very clearly stated near the front-end of the paper, preferably in the research design section.

Working with Tables and Figures:
There are some important formatting issues relating to tables and figures. Every table and figure needs to have an informative title. Do not simply say "Table 1." Add some description that tells the reader what the table or figure is about. For example, you might write something like

Table 1: Vote breakdown for Carter and Reagan with respect to partisanship

or

Figure 1: A jittered scatterplot for presidential vote with respect to respondent education

There are additional points for working with tables that are included in the instructions for the third draft of this paper. Be sure to note those as well when you are doing your results (for the next draft, not this one).

Every time you include a figure, do NOT put it directly in the text of your paper. Put each table and figure on a separate page. All empirical papers should be assembled in the following order. First stack all pages of text. Then include the tables, and then include the figures. Remember the order: text, tables, and figures. Commit this order to memory.

When you write about a table or figure, you need to tell the typesetter where to put it. Always write with the goal that your work will be published. Thus, prepare all of your writing with publication in mind. Follow this rule: After the first paragraph in which a table or figure is mentioned in the text, place an extra space (actually this is an extra double-space, since you are double-spacing your paper) and insert one of the following lines exactly as written.

(Insert table 1 about here.)

(Insert figure 1 about here.)

Do not change anything except the number of the figure or the table in the statements above. Following one of these statements, skip another space (again, a double-space) and then continue with your next paragraph.

Professionalizing the Format:
Formatting is important. Whenever you send something out to a scientific journal with the hope of getting it published, it will be peer-reviewed. If you follow a unique funky format, the reviewers will know you are an amateur and your work will likely be reviewed more skeptically ... which means rejected. If your work is professionally formatted and looks polished, then the reviewers will not know you are just beginning and will be more inclined to give you a fair reading.

Write clearly and forcefully. Memorize the punctuation chapter of Strunk and White's Elements of Style. Use semi-colons very sparingly, and only when you want to add a crucial accent to an important point.

Hand your draft in on time. This is important.

This entire site is Copyright © 1997-2008 by Courtney Brown. All Rights Reserved.
DISCLAIMER
URL: www.courtneybrown.com