Assignment 1: Hypothesis Construction (due Monday, September 17th)

This assignment will give you practice developing a hypothesis you can test for your final project. For the final project, you will do secondary analysis on the 2006 General Social Survey, generate a hypothesis about the relationship between three variables, and test that hypothesis by applying data analytic techniques. To complete the process, you will present the results of your data analysis in a final paper of 8 to 10 pages.

This assignment is the first step in this semester-long process. You will use information from the GSS website to collect data initially on two variables. The first step is to use your browser to go to the GSS website:

This website has data for the cumulative General Social Survey (1972-2006), although you'll use only the 2006 data. Peruse the web site. The list of variables is on the left when you link to the GSS site. You'll have to dig a bit to find all the available variables. For example, to get to "sex" or "race," you need to link to "Respondent's Background Variables," and then "Age, Gender, Race, and Ethnicity." When you click on the variables, it will show up in the "Selected" row at the top left of your screen. When you then click on "View," you will get the codebook information for that variable, and the frequencies for the entire GSS (from 1972 to 2006). Available for the first time this year, you can also use the Search link on the top of the web page. This should facilitate your ability to find interesting variables you can use. Use both these strategies to select two variables.

Example: Before you begin you might want to read examples of well argued, well written final papers by previous 311 students (example #1; example #2). These students earned well deserved As. The first student was interested in the relationship between sex (her independent variable) and attitudes toward extramarital affairs (her dependent variable). She wanted to explain variation in attitudes toward extramarital affairs, arguing that one's sex affects why some people are more tolerant of extramarital sex than others. The GSS has a variable named "xmarsex" that provides data on attitudes toward extramarital sex. Typing in "xmarsex" in the "Selected" row will give you frequencies for that variable, variable name, survey question, and the punch codes associated with each response. She chose "sex" for her independent variable. Typing in "sex" in the "Selected" row will give you the information for that variable. In paper example #2, the student was interested in the effect of strength of religious affiliation on attitudes toward premarital sex. Read through the paper to find details, and then try to find the variables in the GSS. Use these two papers as guides for your own work. Don't use these examples for your variables, choose two variables of your own.

Your turn: Now it's time for you to get started. Select two variables that you believe might be causally related. If you choose wisely, you'll be able to use these variables for the rest of the semester. Construct a theory and develop a hypothesis that describes how you think these two variables are related. Make your hypothesis as specific as possible. Include a short statement about how you came up with your theory and hypothesis. Why do you have the expectation you do?

Important point: please check to make sure that the variables you chose are available for 2006. In addition, check the 2006 frequencies to ensure that you have enough valid cases (at least n=1,800). You can check this by doing frequencies for your variables from the top right of the GSS site, "filtering" for "year(2006)". If there are no frequencies for 2006, or you have too few valid cases, you won't be able to use that variable for that year (it's okay to choose 2004 if necessary; check with us about earlier years). For example, if you check out "xmarsex" (put in "row") for the year 2006 (put "year(2006)" in "Selection Filter") you'll see that there are 1,966 cases in 2006. If you did the same thing for 2004 ("year(2004)") you'd find too few valid cases (only 868). In the first case, you could use xmarsex for 2006 (with an n=1,966), but not for 2004 (n=868). Again, look for a sample size of at least 1800.

Once you decide on the two variables you want, find the frequencies for each of your variables, and print them out. Keep this information for your records. Make sure you get the frequencies for 2006 (or 2004 or 2002, if that's the year you choose).

Print a copy of the codebook information (from "View") and the frequencies, and turn them in with your assignment.

You will have to think ahead and not wait until the last minute to do this assignment. Come in and see us early if you have any questions.

Once you have developed your testable hypothesis, follow it through the research process we talked about in class: describe your theory, the hypothesis you plan to test, your conceptualization of the two variables, and your operationalization of the variables (note: your operationalization will be the actual wording of the GSS questions).

Finally, speculate on which additional variables you might want to include in your future analyses to help you reformulate your theory.

Advice: as you are thinking through which variables to shoose, you probably want to avoid using an attitude as your independent variable. It's hard to explain variation in an attitude with another attitude. Talk to one of us for guidance if you choose to do so.

To help you along, feel free to consult the final paper examples I provided. But, avoid using their language in your writeup--write in your own words! Peruse their tables and graphs for future reference only. DO NOT make such a table for Assignment 1.

ALL ASSIGNMENTS MUST BE TYPED (12 point font, double spaced)!

[Don't forget to turn in the codebook description and a copy of the frequences for 2006 for your two variables.]