For example, the original responses to the question, "do you agree or disagree that abortion should be legalized in the US?" might be 1) strongly-agree, 2) agree, 3) don't know, 4) disagree and 4) strongly-disagree. For this class, we will combine categories so that 2 and only 2 categories are used in the final table. For example: 1) agree and 2) disagree. In general, the best thing to do with responses such as, don't-know, neutral, and uncertain is to simply eliminate individuals who answered in that manner from the analysis. For the most part, I have already done this for you - these responses already should have been converted to "system-missing" and SPSS should automatically eliminate them from analysis.

So, for this stage, combine the survey respondents' answers into 2 categories in some logical manner. For interval variables such as age, you might consider combining all categories below the mean or median into "young" and those at or above the mean or median into "old." You are not limited to this decision. Your hypothesis might deal with the differences between teenagers and adults. In that case, your "teenage" category should include those who are 18 and 19, while the "adults" should be everyone else.

For political party (if you are using this variable), you will have to do something about the "Independent" category as well as those who are "Undecided." You will also have to do something to combine the "Strongly Republican" and "Not so Strongly Republican" categories etc. The easiest thing to do in this case is to call your categories "Republican" and "Everyone Else" or similar. Here, you would place the "Strong Republican" and "Not so Strong Republican" in the same category and all other responses into the category "Everyone Else."  You can make similar decisions with other variables - use your own judgment and imagination. I will not deduct any points for any reasonable decision on how you separate your categories. Just make a written note of how you did this so that you can report these decisions when you write the methods section in an upcoming stage.

1. A frequency table for each of the 3 or 4 original variables that you will be using in your paper. The table should contain the original variable name.

2. A frequency table for each of the 3 or 4 recoded variables after combining categories. This table should contain the new variable name. The table should also not include any responses that you will be eliminating from your analysis.

Follow the instruction below to accomplish these tasks using SPSS software. If you are uncertain about how to open the data file, refer back to the instructions in stage 3.

1. Decide how to divide the responses for each of your variables into two categories (dichotomize).
2. After you have made your decision about how to divide the responses into 2 (and only 2) categories for each of the variables in all three of your hypotheses (the IVs, the DVs, and the Control variables) you will have to "recode" the responses and place the altered responses into another data-column (another variable). Remember that all of the responses are stored as numbers - e.g., "male" is coded as "1" and "female" is coded as "2."
3. Before going further, you should examine the spss output from the first spss assignment and write down the numerical coding for each of the response categories. You can do this from the bottom of the Data Editor window by selecting the "Variables View" tab. The window will change from a display of the GSS data to a display of details about each variable.
4. If you look under the column marked "Values" and click on the icon displaying three dots (...) a window will open that will show you the codes corresponding to each response category. Write these down for future reference. 
5. Next, from the Data Editor Window, go into the "Transform" menu and select "ReCode" and then "Into Different Variables." Again, a window will appear with your variables on the left. Let's start by recoding one variable at a time:
6. Next, highlight one of your variables that requires recoding (dichotomizing). Hit the arrow icon to place this variable into the text box in the middle of the window.
7. On the far right side of the window, there is a text box that is labeled, "Output Variable - Name." Type in any name that you will remember (caution - if you are using a version of spss older than version 12, it will only accept names less than 8 characters long). It may be wise to use the original name and add the letter "R" to the end to designate it as a "recoded" variable. For example, if the original IV is "sex" you could call the new/recoded variable "sexR."
8. If you wish, you can also include a descriptive "label" in the appropriate text box that will provide addition information in your output (experiment with this if you wish). Afterward hit the "CHANGE" icon.
9. Hit the icon that says, "old and new values." With your notes about the coding for the response categories from the step above at your side, type in the first old (existing) value on the left text box and a new value in the right text box. Then hit the "ADD" icon. Do this for each of the responses you wish to recode. For example, if I were using the "FeFam" variable that measured belief in traditional gender roles, I would see that the responses are listed from 0 to9. Responses 1 and 2 represent "strongly agree" and "agree" respectively. Because it makes sense to combine these into a single "Agree" category, I would first place "1" in the old value and a "0" in the new value and hit "ADD." I would then then place "2" in the old value and "0" in the new value and again hit "ADD." The window on the bottom will then indicate that any response marked as either 1 or 2 in my original variable will be "recoded" so that the new response is marked "0."  I would do something similar for the old categories 3 and 4 - recoding them so that they now equal "1."
10. Responses like "don't know," "undecided," "not applicable," etc. should have already been changed to "system-missing" so that they won't appear in our analysis.
11. You can also experiment with the options to include a range of responses to recode to a single new value - this is advantageous especially if you are using age as one of your variables. Figure this out on your own and ask me if you run into problems you can't solve. (Warning - this can be frustrating and confusing until you do this once or twice - don't worry, be patient it will eventually make sense and seem easy after a short time)
12. After you have specified the old and new values for one variable hit the "OK" button. In the "Data Editor" window a new column should appear with the new variable name that you specified above. Check things over to see if everything appears correct. It would be wise to go to the analyze menu and run "frequencies" on the recoded variable, compare with the "frequencies" on the original variable and check to make certain that your recodes are correct.
13. Repeat the above procedures for each of your IVs, DVs, and control variables so that they are all dichotomized. When you re-enter the window for each subsequent variable you will have to hit "reset" to erase the old information before entering the information for the next variable.
14. Run the frequencies command on all of the original variables as well as all of the recoded variables (refer to the instructions in prior stages). Look closely at the output to determine whether your work is correct.
15. Make certain to save a copy of your data set that has the recoded variables for future use.

You should save your output in a file by going into the file menu (on the output window) and selecting "save as." You should also save a copy of the dataset that has your recoded variables by going into the file menu (on the Data Editor Window) and selecting "save as." You can either over-write the existing dataset or create a new one. If you have room on your computer/removable-disk, I would recommend the latter. We will soon use these recoded variables during a subsequent stage.