Exercise Using SPSS to Explore Measurement, Validity, and Relationships Among Variables
Ed
Nelson and Elizabeth Nelson, Department of Sociology
California State University, Fresno
Note to the instructor: The data set used in this exercise is RELG9800 which is a combination of the 1998 and 2000 General Social Surveys. (Some of the variables in the GSS have been recoded to make them easier to use and some new variables have been created.) This exercise uses RECODE and IF in SPSS to create new variables and CROSSTABS to explore the relationships among variables. In CROSSTABS, students are asked to use percentages, Chi Square, and an appropriate measure of association. A good reference on using SPSS is SPSS for Windows Version 9.0 A Basic Tutorial by Richard Shaffer, Edward Nelson, Nan Chico, John Korey, Elizabeth Nelson and Jim Ross. To order this book, call McGrawHill at 18003383987. The ISBN is 0072414456 . There is an online version of the book at SPSS Text . Validity is also discussed and students are asked to use the idea of construct validity to validate the measure they created. A good reference on validity is Reliability and Validity Assessment by Edward G. Carmines and Richard A. Zeller (Sage). You have permission to use this exercise and to revise it to fit your needs. Please send a copy of any revision to the authors.
Authors:
Ed Nelson and Elizabeth
Nelson, Department of Sociology, M/S SS107
California State University, Fresno
Fresno, CA 93740
Phone: 5592782275
Email: ednelson@csufresno.edu
elizn@csufresno.edu
Please contact the authors for additional information.
Goals of Exercise
The goal of this exercise is to create a measure of religiosity. We will also validate our measure. Validity refers to whether we are measuring what we think we are measuring. If we can show that we are measuring what we say we are measuring, that we have validated the measure. Once we have validated the measure, we’ll see how it is related to other variables.
Part I
We’re going to use the General Social Survey (GSS) for this exercise. The GSS is a national probability sample of adults in the United States conducted by the National Opinion Research Center. For this exercise we’re going to use a data set that combines the 1998 and 2000 surveys. Your instructor will tell you how to access this data set.
Religiosity is the strength of an individual’s attachment to his or her religious affiliation. Several questions on the GSS are possible indicants of religiosity. One of the questions asks respondents to estimate the strength of their religious affiliation. This variable in the GSS is called RELITEN. Respondents were also asked how often they attend religious services (ATTEND) and how often they pray (PRAY). These are all possible indicants of religiosity. Instead of choosing one, let’s combine all three variables into one composite variable.
Before you start, run FREQUENCIES in SPSS to get the frequency distributions for the following three variables: RELITEN, ATTEND, PRAY.
Let’s start by reducing the number of categories for each variable by using RECODE in SPSS. The variable RELITEN records the respondent’s selfreported strength of affiliation. The possible categories are strong (value 1), somewhat strong (2), not very strong (3), and no religion (4). Let’s combine somewhat strong, not very strong, and no religion into one category and give that category a value of 2. Now we have two categoriesstrong (1) and not strong (2). (Hint: When you use RECODE in SPSS, recode into the same variable. It’s easier. But if you make a mistake, you will not be able to go back and recode it again. You will have to close SPSS without saving the data set and then reopen the data set to get a fresh, clean copy of the data.)
Now let’s recode ATTEND. Let’s combine every week (value 7) and more than once a week (8) into one category and give this category a value of 1. Combine once a month (4), two to three times a month (5), and nearly every week (6) into another category and give this a value of 2. Finally, combine never (0), less than once a year (1), once a year (2), and several times a year (3) into another category and give this a value of 3. Now we have three categoriesoften (1), sometimes (2), and infrequently (3).
Finally, let’s recode PRAY. Combine several times a day (value 1) and once a day (2) into one category and give that a value of 1. Combine several times a week (3) and once a week (4) into another category and give that a value of 2. Combine less than once a week (5) and never (6) into another category and give that a value of 3. Now we have three categoriesoften (1), sometimes (2), and infrequently (3).
Now that you have recoded these variables, run FREQUENCIES in SPSS to get a frequency distribution for these three variables. Compare these distributions to the distributions you ran before you started to see if you made any mistakes. If you did, you will have to exit SPSS (or close your file) being sure NOT to save it. Then get back into SPSS and open the RELG9800 file again. The reason for this is that you have altered the coding of these three variables and will have to get another copy of the data file to start over. If you saved the data file, then you would have written over the original copy. So be careful.
Once you are sure that the variables have been recoded properly, add value labels for the recoded variables so the output will be easier to read.
Part II
Now that we have reduced the number of categories into a more manageable number, let’s create a new variable, which will be a combination of these three variables. We’ll call this new variable REL. To do this we’ll use the IF command in SPSS.
If an individual says he has a strong attachment to his religious affiliation (recoded value of 1 on RELITEN), attends church often (recoded value of 1 on ATTEND), and prays often (recoded value of 1 on PRAY), then he or she is highly religious. Let’s give these individuals a value of 1 on our new variable REL.
If an individual says he doesn’t have a strong attachment to his religious affiliation (recoded value of 2 on RELITEN), attends church infrequently (recoded value of 3 on ATTEND), and prays infrequently (recoded value of 3 on PRAY), then he or she is not religious. Let’s give these individuals a value of 3 on REL.
Everyone else will be somewhere between highly religious and not religious. Let’s give these individuals a value of 2 on REL.
Our new variable, REL, should have three categories1 represents those who are highly religious, 2 those who are medium in religiosity, and 3 those who are low in religiosity. If a respondent has a missing value for any of the three variables (RELITEN, ATTEND, PRAY), then he or she will automatically be assigned a system missing value for REL.
Run FREQUENCIES in SPSS to get a frequency distribution for your new variable, REL. There is another variable in the data set, RELIGOS, which should be identical to your variable, REL. Run FREQUENCIES for RELIGOS and compare the two distributions. If they are not the same, you made a mistake and will have to start over. See your instructor is you can’t figure out your mistake. Now add value labels for REL.
Part III
We have created a variable, REL, which we claim is a measure of religiosity. But how do we know it measures religiosity? This is a question of validity. Are we measuring what we say we are measuring?
What we can do is look for variables that are likely to be closely related to religiosity and see if they are strongly related. For example, if our measure is a valid measure of religiosity, then we would expect highly religious individuals to be more likely to believe in life after death than less religious individuals. The variable POSTLIFE tells us whether respondents say they believe in life after death. We would also expect highly religious respondents to be less likely to have seen an Xrated movie in the last year (variable is XMOVIE in the GSS).
If our new variable (REL) behaves as we expect it to, then we can claim that we have demonstrated its validity. This is called construct validity. If it does not behave as we expect it to, then it’s a little more complicated. It may be that our measure is not valid. Or it may be that our expectations are wrong. Or it may be there is something else wrong with our survey. But the important point is that if REL behaves as we expect it to, then we have evidence of the construct validity of our new measure.
Write a paragraph indicating whether you think your measure of religiosity, REL, is a valid measure. Indicate your reasoning.
Part IV
Now that we have created a measure of religiosity (REL) and have some evidence that it is valid, we could explore its relationship with other variables. It’s going to be up to you to choose the variable you want to use. Select one other variable that you think ought to be related to religiosity and complete the following steps:
2. Write a paragraph or two that indicates why you think your hypothesis is true. In other words, write an argument in which your hypothesis is the conclusion.
3. Use SPSS to run the crosstabulation of REL and your variable. Think about which is the independent and dependent variable. Remember to get the correct percentages. Use Chi Square and an appropriate measure of association.
4. Write a paragraph interpreting the table that SPSS gave you and indicate whether the data support your hypothesis. Use Chi Square and the measure of association to help you interpret the table.




