In SPSS, how do I analyze the similarity of multiple scores, differentiated by another variable?

Question

I could use some help figuring out how best to analyze my data with SPSS. I'm new to statistical analysis and SPSS, and trying to solve these issues is melting my brain. Sorry if this question is confusing or too nonspecific.

Some of the details of my research have been changed for privacy reasons. I created and distributed a survey to two schools--let's call them Alpha College and Beta University. I'm studying how important students think it is to follow rules with regards to plagiarism, drugs, and sex, and I've hypothesized that, due to a difference in campus culture, students at Alpha College are going to be more consistent in how they view the rules--that is to say, Alpha students are going to find the rules for all three subjects similarly important. On the contrary, I hypothesize that the campus culture at Beta University is a lot less salient when it comes to rules, and so I expect that Beta students will experience more variation in importance across the three subjects.

I measured "rule importance" with three different multiple-item Likert scales according to subject--so I have three variables, plagiarismImportanceScore, drugsImportanceScore, and sexImportanceScore. Each of these Likert scales contained a different number of items, so where a maximum score on plagiarismImportanceScore might be 25, the maximum on drugsImportanceScore might be 15 instead. I've also measured school-- a value of 1 means "Alpha College," and a value of 2 means "Beta University."

How, in SPSS, do I test how similar the scores for plagiarismImportanceScore, drugsImportanceScore, and sexImportanceScore are, differentiated between Alpha students and Beta students?

Ultimately, what I'm looking for is data that essentially says, "Alpha students have more consistent attitudes toward rules as a whole than Beta students."

Any and all help will be appreciated. Thank you.

Answer 1

Based upon your research design, I would hypothesize that one analysis you might want to look into is a logistic regression model. This model is typically conducted with a binary response variable, and predictors are incorporated to see how well they help differentiate between the response categories.

In your case, you are predicting school membership (Alpha | Beta), and trying to use your plagiarism, drugs, and gender scores to do so. A significant predictor in this model would tell you that, controlling for the other predictors, it accounts for a proportion of the variability in the log odds of attending one school or the other. You can further interpret the individual coefficients to understand what one school is doing more or less of.

To get you started, I would recommend looking at the logistic regression chapter in Tabachnick and Fidell's Using Multivariate Statistics, or (for an applied example using SPSS, without much of the mathematical background) Andy Field's Discovering Statistics with SPSS.