Chapter 5 of the SPSS book covered making controlled comparisons. From my experience with political science papers so far, control variables are a key component of the research design. So, it is nice to know how to deal with them in SPSS. SPSS taught the technical details of how to analyze control variables, based on the concept that in order to help make correct interpretations of controlled comparisons, the researcher should evaluate the relationship between the control variable and the dependent variable, controlling for the independent variable. Although I was aware of the basic concept of control variables before, I learned that, when making a controlled comparison, the researcher may discover that the relationship between the dependent variable and independent variable is spurious, additive, or interactive. I already knew what the term “spurious” meant, but I missed several questions on the worksheet before I really learned the difference between additive and interactive relationships. Thyne’s textbook helped with my understanding of these terms.
Chapter 6 of SPSS focused on making inferences about sample means. It discussed T Test procedures, which I had not studied since freshman-year statistics classes. I found that I was fairly easily able to recall what I had learned about the 95% confidence intervals and the P-value from freshman year. Because of this, I found the exercises to be fairly straightforward. Thyne’s textbook further explained the concept of statistical significance. Overall, this chapter was easy to grasp because of my previous experience with the concepts.
Chapter 7 of SPSS and chapter 9 of Thyne’s book delved deeper into cross-tabulation analysis. I enjoyed learning about the chi-square test, and how it compares the observed and expected values. However, I found the measures of association to be particularly interesting. I had not studied them in a statistics class before. For example, Somers’s d reports different measures of the strength of a relationship between two ordinal-level variables. I also found the concept of the proportional reduction in error (PRE) measure to be especially potentially useful for my own future analyses, since a PRE measure reports the extent to which the values of the independent variable predict the values of the dependent variable. The exercises required a little bit more time and focus this time. I still do not know from memory when I am supposed to use a measure such as Somers’s d versus when I am supposed to use a measure such as Cramer’s V, but I do know that it depends on the whether the dependent and independent variables are ordinal or nominal.
Also, our group submitted our theory, hypothesis, and definitions of key terms last week. I feel satisfied about all three of these parts so far. My only concern is that the theory is not fleshed out enough. I am fairly certain that we covered all of our terms and defined them sufficiently; we may have even defined more terms than were actually necessary.