# Study

**Chapter Summary**

Each of the statistical techniques presented in this book made certain assumptions about the data. The correct measure of association depends on the level of measurement of the variable as well as the shape of the distribution. Similarly the choice of the proper measure of association depends on the level of measurement of your variables as well as the nature of the relationship. Those statistics which assume a linear relationship do not do a very good job describing non-linear relationships. In addition, regardless of the measure of association, your model needs to be correctly specified. Keep in mind that political factors tend to be interrelated. If you model a political process as being a simple interaction between your independent and dependent variables, you might not get an accurate measure of their relationship. If there is another variable that is highly connected to both variables, its effect on the dependent variable will get tied into the measure of the association between your two variables. In the end, your ability to use statistics to understand the political world depends on how well you connect your measures and statistical techniques to the political phenomena you are trying to describe.

**Learning Objectives**

After reading this chapter, you should:

- Understand how the way we measure political concepts affects our statistical analysis
- Understand that we can only draw conclusions at the level of the unit of analysis of the data
- Understand that the level of measurement of the data affects how the distribution of a variable can be described as well as its association with another variable
- Be able to describe the pattern of a relationship between variables
- Be able to identify the statistical significance of the relationship by using the correct statistic for the variables
- Be able to evaluate the substantive significance of the relationship by using the correct statistic for the variables
- Understand that statistics are probabilistic in nature—you cannot prove anything!
- Be more confident in your ability to use statistics honestly and evaluate the honesty of statistic you see in the news