Source materials for the following questions will be found in Adventures in Criminal Justice Research. Dowdall, Logio, Babbie, and Halley. At p. 31.

Click on the BACK button on your Browser to go back to the Pass or Prepared? on Nominal, Ordinal, Interval.

1. With what term should we associate "nominal" variables?

   One Plausible Answer

   Nominal. Naming. With nominal data we operate on the simplistic theory that the categories are mutually exclusive. That is, if you are a Republican, you may not be a Democrat. If you are male, you may not be female. If you belong to the category Protestant, you may not also belong to the category Catholic, or Jewish, or Other.

   Notice that the term Other covers a multitude of sins. Since Other covers everyone not in the other three categories of Protestant, Catholic, Jewish, that means your categories cover everyone. They are exhaustive.

   "Mutually exclusive" and "exhaustive" sound just wonderful. I should like, however, to respectfully suggest that the underlying theory from which you are operating may leave much to be desired. For example, if you want to know how people at a given college connect to the Internet, your categories might be the following:

   • DSL line
   • Modem
   • Other

   But that set of categories makes the unstated assumption that everyone at the college does connect to the Internet. Quite an assumption. Beware of letting your perspective get in your way when you choose categories.

   What we can do statistically with nominal data is look at the frequency of each category. We can have a modal category, the category with the most people. We can do bar graphs or charts to present a visual sense of how the variable is distributed.

2. With what term should we associate "ordinal" variables?

   One Plausible Answer

   Order. Ranking. Any variable that we are willing to measure on a scale of high, medium low, strongly agree to strongly disagreee, or any other scale for which we can rank the values is measured ordinally.

   Why does that matter so much? Well, in class we spoke of the importance of measures of association. Measures that give us an estimate of how much better we can rank one variable, like IQ, if we know the rank on another variable, like achievement test scores. Or how much better we can rank achievement categories if we know IQ categories.

   . . .

3. With what term should we associate "interval" variables?

   One Plausible Answer

   Interval. Equal intervals. Measurement like number of units at school, number of years of school, age in years, income in thousands of dollars, etc.

   . . .

---

---