In statistics, data are classified into four types which determine how someone would interpret them and what tests can be performed on them. The four types of variables are nominal variables, ordinal variables, interval variables, and ratio variables.
**Nominal Variables**are variables for which there are only a few discrete values or "bins" into which the data fall. A person's sex, their favorite color, and the brand of their car would all be nominal variables. A necessary feature of a nominal variable is that any given data point lies in one and only one bin, that is, the bins cover the entire range of possibilities and they do not overlap.**Ordinal Variables**are made up of a series of bins into which the data fall. These bins are in a certain order, so that one could say that one bin is "greater than" or "less than" another. A person's health, if classified into several levels, would be an ordinal variable.**Interval Variables**are measured in discrete intervals from an arbitrary zero point. Since their scales are even, then, the Celsius and Farenheit temperature scales is an interval variable.**Ratio Variables**are measured continuously from an absolute zero point, so that there are no negatives, and each value can be said to be related by a ratio to any other.
The tests that can be run on a variable are determined by what kind of variable it is. There are two very general types of tests: parametric and non-parametric tests. Parametric tests deal with interval and ratio variables; most of the sites in DIG Stats are based on parametric tests. Non-parametric tests deal with nominal and ordinal variables; the only non-parametric test in DIG Stats is the chi-squared test. Original work on this document was done by Central Virginia Governor's School student Richard Barnes (Class of '00). Copyright © 1999 Central Virginia Governor's School, Lynchburg, VA |