Levels of Measurement

Essentially, a measure's level of measurement is determined by the nature of it's categories.  When we develop an operational definition, what we are doing is attempting to categorize our observations based on some important observable trait.  The point is that we are making some kind of distinction between the cases in our study.

Level of measurement has to do with the nature of those distinctions.

There are
4 levels of measurement:
  1. Nominal
  2. Ordinal
  3. Interval
  4. Ratio

Determining the level of measurement of a given variable is a straightforward process of identifying certain characteristics of its categories.

Ask yourself these questions, in order, about your measure to determine the level of measurement:

  1. Do the categories of the variable have a rank order?
    That is, can you talk about something having "more" or "less" of the concept being measured?
    • ý No - The variable is a nominal level measure.
      The categories simply represent names of attributes, and any numerical
      values we assign to the categories are completely arbitrary.
      Example:   Race - 1) White   2) Black  3) Other
    • þ Yes - Go to Question #2

  1. Are the "distances" or "intervals" between the categories meaningful?
    In other words, do the numbers associated with the categories have any substantive meaning with the respect to the concept, other than simply showing rank order?
    • ý No - The variable is an ordinal level measure.
      The categories can be ordered from "low" to "high," and the numbers that we assign to them
      are arbitrary except to the extent to which they represent that rank order.
      Example:   Political Trust - 1) Low   2) Medium  3) High
    • þ Yes - Go to Question #3
  1. Does the scale have a meaningful zero?
    That is, does a zero on the measure mean the absence of the concept being measured?
    • ý No - The variable is an interval level measure.
      The categories only have meaningful intervals between them.
      Example:   Temperature - The numbers on the scale are meaningful in terms of their order and distance (i.e., you can say that it is 30º warmer in Florida than it is in Minnesota), but 0º F does not mean the absence of the concept being measured, heat.
    • þ Yes - The variable is a ratio level measure.
      You can construct a meaningful ratio, or fraction, with the values.
      Example:   Income - The numbers on the scale are meaningful in terms of their order and distance, and $0 on the scale means the absence of income.