What is an ordinal measurement scale and when is it used?

Answer 1

An ordinal level measurement is a scale that differentiates whatever is measured in order, but does not differentiate the size of the difference.

A scale that differentiates both order and the size of the difference is called an interval level measurement. In an interval level measurement the interval between two points has mathematical meaning that can show the size of the difference between two measures.

In the example below, the weight-rank variable shows which subject is heavier or lighter than the other, however the mathematical difference between the numbers in the Weight-Rank variable do not reflect the size of the weight difference between the subjects. Weight-rank is ordinal.

The Weight variable, on the other hand, does mathematically reflect the size of the difference (or interval) between subject. The Weight variable is therefore an interval measurement.

Weight-

Rank Weight

subject 1 1 120

subject 2 2 125

subject 3 3 150

subject4 4 225

SCALES AND INDIVIDUAL LIKERT SCALE QUESTIONS

It should be noted that individual likert style questions are considered ordinal. However, once you sum more than one likert style question, the resulting measurement is considered interval.

WHO CARES?

It is important to differentiate the level of measurement and ordinal and interval measures in particular because it effects the descriptive statistics and the statistical tests you can use on that measure.

Some Implications:

1. Ordinal level data measures central tendency with the median.
2. Interval level data measures central tendency with the mean.
3. Ordinal level data requires a non-parametric test to test for differences.
4. Interval data is "parametric"and meets the assumptions (or parameters) to test means with a t-test.
5. Interval level data is required to use Pearson correlation or ordinary least squares linear regression.
6. Ordinal level data uses a non-parametric version of the correlation coefficient to test association.