Definitions
of Scales:
The
first empirical task is measurement. There are four measurement
scales which differ in the type of quantitative information they provide.
1. Nominal scales: In nominal scales, numbers are used as labels. They convey no quantitative information about an underlying attribute. Instead, they provide categorical information. All you can say about nomially scaled scores is whether or not they are the same (that is, belonging to the same category) or different (that is, belonging to different categories).
Example:
male = 0
female = 1
2. Ordinal Scales: In addition to saying whether scores are the same or different, numbers on an ordinal scale convey information about magnitude. You can say whether or not two scores are equivalent (i.e., tied) and in addition, you can rank order scores such that one score can represent a greater or a lesser amount of an attribute than another score.
First
place >>>>>>>>>>>
>>>>>>>>>>>>>
Last place
3. Interval Scales: In addition to providing information about equivalence and relative magnitude of an attribute, interval scales tell us how much greater or less one score is than another score. In interval scales, each unit of measure represents a constant amount of attribute. In other words, the SIZE OF THE INTERVALS BETWEEN SCORES IS CONSTANT AND MEANINGFUL!!!! Moreover, in interval scales, 0 does not mean the complete absence of an attribute.
![]() |
Note that zero degrees Celsius does not indicate a complete absence of heat |
** -15 degs Celsius is 45 degrees cooler than +30 degs Celsius**
4. Ratio Scales: Ratio scales are interval scales with an absolute zero point. Thus, in addition to providing information about the difference between two persons or things in the amount of an attribute they have, ratio scales allow us to make statements about the ratio of an attribute that one person or thing has relative to another. That is, with ratio scales statements such as 'one pencil’s length is twice that of another pencil' are meaningful.
PENCIL B
In other words,the zero point for a ratio scale, such as length measured in centimeters, is non arbitrary.
For most basic statistics, the distinction between interval and ratio level measurement is not that important .