4 Levels of Measurement
Measurement may be classified into four different levels, based on the characteristics of order, distance and origin.
Table of Content
This level of measurement consists in assigning numerals or symbols to different categories of a variable. The example of male and female applicants to an MBA program mentioned earlier is an example of nominal measurement. The numerals or symbols are just labels and have no quantitative value.
The number of cases under each category is counted. Nominal measurement is therefore the simplest level of measurement. It does not have characteristics such as order, distance or arithmetic origin.
In this level of measurement, persons or objects are assigned numerals which indicate ranks with respect to one or more properties, either in ascending or descending order.
Example: Individuals may be ranked according to their “socio- economic class”, which is measured by a combination of income, education, occupation and wealth. The individual with the highest score might be assigned rank 1, the next highest rank 2, and so on, or vice versa.
The numbers in this level of measurement indicate only rank order and not equal distance or absolute quantities. This means that the distance between ranks 1 and 2 is not necessarily equal to the distance between ranks 2 and 3.
Ordinal scales may be constructed using rank order, rating and paired comparisons. Variables that lend themselves to ordinal measurement include preferences, ratings of organizations and economic status. Statistical techniques that are commonly used to analyze ordinal scale data are the median and rank order correlation coefficients.
This level of measurement is more powerful than the nominal and ordinal levels of measurement, since it has one additional characteristic, which is equality of distance. However, it does not have an origin or a true zero. This implies that it is not possible to multiply or divide the numbers on an interval scale.
Example: The Centigrade or Fahrenheit temperature gauge is an example of the interval level of measurement. A temperature of 50 degrees is exactly 10 degrees hotter than 40 degrees and 10 degrees cooler than 60 degrees.
Since interval scales are more powerful than nominal or ordinal scales, they also lend themselves to more powerful statistical techniques, such as standard deviation, product moment correlation and “t” tests and “F” tests of significance.
This is the highest level of measurement and is appropriate when measuring characteristics which have an absolute zero point. This level of measurement has all three characteristics order, distance and origin.
Examples: Height, weight, distance and area etc. are measured by natural numbers. Since there is a natural zero, it is possible to multiply and divide the numbers on a ratio scale. Apart from being able to use all the statistical techniques that are used with the nominal, ordinal and interval scales, techniques like the geometric mean and coefficient of variation may also be used.
The different levels of measurement and their characteristics may be summed up. In the table below:
|Levels of Measurement||Characteristics|
|Nominal||No order, distance or origin|
|Ordinal||Order, but no distance or origin|
|Interval||Both order and distance, but no origin|
|Ratio||Order, distance and origin|
Characteristics of Good Measurement
A good measurement tool must possess the following characteristics:
- Uni-dimensionality: This means that the measurement scale should not measure more than one characteristic at a time. For example, a scale should measure only length and not both length and temperature at the same time.
- Linearity: A good measurement scale should follow the straight line model.
- Validity: This means that a measurement scale should measure what it is supposed to measure.
- Reliability: This refers to consistency. The measurement scale should give consistent results.
- Accuracy and Precision: The measurement scale should give an accurate and precise measure of what is being measured.
- Simplicity: A measurement tool should not be very complicated or elaborate.
- Practicability: The measurement tool should be easy to understand and administer. There should be proper guidelines regarding its purpose and construction procedure, so that the results of a test can be interpreted easily.