Statistics - (Scales of measurement|Type of variables)

Thomas Bayes


According to Stevens theory (On the Theory of Scales of Measurement), the four scales of measurement are:

  • nominal
  • ordinal
  • interval
  • ratio

Each of these four levels refers to the relationship between the values of the variable.

It is also helpful to think of the scales of measurement as building in complexity, from the most basic (nominal) to the most complex (ratio).

Each higher level of measurement includes aspects of those before it.

Level Scale Application Basic Empirical Operations Statistics
1 Nominal Categorization Equality Number of cases , mode, Contingency correlation
2 Ordinal Order Greater or less median, percentile
3 Interval Distance Equality of difference or difference Mean, Standard deviation, Rank-order correlation, Product-moment correlation
4 Ratio Absolute zero Equality of Ratios Coefficient of variation

Table 1

Table 1 of Stevens theory (On the Theory of Scales of Measurement)

Level Of Measurement Table 1


Levels are type of measurement.


A Nominal Measurement is the type of measurement in which the values of the variable are names and not numerical at all. It' s used to assign individual cases to categories.

The Country of Origin is a nominal variable

More … Statistics - (Discrete | Nominal | Category | Reference | Taxonomy | Class | Enumerated | Factor | Qualitative | Constant ) Data


The Ordinal Measurement involves collecting information in which the order is significant and is used to rank order cases.


Interval measurements give:

  • distance between two values
  • and order (rank) between this distances


Ratio Measurement gets its name from the fact that meaningful fraction (or ratio) can be constructed with a ratio variable.

Documentation / Reference

Recommended Pages
Thomas Bayes
Data Mining - Multi-class (classification|problem)

Multiclass classification is used to predict: one of three or more possible outcomes and the likelihood of each one. Generally, there is no notion of closeness because the target class is nominal....
Data System Architecture
Dimensional Data Modeling - Measure

In a dimensional model, a measure is a quantitative attribute of a fact (in a fact table) that is not a foreign key that creates a relationship to a dimension. A measure permits to quantify. A calculated...
Data System Architecture
Ordinal Data

Ordinal data are data that can be ordered on an ordinal scale in order to rank them. Each ordinal data can therefore be translated into a number . This is a functionality/behavior of the data type...
Thomas Bayes
Statistics - (Fraction|Ratio|Percentage|Share) (Variable|Measurement)

Ratio measurement gets its name from the fact that meaningful fraction (ratio or rational number) can be constructed with a ratio variable. Ratio is the deepest, most meaningful level of measurement,...
Thomas Bayes
Statistics - (Interval|Delta) (Measurement)

An interval (measurement|variable) gives: distance (a range) between two values and order (rank) between this distances The following properties applied to interval measurement: The interval...
Thomas Bayes
Statistics - Measure

A measure is an dependent variable of an experimentation and have four level of measurement. See also:
Card Puncher Data Processing
Statistics - Relative

In the temperature scale, the temperature 0C at which water freezes is an arbitrarily chosen number and does not represent the absence of temperature. As a result, the ratio between temperatures is relative....
Card Puncher Data Processing
Statistics - Sample (Variable | Attribute | Feature)

A (statistician|data miner) studying a population would be interested in collecting information different characteristics of the subject (like their length, or weight, or age) in a sample. Those characteristics...

Share this page:
Follow us:
Task Runner