Types of Data

Geographers have identified four types of measurement scales, or data types: NOMINAL, ORDINAL, MEASURE and COUNT. The distinction between these data types is based on the amount of information or the qualitative characteristics of the information carried by the data. Nominal data carry the least information, usually no more than a distinction between one datum and another, whereas Count data carry the greatest amount of information, including the amount of some variable measured on an absolute scale. These different data types are defined below.

NOMINAL (Names, Uniqueness)

Description: Nominal scales distinguish one item from another, but they do not rank or quantify data. Nominal values should be mutually exclusive, and can be exhaustive.

Examples: Soil Name, City Name, Polygon Identification Number

Test: A Chernozemic soil is not more or less than a Brunisolic soil; it is simply different.

PARS method: Nominal/ordinal interpolation

ORDINAL (Rank order)

Description: Ordinal scales identify the relative magnitudes, but they do not quantify exact differences between values. Ordinal scales include numeric data that has been grouped into classes.

Examples: Income = (low , medium , or high) Slope = ( A , B ); where A = 0-4%, and B = 5-9%

Test: A is less than B, but the difference between A and B is not strictly quantifiable.

PARS method: Nominal/ordinal interpolation

MEASURE (relative values)

Description: Measure scales include average values and densities.

Examples: mean snowfall, farms per square kilometer

Test: When a homogeneous polygon is subdivided, both halves assume the numeric value of the parent polygon.

PARS method: Measure interpolation

COUNT (total values)

Description: Count scales include absolute measurements (total counts) of continuous or discrete phenomena measured on an entire polygon basis.

Examples: hectares of woodland, number of farms.

Test: When a homogeneous polygon is subdivided, the numeric value of the parent must be divided into two parts.

PARS method: Count interpolation

Source: modified from Ballard and Schut, 1995