Read Edit History
Tags: statistics




Variable are descriptive elements; data is described through the use of variables. Each item (usually a record in database language) can be described by several variables (attributes in database language) These variables are categorized in four different scales: nominal, ordinal, interval and ratio.

Full text

Nominal variables are variables that describe categories without any implication of hierarchical relationships. For example, you could describe an archaeological feature as a pit, a ditch, a tomb or a wall.

Ordinal variables are variables that describe categories with a certain hierarchical relationship, though not defined in measurable properties. An ordinal scale would be village/town/city or bad visibility/medium visibility/good visibility.

Interval variables are variables that use a fixed numerically expressed distance to describe a hierarchical relationship, but do not have a meaningful fixed datum (or 0-point). A classic example in archaeology is chronology, where one year is a fixed measurement. This numerical interval allows relative comparisons, but does no allow calculations. E.g. you can say that between 600 BC, 400 BC and 200 BC each fall 200 years so there is a fixed distance, but you cannot say that 400 BC is twice as long ago as 200 BC.

Ratio variables are variables that represent a fixed numerically expressed distance with a fixed datum, which allows calculations. A ratio scale would be the thickness of sherds, the amounts of finds in a feature or the age of an artefact. In the age example you see the difference between interval and ratio; the dates are not necessarily bound to a fixed datum (the year 0 does not represent an ‘absence’ of the measured concept, the date), but stating that an artefact is 1000 or 500 years old does allow a you to say it is twice as old, using the present as a 0-point (which is the ‘absence’ of the measured concept, the age).

Research topics: Software & Technology