A variable whose values are alphanumerical codes for different categories of observations or for different qualitative outcomes, which may or may not be ordered or ranked: for example, sex (male/female coded as M/F), preferred mode of transportation (cycle/bus/taxi coded as 1/2/3), opinions from a survey (say, ‘strongly opposed’, ‘opposed’, ‘neutral’, ‘support’, ‘strongly support’ coded as 0/1/2/3/4). When used in a regression they are recoded as a binary variable (often referred to as a dummy variable), with value 1 for a particular observation if it belongs to a given category and 0 if it does not.