ABSTRACT
To what extent can we say that a variable is ordinal? In other words, what is the degree of ordinality of a variable? The variable age, for instance, may have categories that represent a discretized version of an underlying continuous distribution, while the variable family size has categories that truly reflect an underlying discrete distribution. On the other hand, we may have nebulous ordinal variables like the variable having categories: {strongly agree, disagree, …, strongly disagree} or a variable with categories: {social class 1, social class 2, …}. In the former, a definite ordering does exist even if respondents all have different conception of the locations of the relevant cutoff points. In the latter however, the existence of an underlying continuum for the purportedly ordinal variable is itself open to question.
