Ordered outcomes

An ordered outcome is a categorical variable with three or more categories for which the order of categories is not treated as arbitrary.

For example, the most common way of asking survey respondents about their general health is to ask something like, “In general, would you say that your health is excellent, very good, good, fair, or poor?” Respondents are meant to understand these answers as constituting an order, where excellent > very good > good > fair > poor.

We could reverse the order of these categories – that part is arbitrary – but it would be totally confusing if we switched it up some other way, like asking respondents “In general, would you say that your health is excellent, fair, good, poor, or very good?”

In codebooks, numeric values are often assigned to ordered variables, and those will usually reflect the ordering of the categories as consecutive integers. For example, a codebook might list responses to this general health scale as: 1=excellent, 2=very good, 3=good, 4=fair, and 5=poor. When we treat an outcome as ordered, we consider the specific values used to assign numbers to categories as arbitrary so long as they preserve the category ordering. That is, numbering the set of categories [1, 5, 19, 37, 239] carries the same information as [1, 2, 3, 4, 5]. We do not, in other words, wish to treat the magnitude of the interval between the values as itself meaningful.