Nominal and Ordinal Categories

Why Some Categories Have an Order

Categorical variables describe which group or kind an individual belongs to. But categories differ in an important way: some have no natural order, while others can be ranked meaningfully. Knowing which kind you have helps you describe the data without adding meaning that the recorded values do not support.

For example, a student's major might be Biology, History, or Economics. These are different groups, but there is no general sense in which one major is higher or lower than another. By contrast, responses such as “very dissatisfied,” “dissatisfied,” “satisfied,” and “very satisfied” have a meaningful order from less satisfaction to more satisfaction. They remain categories, but their order conveys information.

The distinction is about the meaning of the categories in context, not their spelling, position in a table, or use of numbers as codes. An analyst can arrange nominal categories alphabetically, but that arrangement does not make them ordinal. Conversely, categories can be ordinal even when their names are words rather than numbers.

Nominal Categories: Different, Not Ranked

A nominal variable classifies each individual into one of several labels. Examples include academic major, blood group, type of transportation, and region of residence. The categories distinguish groups, but none naturally comes before or after another in a ranking.

A table may display nominal categories in any convenient order: alphabetically, by frequency, or in an order chosen for presentation. That is a practical arrangement, not a ranking carried by the variable itself. If a table lists Biology before Economics, it does not say that Biology is “less than” Economics.

Nominal categories can still be compared in useful ways. One can count how many individuals are in each category, compare category frequencies, or identify the most common category. Those comparisons concern how often categories occur, not an order or numerical distance among the category labels.

Ordinal Categories: A Meaningful Ranking

Ordinal categories have an order that is meaningful for the concept being recorded. Examples include levels of satisfaction, stages of a process, and health-severity levels such as mild, moderate, and severe. The categories can be ranked, but that ranking alone does not tell us how far apart adjacent categories are.

This limitation matters when interpreting the data. If one response is “very satisfied” and another is “satisfied,” the first is higher on the stated scale. But the labels alone do not establish that the difference between those responses is the same size as the difference between “satisfied” and “neither satisfied nor dissatisfied.” They also do not tell us that either difference has a numerical size.

The order of an ordinal variable should be stated in a way that matches its meaning. For satisfaction, more satisfied is higher. For symptom severity, more severe is higher, even though that direction may represent a worse condition. “Higher” refers to the chosen ranking, not necessarily to something more desirable.

Worked Classification: Health Severity

Why the Order Does Not Supply a Measurement Scale

The distinction can be justified by separating two questions. First, can the categories be ranked? Second, does the information supplied define numerical distances between them? Ordinal categories answer the first question yes and the second question no. Nominal categories do not supply a meaningful ranking in the first place.

This reasoning explains why an ordinal scale is not automatically a quantitative measurement. A rank tells us which category comes before another. A measurement with meaningful numerical distances would require additional information about how differences are defined. The category names alone do not provide that information.

A common way to encode categories is to assign numbers, such as 1 to “mild,” 2 to “moderate,” and 3 to “severe.” This can be convenient for storing or sorting responses. The codes preserve the order if assigned consistently, but the numbers do not prove that the gaps are equal. Likewise, assigning codes 1, 2, and 3 to Biology, Economics, and History would not make majors ordinal; the code values would be labels chosen for convenience.

A Reliable Classification Check

When deciding whether a categorical variable is nominal or ordinal, describe what the values mean and test the proposed order in context. A practical sequence is:

• Name the categories using their meanings, not just their codes.
• Ask whether one category can be meaningfully ranked above or below another for the variable being recorded.
• If no such ranking exists, classify the variable as nominal.
• If a meaningful ranking exists, classify it as ordinal, while avoiding claims about equal or known distances between levels.

Context can matter. A collection of words does not have one fixed type outside the variable they describe. The labels “low,” “medium,” and “high” are ordinal when they represent increasing levels of a characteristic. A list of department names, even if someone displays the names alphabetically, remains nominal. The question is always whether the recorded categories themselves have a meaningful order for the characteristic under study.

Common Mistakes

• Treating display order as a natural order. Sorting majors alphabetically does not rank one major above another. Alphabetical order is only an arrangement.
• Treating number codes as measurements. If 1 means “mild” and 3 means “severe,” the numbers label ordered categories; they do not establish that level 3 has three times the severity of level 1.
• Assuming ordinal steps are equal. A ranking tells us which category is higher, but not the size of the difference between adjacent categories.
• Calling every categorical variable nominal. Satisfaction and severity levels are categorical, but their levels also have meaningful order, so they are ordinal.
• Confusing higher rank with better outcome. “Higher severity” is a higher position in the ranking but usually describes a less favorable condition.