Ordinal data is formally classified as qualitative (categorical) data, not quantitative. It falls into the broader category of categorical variables alongside nominal data, with one key distinction: ordinal categories have a meaningful rank or order. That ranking is what makes ordinal data feel numerical, and it’s why this question comes up so often.
Why Ordinal Data Is Categorical
Variables are generally split into two camps: qualitative (categorical) and quantitative (numerical). Categorical data sorts observations into groups. Quantitative data measures or counts something with actual numbers. Ordinal data sorts observations into groups that happen to have a logical sequence, but the “distances” between those groups aren’t equal or measurable. That’s what keeps it on the categorical side of the line.
Think about education level: elementary school graduate, high school graduate, some college, college graduate. These categories have a clear order from less to more education, but the gap between “elementary school graduate” and “high school graduate” isn’t the same measurable quantity as the gap between “some college” and “college graduate.” You can rank them, but you can’t do math with them the way you would with, say, years of education. The same logic applies to economic status categories like low, medium, and high, or a patient rating pain on a scale of mild, moderate, and severe.
What Makes It Different From Nominal Data
Nominal data is the other type of categorical variable, and it has no inherent order at all. Eye color (blue, brown, green) or blood type (A, B, AB, O) are nominal because rearranging the categories changes nothing about what the data means. With ordinal data, the order matters. “Strongly agree” ranks higher than “agree,” which ranks higher than “neutral.” Shuffle those labels and you lose information.
This is the defining feature of ordinal data: it preserves rank but not distance. You know that first place beat second place, and second beat third, but you don’t know by how much.
Why People Think It’s Quantitative
The confusion usually starts with numbered scales. A five-point Likert scale coded as 1 through 5 (strongly disagree to strongly agree) looks like numerical data. Researchers frequently assign numbers to ordinal categories for analysis, and in many studies those numbers get averaged, graphed, and run through statistical tests designed for quantitative data. This creates a reasonable impression that ordinal data is quantitative, but the numbers are stand-ins for categories, not true measurements.
The core problem is that the intervals between points on an ordinal scale aren’t guaranteed to be equal. The psychological distance between “strongly disagree” and “disagree” may not be the same as the distance between “neutral” and “agree.” Because those gaps are unequal and unmeasured, standard arithmetic operations like addition and subtraction aren’t technically valid. You can perform logical operations (greater than, less than) but not arithmetic ones.
The Likert Scale Debate
In practice, researchers break the “rules” constantly, and there’s genuine academic disagreement about whether that’s acceptable. Some statisticians argue that if your sample size is large enough (at least 5 to 10 observations per group) and the data are roughly normally distributed, you can safely use parametric tests like t-tests on Likert scale data. There’s compelling evidence that parametric tests tend to give the right answer even when their assumptions are violated to an extreme degree, making them more robust than the nonparametric alternatives that ordinal data technically requires.
That said, using means to describe ordinal data can be misleading unless the responses follow a normal distribution. A frequency distribution showing how many people chose each response is usually more informative than reporting that the average response was 3.4 on a 5-point scale.
So ordinal data lives in a gray zone. It’s categorically qualitative by definition, but in practice it often gets treated as quantitative when certain conditions are met. Whether that’s appropriate depends on who you ask and what you’re trying to do with the data.
Which Statistics Work for Ordinal Data
Because ordinal data has rank but not equal intervals, the appropriate summary statistics are the median (the middle-ranked value) and the mode (the most frequent value). The mean is technically off-limits because calculating an average assumes equal spacing between values.
For comparing groups, the standard toolkit includes nonparametric tests built around ranks rather than raw values. The Mann-Whitney U test compares two groups, the Wilcoxon signed-rank test handles paired data, and the Kruskal-Wallis test extends the comparison to three or more groups. For measuring association between two ordinal variables, Spearman’s rank correlation is the go-to choice. These tests strip the data down to its rank order and work from there, which respects the structure of ordinal measurement. Interestingly, when data does meet normality assumptions, the efficiency of these rank-based tests reaches about 95.5% compared to their parametric equivalents, meaning you lose very little statistical power by using the “correct” nonparametric approach.
Visualizing Ordinal Data
Bar charts are the standard choice for displaying ordinal data. Each category gets its own bar, and the bars are arranged in their natural order from lowest to highest (or vice versa). The height of each bar shows how many observations fall into that category. This works well because bar charts use alignment and length, which are easy for readers to interpret, and they show exact values clearly.
Histograms, by contrast, are designed for continuous quantitative data where adjacent bars touch to show a continuous range. Since ordinal categories are discrete groups with gaps between them, keeping the bars separated in a standard bar chart is more accurate. If you’re presenting survey results, a horizontal bar chart with categories listed from top to bottom in their ranked order is one of the clearest formats available.
Quick Comparison Across Measurement Scales
It helps to see where ordinal fits among all four levels of measurement:
- Nominal: Categories with no order. Examples: blood type, country of birth. You can only count frequencies.
- Ordinal: Categories with a meaningful rank but unequal intervals. Examples: pain severity (mild, moderate, severe), satisfaction ratings, education level. You can rank observations but not measure the distance between them.
- Interval: Numeric data with equal intervals but no true zero. Examples: temperature in Celsius, calendar years. You can measure differences but ratios don’t work (40°C is not “twice as hot” as 20°C).
- Ratio: Numeric data with equal intervals and a true zero. Examples: weight, height, income. All arithmetic operations apply.
Nominal and ordinal sit on the qualitative side. Interval and ratio sit on the quantitative side. Ordinal is the highest-ranked qualitative scale, which is exactly why it gets mistaken for quantitative data so often. It carries more information than nominal data but not enough to cross into true numerical measurement.