What are Different Scales of Measurement? Nominal, Ordinal, Interval, and Ratio Scale with Examples

Measurement scales sort variables into four levels—nominal, ordinal, interval, and ratio—based on which mathematical properties their assigned numbers can legitimately support. The practical payoff is straightforward: the scale determines what kinds of statistical operations and tests are valid later, because higher scales allow more meaningful arithmetic. In short, identifying a variable’s scale is a prerequisite for choosing the right analysis.

Nominal scale variables carry only identity. Gender is the example: assigning numbers like 1 for “male” and 2 for “female” labels categories, but there is no meaningful order between them. Ordinal scale variables add “weight,” meaning the categories have a natural ranking. Position illustrates this: 1st, 2nd, and 3rd imply order, and swapping them (e.g., 2nd, 1st, 3rd) breaks the inherent ordering. However, ordinal variables still don’t guarantee equal spacing between ranks.

Interval scale variables include identity, weight (order), and equal interval spacing. Age in years is used to show this: differences between categories can be treated as equal when grouped into ranges like 21–30, 31–40, 41–50, and 51–60, each separated by a consistent difference (10 years). But interval variables lack “true zero.” Age cannot be zero in the same way that temperature can’t be treated as a true baseline for all purposes; the transcript’s key point is that without a meaningful zero, ratio-style comparisons (like “twice as much”) aren’t justified.

Ratio scale variables inherit all interval properties and add a final requirement: a true zero. Return on investment (ROI) is the ratio example. ROI can be grouped into ordered ranges (e.g., 1–10, 11–20, 21–30, 31–40), and the intervals are treated as equal. Crucially, ROI can reach zero when the investment is gone—unlike an arbitrary zero such as 0° Centigrade, which doesn’t represent an absolute absence of temperature. Because ratio variables have true zero, they support the full set of arithmetic operations (addition, subtraction, multiplication, division), enabling stronger quantitative comparisons than nominal, ordinal, or interval variables.

The transcript also lays out a method for classification. First, identify the variable’s possible values (e.g., gender: male/female). Second, assign numerical codes to represent those values, which establishes identity. Third, check whether the values have an inherent order (weight). Fourth, test whether the spacing between categories is equal (equal interval). Finally, determine whether the variable can take a true zero. Once the scale is known, analysis choices follow: ratio variables can support operations like division and multiplication and are more compatible with techniques such as multiple regression or independent-samples t-tests, while lower scales restrict what comparisons are statistically meaningful.