REVISION LESSON 4 ON SCALES OR LEVELS OF MEASUREMENT: NOMINAL || ORDINAL || INTERVAL || RATIO.
Based on RESEARCH METHODS CLASS WITH PROF. LYDIAH WAMBUGU's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.
Scales of measurement determine which statistical tools are appropriate because they define what the data values mean.
Briefing
Scales (or levels) of measurement determine what statistical tools can be used, because they define how variables were coded—either as categories or as numbers. The core idea is straightforward: before analyzing data, researchers must identify whether each variable is measured on a nominal, ordinal, interval, or ratio scale, since the scale dictates what kinds of comparisons and calculations make sense.
At the foundation is the nominal scale, the lowest level. Nominal measurement is for identification only—names or labels that sort observations into categories without implying quantity. Gender, eye color, skin color, race, and names are typical examples. Even if numbers are assigned (e.g., marital status coded as 1 = single, 2 = married, 3 = divorced), those numbers function only as tags. Swapping the codes would not change meaning, because there is no inherent order and no quantitative distance between categories.
Next comes the ordinal scale, which adds order. Ordinal variables rank observations from high to low or low to high, but the gaps between ranks are not equal. Educational level illustrates this: basic education, college/tertiary education, and university education can be ordered from lowest to highest (or reversed), yet the difference between adjacent categories is not guaranteed to be the same. The same logic applies to grouped age bands such as “below 19,” “20 to 25,” “26 to 30,” “31 to 35,” and “above 36.” These categories can be ordered, but the numerical spacing between them is not treated as equal.
The interval scale goes further by requiring equal intervals between values, while still lacking an absolute zero point. Temperature in degrees Celsius is the classic example: the difference between 10°C and 20°C matches the difference between 20°C and 30°C, but 0°C does not mean “no temperature” in an absolute sense. Attitude scales (like Likert-type responses) also fit this logic when coded so that the step from one response option to the next is treated as equal (e.g., strongly agree to agree to undecided to disagree to strongly disagree), even though there is no absolute “zero attitude.”
Finally, the ratio scale includes equal intervals and an absolute zero, meaning the zero point represents the true absence of the measured property. Distance and weight are the examples used most clearly: 0 kilometers means no distance, and 0 kilograms means no weight. Because ratio scales have a meaningful zero, they support the widest range of quantitative interpretations.
A practical takeaway ties the scales together: higher-level measurements can often be converted into lower-level ones (for example, age measured on a ratio scale can be grouped into ordinal categories like age bands), but converting a lower scale into a higher one is not valid. The lesson ends by emphasizing that getting the scale right is essential for choosing the correct analysis method.
Cornell Notes
Scales of measurement determine which statistical methods are appropriate because they define what the numbers (or categories) actually mean. Nominal variables are labels for identification only, such as gender or marital status; assigned numbers carry no quantitative value. Ordinal variables have a meaningful order (low to high) but unequal gaps between ranks, such as educational level or grouped age bands. Interval variables have equal spacing between values but no absolute zero, like Celsius temperature or coded attitude responses. Ratio variables include equal spacing and an absolute zero, such as distance and weight, where zero means the absence of the property. Higher scales can be converted into lower ones (e.g., age into age groups), but not the reverse.
Why do nominal variables treat numbers as “just labels,” and what does that imply for analysis?
What distinguishes ordinal measurement from nominal measurement?
How does interval measurement differ from ordinal measurement?
What does “absolute zero” mean in ratio measurement, and why does it matter?
Why is converting age from ratio to ordinal (age bands) acceptable, but converting ordinal back to ratio is not?
Review Questions
- Give one example each of nominal, ordinal, interval, and ratio measurement, and state the key rule that differentiates each.
- Explain why Celsius temperature is interval rather than ratio, using the meaning of zero.
- If educational level is coded as 1, 2, 3, what assumptions are valid (and invalid) about the differences between these codes?
Key Points
- 1
Scales of measurement determine which statistical tools are appropriate because they define what the data values mean.
- 2
Nominal variables are identification categories; numeric codes for them are labels with no quantitative value.
- 3
Ordinal variables have a rank order, but the gaps between ranks are not equal, so equal-distance calculations are not justified.
- 4
Interval variables have equal spacing between values but no absolute zero point, so zero does not mean absence of the property.
- 5
Ratio variables include an absolute zero and equal spacing, making them suitable for comparisons that rely on true absence (e.g., 0 distance, 0 weight).
- 6
Higher measurement levels can be converted into lower ones (e.g., age into age bands), but converting lower levels into higher ones is not valid because information is lost.