Key Postulates of Measurement: Foundational Concepts in Research

Every research project, from a household income survey to an impact evaluation of a government scheme, begins with a deceptively simple act: assigning numbers or labels to things we observe. But measurement is not just about reading a scale or ticking a box. It rests on a set of logical rules that decide whether the numbers we assign actually mean anything. These rules are called the postulates of measurement, and they form the silent foundation beneath every dataset, questionnaire, and statistical test you will ever encounter. Understanding them helps you see why a “5 out of 10” rating or a ranking of cities by liveability is meaningful in the first place.

What measurement really means 🔗

In research, measurement is the process of assigning numbers or symbols to the characteristics of objects, people, or events according to a fixed set of rules. The key word is rules. Without consistent rules, the numbers we collect would be arbitrary and impossible to compare. Measurement turns vague observations like “this neighbourhood feels crowded” into structured data that can be analysed, compared, and acted upon.

The postulates of measurement are the logical assumptions that make this assignment of numbers valid. They tell us what we are allowed to say about objects once we have measured them. According to the foundational treatment of measurement in IGNOU’s research methodology material, these postulates govern when two things can be treated as equal, how equality carries across several objects, and how we establish that one object has more of a characteristic than another. Three postulates do most of this work: equality, transitivity, and the ordinal postulate.

The postulate of equality 🔗

The first and most basic rule is the postulate of equality. It states that for any two objects, we must be able to say either that they are the same on a given characteristic, or that they are not the same. In formal terms, for two objects a and b, either (a = b) or (a ≠ b), but never both at once.

This sounds obvious, but it carries real weight. Measurement begins by sorting objects into categories where members are treated as equivalent. When a census enumerator records a person’s religion, mother tongue, or occupation category, they are applying the postulate of equality. Two people recorded under the same category are being declared “equal” on that characteristic, while a person in a different category is “not equal.”

Why equality comes first 🔗

Equality is the entry point to all measurement because every higher level of measurement still depends on it. Before you can rank cities or measure distances between values, you must first be able to decide whether two things belong in the same bucket. This is exactly the logic behind the nominal scale, the simplest level of measurement, where variables are merely named or labelled without any order. As described in a widely used explainer on levels of measurement, a nominal scale labels variables into distinct classifications without involving any quantitative value or sense of “more” and “less.”

Consider how the Census of India classifies workers as main workers, marginal workers, or non-workers. These are categories built on the postulate of equality. Two main workers are treated as equal on the employment-status characteristic, even though their actual jobs may differ enormously. The postulate does not claim they are identical in every way, only that they share the specific characteristic being measured.

The condition of mutual exclusivity 🔗

A crucial part of the equality postulate is that an object cannot be both equal and unequal to another on the same characteristic at the same time. This forces categories to be mutually exclusive. A respondent cannot simultaneously be classified as urban and rural, or literate and illiterate, within a single variable. If your categories overlap, the postulate of equality is violated, and any analysis built on that variable becomes unreliable. This is why good questionnaire design insists on response options that do not overlap and that together cover every possibility.

Transitivity in measurement 🔗

The second postulate extends equality across more than two objects. Transitivity states that if object a equals object b, and object b equals object c, then object a must also equal object c. Symbolically, if (a = b) and (b = c), then (a = c).

This postulate is what allows a researcher to build comparable sets. Without transitivity, equality would be a fragile, pairwise affair that collapsed the moment you added a third object. With it, you can confidently group large numbers of objects together knowing that everyone inside the set shares the characteristic equally. As the IGNOU material notes, transitivity enables a researcher to establish the quality of set members on a characteristic by comparing them through a common reference point.

How transitivity builds reliable categories 🔗

Imagine a survey team coding the education level of thousands of respondents. If person A and person B are both classified as “graduate,” and person B and person C are both classified as “graduate,” transitivity guarantees that A and C also belong together. This is what makes it possible to count, tabulate, and compare groups. Every frequency table and cross-tabulation you have ever seen quietly relies on transitivity holding true.

When transitivity breaks down 🔗

Transitivity is not just an abstract assumption; researchers actively test whether it holds, because real-world data can violate it. In advanced methods such as network meta-analysis, where treatments are compared indirectly through common comparators, the transitivity assumption is treated as a cornerstone of valid conclusions. Methodologists have shown that intransitivity can be surprisingly common when the groups being compared differ in important background characteristics. When that happens, indirect comparisons become biased and the results can mislead. The lesson for any researcher is that transitivity must be earned through careful design, not simply assumed.

Experimental research has even tested whether human judgements obey transitivity. Studies on decision-making have found that, while occasional violations appear, the transitivity postulate generally holds up when measurement is done carefully, lending support to its use as a foundational rule.

The ordinal postulate 🔗

The first two postulates deal only with sameness and difference. The ordinal postulate introduces the idea of order. It states that objects can be ranked according to how much of a characteristic they possess. If object a has more of a characteristic than b, and b has more than c, then a must have more than c. Symbolically, if (a > b) and (b > c), then (a > c).

This is transitivity applied to inequality rather than equality. It lets us move beyond saying “these are the same or different” to saying “this one ranks higher than that one.” The ordinal postulate is the logical basis for the ordinal scale, where data can be categorised and ranked, but the exact distance between ranks is not known. A widely cited guide on levels of measurement explains that ordinal data can be ordered, yet the gaps between adjacent ranks cannot be measured precisely.

Ranking without equal distances 🔗

The defining limitation of ordinal measurement is that the intervals between ranks are not necessarily equal. Think of a customer-satisfaction question with options ranging from “very dissatisfied” to “very satisfied.” You know that “satisfied” outranks “neutral,” but you cannot claim the jump from “neutral” to “satisfied” is the same size as the jump from “satisfied” to “very satisfied.” The popular Likert scale used in countless social surveys is a classic ordinal instrument for exactly this reason.

Everyday examples in India are easy to find. The order of finish in a marathon, the ranking of states in NITI Aayog indices, or the seniority order in a civil service cadre are all ordinal. They tell you who is ahead and who is behind, but not by how much. As a statistics educator notes, the four scales of measurement build on one another, and recognising which scale your data uses helps you avoid analysis mistakes such as calculating a meaningless average of ranks.

Why the ordinal postulate matters for analysis 🔗

The scale of measurement directly limits which statistics you may legitimately compute. Ordinal data supports the median and percentiles but not the arithmetic mean, because the unequal intervals make addition and averaging unjustified. Choosing the wrong statistic, such as averaging a set of satisfaction ranks as though they were temperatures, produces numbers that look precise but mean nothing. Respecting the ordinal postulate keeps your conclusions honest.

How the postulates fit together 🔗

These three postulates form a logical ladder. The postulate of equality lets you sort objects into categories. Transitivity lets you trust that those categories hold consistently across many objects. The ordinal postulate lets you arrange the categories in a meaningful order. Each rung supports the one above it, which is why the higher levels of measurement, interval and ratio, still carry all the properties of the lower ones.

This layered structure traces back to the framework developed by psychologist Stanley Smith Stevens in 1946, who argued that all scientific measurement falls into four levels, with each level building on the properties of the previous one. The interval scale adds equal spacing between points, and the ratio scale adds a true zero that makes statements like “twice as much” valid. But none of these higher capabilities would mean anything without the foundational postulates of equality, transitivity, and order sitting underneath.

For anyone designing a study, the practical takeaway is clear. Before collecting a single observation, ask which postulates your variable satisfies. Are your categories mutually exclusive? Do they hold consistently across all respondents? Can they be ranked, and if so, are the distances between ranks equal or not? The answers determine the scale of measurement, and the scale determines which analyses will yield trustworthy results. Getting this right at the design stage prevents costly errors later, when flawed measurement quietly corrupts every conclusion that follows.

What do you think? When you next see a ranking of cities, schemes, or institutions in the news, can you identify whether it respects the ordinal postulate, or whether it wrongly treats unequal gaps as if they were equal? And in a survey you might design yourself, how would you make sure your response categories satisfy the postulate of equality?

References

1. https://egyankosh.ac.in/bitstream/123456789/39230/1/Unit-1.pdf
2. https://www.questionpro.com/blog/nominal-ordinal-interval-ratio/
3. https://link.springer.com/article/10.1186/s12874-024-02436-7
4. https://www.sciencedirect.com/science/article/abs/pii/0030507377900666
5. https://researcher.life/blog/article/levels-of-measurement-nominal-ordinal-interval-ratio-examples/
6. https://statisticsbyjim.com/basics/nominal-ordinal-interval-ratio-scales/
7. https://careerfoundry.com/en/blog/data-analytics/data-levels-of-measurement/