Measurement Techniques: From Physical Devices to Abstract Parameters

Every time you check the temperature, weigh vegetables in a market, or read a satisfaction score on a survey, you are relying on measurement. Measurement is simply the process of assigning numbers to things so we can compare, analyse, and make decisions. But the techniques we use are not all the same. Weighing a packet of rice is very different from measuring how “satisfied” a reader is with a library, or how “intelligent” a student is. This post walks through the full spectrum of measurement techniques, starting with familiar physical devices, moving into the trickier world of abstract and subjective parameters, and finishing with the digital tools that now dominate modern science.

What measurement really means 🔗

At its core, measurement is the act of attaching a numerical value to a property of an object or event. The philosophy of measurement shows that even something as basic as weighing relies on two ideas: ordering (one object is heavier than another) and concatenation (placing weights together on a pan adds them up). When a property allows both ordering and addition, we can build a meaningful numerical scale around it. This is why physical measurement feels so reliable, and why measuring something abstract like “service quality” needs extra care.

The key insight is that not every number means the same thing. A jersey number, a ranking, a temperature reading, and a weight are all numbers, but they carry very different amounts of information. Understanding this difference is the foundation of every measurement technique discussed below.

Physical measurement techniques 🔗

Physical measurement deals with tangible quantities such as length, mass, temperature, volume, and time. These are the techniques most people learn first in school, and they share a common feature: a physical device responds to a property, and the response is read off against a calibrated scale.

Measuring mass with a balance 🔗

A balance works on a beautifully simple principle. An equal-arms balance compares an unknown object against known weights until the two pans level out. The triple beam balance commonly used in laboratories uses three sliding riders along marked beams; the pointer must rest at zero before any object is placed, and the riders are moved until balance is achieved. The mass is then read directly in grams. Because weights can be physically added together on a pan, mass measurement produces what theorists call an additive scale, where addition, multiplication, and division all make sense.

Measuring temperature with a thermometer 🔗

Temperature is more subtle. A liquid-in-glass thermometer does not measure heat directly. Instead, it relies on a property that changes consistently with temperature: as the liquid (mercury or alcohol) warms, it expands and rises up a thin tube, and the position is read against a printed scale. The principle of inferring temperature from a mechanical or physical change underpins many instruments, including bimetallic strips and pressure thermometers.

There is a catch, though. Two thermometers will not agree unless they follow the same agreed scale. This is why temperature scales like Celsius, Fahrenheit, and Kelvin are built by fixing reproducible reference points, such as the freezing and boiling points of water at standard pressure. Metrologists, the experts in the science of measurement, even maintain standard instruments like the constant-volume gas thermometer to calibrate ordinary ones.

Reading a numerical result 🔗

A common thread runs through all these devices: the result must be read off a graduated scale. To read any instrument correctly, you first work out the scale increment, the value of each small division. The method is the same for a thermometer, a graduated cylinder, or a spring scale: take two labelled marks, subtract their values, and divide by the number of intervals between them. This simple arithmetic turns a physical position into a precise number, and it is why physical measurement is considered the most objective form of measurement.

Abstract and subjective measurements 🔗

Now we move into harder territory. How do you measure intelligence, satisfaction, attitude, or the relevance of a search result? These parameters are intangible. You cannot place “satisfaction” on a balance or read “intelligence” off a thermometer. Yet social sciences, psychology, and Library and Information Science constantly need to measure them. The solution is to use carefully designed scales and instruments that convert subjective judgements into numbers.

The four levels of measurement 🔗

The most important framework here was introduced by psychologist Stanley Smith Stevens in 1946. He argued that all measurement falls into one of four levels of measurement, and this classification is still used across psychology, statistics, and the social sciences today.

Nominal scale: Numbers are used only as labels for categories with no order. Assigning 1 to “fiction” and 2 to “non-fiction” in a catalogue is nominal; the numbers carry no ranking.

Ordinal scale: Categories now have a meaningful order, but the gaps between them are unknown. A satisfaction rating of “poor, fair, good, excellent” is ordinal. We know excellent beats good, but we cannot say by how much.

Interval scale: The order is meaningful and the gaps between values are equal, but there is no true zero. Temperature in Fahrenheit or Celsius is the classic example; the difference between 10 and 20 degrees equals the difference between 50 and 60 degrees, but zero degrees does not mean “no temperature”.

Ratio scale: This has all the properties of interval data plus a true zero, which makes ratios meaningful. Mass, length, and time are ratio measurements. Twenty kilograms really is twice ten kilograms.

This hierarchy matters because the level of measurement decides what statistics you can legitimately use. Stevens himself cautioned that computing an average only makes full sense for interval or ratio data, and that most psychological scales are really only ordinal. The same debate applies whenever we attach numbers to opinions.

Measuring intelligence and attitudes 🔗

Intelligence is measured through standardised tests that produce a score such as an IQ. These scores behave like interval data: the difference between scores of 100 and 110 is treated as equal to the difference between 110 and 120, which allows meaningful comparison. Attitudes and opinions are usually captured with rating scales, most famously the Likert scale, where respondents choose options like “strongly agree” through “strongly disagree”. These responses are ordinal in nature, even though researchers often assign numbers to them and calculate averages.

Measuring library service quality 🔗

Library and Information Science offers an excellent example of measuring an abstract idea. How good is a library’s service? This question cannot be answered with a single physical reading, so the profession developed survey-based instruments. The widely used LibQUAL+ tool, adapted from the SERVQUAL model, measures service quality across dimensions such as the affect of service, reliability, the library as a place, and access to information.

LibQUAL+ works by capturing the gap between what users expect and what they perceive they actually receive. Users rate each item on a scale, and the difference reveals where the library is falling short. A study of an academic library using this framework found that service quality fell short of user expectations and even differed by gender, showing how a subjective concept can be turned into actionable, comparable data. Crucially, this lets library managers move beyond simple counts like budget or number of books toward genuine, user-centred quality measures.

Measuring relevance in information retrieval 🔗

Relevance is one of the most slippery concepts in our field. When you search a database, how do we measure whether the results are “relevant”? Information scientists handle this by using human judges to label documents as relevant or not relevant for a given query, and then computing measures such as precision (the fraction of retrieved documents that are relevant) and recall (the fraction of relevant documents that were retrieved). This converts a deeply subjective judgement into reproducible numbers that allow one search system to be compared against another.

Digital scales and devices 🔗

The most visible change in modern measurement is the shift from analog to digital instruments. An analog instrument indicates a value through continuous physical movement, such as a needle on a dial or liquid in a tube. A digital instrument instead converts the measured quantity into discrete numerical values shown directly on a screen.

How digital measurement works 🔗

At the heart of most digital devices is a sensor and an analog-to-digital converter (ADC). The sensor responds to a physical property, producing an electrical signal. The ADC then converts this continuous signal into a digital number that a microcontroller or computer can process and display. The resolution of the reading depends largely on the quality of the ADC; a higher-bit converter can detect much smaller changes in the measured quantity.

Why digital devices took over 🔗

Digital instruments offer several practical advantages. Because they have few or no moving parts and display a value directly, the human error involved in reading them is very small. There is no need to interpret a needle position or interpolate between scale markings, which removes a major source of reading error. Digital sensors can also filter out signal noise and include built-in processing for calibration, making readings more consistent. They interface easily with computers, which is what makes automatic data logging and large-scale data analysis possible.

None of this makes digital instruments perfect. Conversion introduces its own errors from factors like temperature and quantization, and an unstable reading can flicker when a measurement sits at the edge of the device’s accuracy limit. The accuracy of a digital instrument ultimately depends on the quality of all its electronic components, not just its tidy screen.

Why this matters for information work 🔗

The digital shift is not just about gadgets. Once measurement becomes digital, data can be captured, stored, and transmitted at enormous scale. This is exactly what makes fields like informetrics and scientometrics possible. Citation counts, download statistics, usage logs, and web analytics are all digital measurements collected automatically. The same logic that turns a thermometer reading into a number now turns scholarly activity into measurable, comparable data, allowing researchers to study the growth and impact of knowledge itself.

From devices to parameters: the common thread 🔗

Across all three domains, the goal is identical: to represent some property of the world as a number we can work with. Physical measurement does this directly through calibrated devices and additive scales. Abstract measurement does it indirectly through carefully designed scales that respect the four levels of measurement. Digital devices do it electronically, converting signals into precise figures at scale. The technique changes, but the underlying purpose, turning the world into comparable numbers, stays constant. Recognising which type of measurement you are dealing with is what keeps your numbers honest and your conclusions valid.

What do you think? When you read a number, such as a library satisfaction score or an IQ result, do you stop to ask which level of measurement produced it? And as more of our world is measured digitally and automatically, do you think we risk trusting numbers more than the messy realities they are meant to represent?

References

1. https://plato.stanford.edu/entries/measurement-science/
2. https://teksguide.org/resource/measurement
3. https://www.cedengineering.com/userfiles/E02-012%20-%20Principles%20and%20Methods%20of%20Temperature%20Measurement%20-%20US.pdf
4. https://phys.libretexts.org/Bookshelves/University_Physics/University_Physics_(OpenStax)/University_Physics_II_-_Thermodynamics_Electricity_and_Magnetism_(OpenStax)/01:_Temperature_and_Heat/1.03:_Thermometers_and_Temperature_Scales
5. https://en.wikipedia.org/wiki/Level_of_measurement
6. https://careerfoundry.com/en/blog/data-analytics/data-levels-of-measurement/
7. https://www.psyctc.org/psyctc/glossary2/ratio-scaling/
8. https://www.libqual.org/documents/admin/blixrud.pdf
9. https://pmc.ncbi.nlm.nih.gov/articles/PMC10730435/
10. https://www.processsensing.com/en-us/blog/digital-versus-analog-sensors.htm
11. https://www.sensortips.com/featured/analog-digital-conversion-impact-sensor-accuracy/
12. https://www.electricalandcontrol.com/electronic-instruments/
13. https://www.linkedin.com/pulse/analog-vs-digital-sensors-unveiling-differences-khadar-baba-shaik