PSLE Science Reality Lab 356 | Expanded Uncertainty and k=2

PSLE-SCI-REALITY-0356

Wait, What? The “2” on a Calibration Certificate May Have No Degrees Attached to It

A calibration certificate reports a temperature result and then gives a line like this:

Expanded uncertainty U = 0.4°C, coverage factor k = 2.

A student circles the 2 and says, “So the thermometer may be wrong by 2°C.”

That mixes two different kinds of numbers. The 0.4°C is the reported expanded uncertainty in the temperature unit. The k = 2 is a dimensionless coverage factor used in forming that expanded uncertainty from a standard uncertainty. It is not a second temperature reading and it is not automatically a ±2°C error.

NIST Technical Note 1297 explains the relationship as U = k × u_c, where U is expanded uncertainty and u_c is combined standard uncertainty. Under commonly used conditions, k = 2 is associated with an interval having an approximate coverage probability near 95%. That statement has conditions and should not be turned into a universal guarantee.

Quick Answer

Read the measurement result and its unit first.
Find the reported expanded uncertainty U; that is the quantity expressed in the measurement unit.
Read k as the coverage factor used to calculate U, not as another error in degrees.
Check the stated coverage probability and method instead of assuming every k = 2 statement means exactly the same thing in every situation.
Do not call uncertainty “the mistake”. Uncertainty describes the remaining range of plausible values associated with the measurement result under the stated evaluation.

The Exact Learner Job This Page Owns

This page owns one real-world communication-object problem: reading a calibration certificate that reports expanded uncertainty and a coverage factor without turning the coverage factor into a measurement error or unit-bearing result.

It does not own measurement uncertainty as a whole. Existing PSLE Science pages remain the owners of measurement, precision, repeatability and evidence limits. Reality Lab applies those ideas to the kind of certificate, report or test record a learner may meet outside an examination.

Original Reality Lab Case: The Freezer Thermometer Certificate

This is an original composite case built for learning. A fictional calibration laboratory checks a thermometer near 0°C. Its certificate contains:

Reported item	Value
Indicated temperature	0.2°C
Correction	−0.1°C
Expanded uncertainty, U	0.4°C
Coverage factor, k	2
Stated coverage	Approximately 95% under the laboratory’s stated method

A learner should not add 2°C to anything merely because k = 2. The unit-bearing uncertainty has already been reported as 0.4°C.

Decode the Line Before Doing Any Arithmetic

Symbol	Scientific job	Unit?
y	Measurement result or estimate	Same unit as the measured quantity
u_c	Combined standard uncertainty	Same unit as the measured quantity
k	Coverage factor	No measurement unit
U	Expanded uncertainty, formed from k × u_c	Same unit as the measured quantity

The first Reality Lab habit is therefore unit discipline. A number without °C cannot suddenly become 2°C just because it appears beside a temperature result.

The Multiplication Check: Where Did U Come From?

If a certificate reports U = 0.4°C and k = 2, the combined standard uncertainty used for that line may be approximately 0.2°C because 2 × 0.2°C = 0.4°C. The factor expands the uncertainty interval; it is not an additional independent uncertainty to be added afterward.

So this reasoning is wrong:

“U is 0.4°C, and k is 2, so total uncertainty is 2.4°C.”

The coverage factor has already played its role in producing U.

The Coverage Check: “About 95%” Needs Conditions

NIST explains that when a normal distribution is a suitable model and the combined standard uncertainty is sufficiently reliable, k = 2 corresponds to an interval with a level of confidence of approximately 95%. NIST also explains that other values of k may be used for documented applications and that the relationship between k and coverage depends on the uncertainty model.

For a Primary 5/6 learner, the important habit is not calculating probability distributions. It is noticing that k belongs to a stated uncertainty method. The certificate should tell you what the laboratory means.

The “Wrong by” Check: Uncertainty Is Not a Confession of an Error

Suppose a corrected result is 20.0°C with expanded uncertainty 0.4°C. It is tempting to say, “The lab thinks it made a 0.4°C mistake.” That wording is misleading. The lab is reporting a measurement result together with a quantified uncertainty interval based on known sources of variation and uncertainty.

The true value is not revealed and then hidden again. Measurement uncertainty describes how confidently the result can represent the quantity under the stated measurement conditions.

The Correction Check: Correction and Uncertainty Are Different Jobs

A calibration certificate may report a correction and an uncertainty. The correction adjusts for a measured systematic difference between an instrument indication and a reference. The uncertainty tells us about the remaining doubt around the corrected result.

Do not merge them because both may contain plus-or-minus signs or decimal values.

The Range Check: Does One Uncertainty Apply Everywhere?

A thermometer can have different calibration uncertainties at different temperatures. A balance can behave differently at different loads. A sensor may be calibrated only over a particular range. Therefore, one uncertainty line should not automatically be copied to every possible use.

Always check which calibration point, range and method the uncertainty belongs to.

The Traceability Check: A Small U Does Not Prove Everything Else Is Good

A small reported uncertainty can be valuable evidence about the calibration measurement, but it does not prove that the instrument is being used correctly later. Storage, damage, drift, environmental conditions, poor technique or using the instrument outside its range can still affect later results.

Reality Lab separates quality of the calibration result from quality of every future use.

What Evidence Would Strengthen a Calibration Claim?

The certificate identifies the quantity, range and calibration point.
The result, correction, expanded uncertainty and coverage factor are clearly distinguished.
The certificate states the coverage probability or uncertainty convention.
The calibration method and reference standards are documented.
The instrument was later used within relevant conditions and has not been damaged or altered.

What Would Weaken It?

A reader turns k = 2 into ±2°C.
The unit-bearing U is ignored.
A certificate from one range is used to claim equal performance everywhere.
The calibration date is treated as proof that no drift can occur.
Uncertainty is described as if it were the exact size of a known mistake.

Worked Case 1: U = 0.4°C, k = 2

A certificate reports a corrected temperature of 10.0°C with U = 0.4°C and k = 2. The sensible reading is that the laboratory reports an expanded uncertainty of 0.4°C using a coverage factor of 2. The 2 has already been used in forming U and carries no °C unit.

Worked Case 2: Same k, Different U

Instrument A has U = 0.2°C, k = 2. Instrument B has U = 1.0°C, k = 2. Same coverage factor does not mean same uncertainty. The factor describes how the interval was expanded; the reported U tells you the size of that interval in the measurement unit.

Worked Case 3: Different k Values

Two certificates use different documented coverage factors. You cannot compare the printed U values fairly without checking the underlying uncertainty method and stated coverage. The same symbol can be used under different documented requirements.

Worked Case 4: Calibration Today, Measurement Next Year

A sensor was calibrated with a small uncertainty, then dropped and used for a year without checks. The old certificate remains genuine, but it does not guarantee the later reading. Evidence has a time and condition boundary.

Tempting Reasoning That Fails

“k = 2 means ±2 units.” k is dimensionless.
“U and k should be added.” k is used to calculate U.
“95% means every future reading is 95% correct.” Coverage belongs to the reported uncertainty interval under stated assumptions.
“Uncertainty is the error.” The exact error is generally not known.
“A calibration certificate guarantees the instrument forever.” Later use has its own conditions and risks.

Model and Measurement Limits

Uncertainty evaluation combines information from repeated measurements and other known sources such as standards, resolution, environmental effects and calibration models. The exact uncertainty budget can be technically complex. Primary students do not need to reproduce that calculation.

What they can do is preserve the structure of the evidence: result, unit, uncertainty, coverage factor, range, method and conditions.

How Far Can the Conclusion Travel?

A calibration certificate can support a carefully bounded statement about a measurement result and its uncertainty at stated conditions. It cannot by itself prove that every future reading is correct, that the instrument cannot drift, or that k is an error in the measured unit.

PSLE-Style Transfer Case

A fictional balance certificate states: “Expanded uncertainty U = 0.06 g, k = 2.” A pupil says the balance may be wrong by 2 g.

Question: Explain why this interpretation is incorrect.

Reasoned answer: The reported expanded uncertainty is 0.06 g. The value k = 2 is a coverage factor used to calculate that uncertainty and has no gram unit. It does not mean an additional 2 g error.

Explained Practice

Practice A: U = 0.8°C, k = 2. Which number has the temperature unit? U = 0.8°C.

Practice B: Two certificates both have k = 2 but different U values. Are the instruments equally uncertain? Not necessarily; compare U on the relevant measurement basis.

Practice C: A device is calibrated at 20°C. Can you assume the same uncertainty at 200°C? Not without evidence that the calibration range and uncertainty statement cover that condition.

Delayed Independent Return: U-N-I-T

U — Uncertainty: What U is actually reported?
N — Number role: Is k a multiplier or a measured quantity?
I — Interval: What coverage statement belongs to it?
T — Test conditions: Which range and calibration point does it apply to?

Parent and Tutor Teaching Guide

Write “0.4°C” on one card and “×2” on another. Ask which card can be added to a temperature and which card is an instruction for multiplication. Then build the simple relationship 0.2°C × 2 = 0.4°C. The aim is not to teach uncertainty statistics; it is to stop a multiplier from masquerading as a measured quantity.

Next, place three certificates side by side with different U values but the same k. Ask the learner what stayed the same and what changed. This reveals that the coverage convention and the size of the reported uncertainty are separate pieces of information.

Authoritative Sources

The Quiet Return

A scientific report can place several numbers beside one another without giving them the same job.

Keep the unit attached to the quantity, and a coverage factor will stop pretending to be a temperature.

Small Group Tutorials

PSLE Science Reality Lab Vol No.356 | “U = 0.4°C, k = 2” — Does k = 2 Mean the Thermometer Can Be Wrong by 2°C?