PSLE Science Reality Lab Vol No.157 | “100% Inspected” — Does That Guarantee Every Defect Was Found?

PSLE-SCI-REALITY-0157

Wait, What? Every Unit Was Looked At

A factory makes 4,000 clear containers. Before packing, every single container passes through a visual-inspection station. The quality label says:

100% INSPECTED.

A reader translates that into:

“Then every possible defect must have been found.”

That conclusion is stronger than the evidence.

“100% inspected” can describe coverage: every unit went through an inspection step. It does not automatically describe detection perfection: whether the inspection method could find every relevant defect under every condition.

A tiny crack may be hidden by glare. A faint mark may fall below a camera’s detection capability. A trained inspector may miss a borderline feature. A method designed to find surface scratches may not detect an internal defect at all.

Reality Lab habit: “Every unit went through the check” and “every defect was detected” are two different scientific claims.

Quick Answer

100% inspection means all units were subjected to the stated inspection process, if the label is accurate.
It does not automatically mean the process detects every possible defect with perfect reliability.
Detection depends on what defect is defined, how large or obvious it is, the instrument or observer, the viewing conditions and the product itself.
A method can miss real defects; it can also sometimes reject acceptable items.
Inspector training, qualification, challenge samples and method validation can strengthen confidence in detection performance.
A visual inspection cannot automatically prove absence of invisible chemical, mechanical or internal problems that it was not designed to detect.
The phrase “100%” describes coverage only when attached to the correct noun: 100% of units inspected, not 100% of defects found.
A careful conclusion preserves both what was checked and what the checking method can realistically reveal.

The Exact Learner Job This Article Owns

This article owns one real-world evidence-transfer job: how a Primary 5/6 learner should evaluate a “100% inspected” product-quality statement without converting complete inspection coverage into a guarantee of perfect defect detection.

It does not teach industrial inspection engineering, medical-device regulation or pharmaceutical manufacturing. It also does not replace the existing PSLE Science owners for observation versus inference, method limitations or measurement limits. Reality Lab applies those ideas to a communication object learners may encounter on quality reports, packaging or factory descriptions.

Original Reality Lab Case: The ClearSpring Container Line

The following case is fictional. The numbers are constructed for teaching.

ClearSpring Works produces 4,000 transparent containers. Each container passes beneath a bright inspection lamp and is rotated once while an inspector looks for three specified visible defects:

a crack longer than 3 mm;
a dark foreign particle visible through the container wall;
a missing cap-seal ring.

All 4,000 units pass through the station. Thirty-six are rejected. The remaining containers are labelled as having undergone 100% visual inspection.

Later, a separate laboratory test finds that one accepted container has a tiny internal wall defect that was not visible under the inspection conditions.

Did the phrase “100% inspected” become false?

Not necessarily. If every unit genuinely passed through the stated visual inspection, the coverage claim can still be true. What failed was the stronger interpretation that the inspection was capable of detecting every possible defect with certainty.

Observed, Checked, Inferred and Overclaimed

Layer	ClearSpring example
Process fact	All 4,000 units passed through the stated visual inspection station
Observed result	36 units were rejected for specified visible defects
Reasonable claim	100% of the units were subjected to that inspection process
Question still open	How reliably could the method detect each relevant defect under the stated conditions?
Unsupported upgrade	Every possible defect in every unit was guaranteed to be found

The Most Important Word After “100%”

Percentages are incomplete without knowing what the denominator represents.

In “100% inspected”, the denominator is normally the units: all units entered the inspection process.

But a reader may silently replace the denominator:

100% of units inspected;
100% of defects detected;
100% of product properties checked;
100% certainty that no future failure will occur.

Those are different claims. Only the first is directly expressed by the phrase.

Coverage Is Not Capability

Imagine a torch shining over every square centimetre of a table. The whole surface receives light. That does not mean the torch can reveal a microscopic scratch too small for the eye to resolve.

Inspection works similarly.

Coverage asks whether every unit was checked.

Capability asks whether the checking method can reliably detect the feature of interest.

A process can have complete coverage but limited capability for certain defects.

The Defect Has to Be Defined

“Find defects” sounds simple until we ask what counts as a defect.

A visual inspection could be designed for:

large scratches;
visible cracks;
missing components;
colour changes;
foreign particles above a visible size;
shape deformation.

The same visual method might be unsuitable for:

chemical composition;
internal strength;
microscopic cracks below visibility;
electrical resistance;
future fatigue failure;
contamination that cannot be seen.

Therefore a claim about inspection quality must remain attached to the characteristic the inspection was designed to detect.

Detection Can Be Probabilistic

For some real inspection tasks, especially visual ones, detection is not perfectly deterministic. A defect may be found most of the time under defined conditions without being found every single time.

Recent U.S. FDA material on visible-particle inspection explicitly describes visual detection as a probabilistic process and stresses trained personnel, suitable methods and qualification. NIST has also developed reference materials to help train and qualify visual-inspection analysts.

For a Primary learner, the essential idea is enough:

Passing every unit through an imperfect detector does not turn the detector perfect.

Size Matters to Detection

A large dark mark may be easy to see. A tiny faint mark may be difficult. Two defects of the same physical type can therefore have different detection chances because their size, contrast or position differs.

This is why an inspection statement should not be interpreted without asking:

What size range was the method expected to detect?
How much contrast did the defect have against the background?
Could part of the object hide the defect?
Was the object moving?
How long was each unit viewed?

Lighting and Background Can Change What Is Visible

Take a transparent plastic sheet with a faint scratch. View it against a bright white wall. Then place it against a dark background and tilt the light.

The scratch can become easier or harder to see even though the scratch itself has not changed.

Inspection conditions are therefore part of the method. “Looked at” is not a complete scientific description.

Human Inspectors Need Qualification Too

A human observer brings strengths and limitations. Training can improve consistency. Challenge samples with known defects can test whether inspectors detect relevant features under realistic conditions.

But qualification should not be turned into another perfection claim. Passing a qualification exercise supports confidence that an inspector can perform a specified task under stated conditions. It does not guarantee that no error will ever occur in future work.

Automated Cameras Also Have Limits

Replacing a person with a camera does not remove the need for evidence.

An automated inspection system depends on:

camera resolution;
lighting;
focus;
viewing angle;
image-processing settings;
detection thresholds;
the examples used to qualify the system;
the types of defects it was designed to recognise.

A very sensitive threshold might catch more true defects but also reject more acceptable units. A loose threshold might accept more units but miss faint defects. Inspection design can involve trade-offs.

False Accept and False Reject

Consider two possible mistakes.

False accept: a unit with the relevant defect is accepted because the defect is missed.
False reject: an acceptable unit is rejected because a harmless feature is mistaken for a defect.

A quality system tries to manage both. Making a detector more aggressive can reduce one kind of error while increasing another.

The important lesson is not the terminology. It is that a real decision method can make errors in more than one direction.

“No Defect Seen” Is Not Always “No Defect Exists”

Suppose an inspector sees no crack.

The direct observation is:

No crack meeting the inspection’s visible detection conditions was observed.

The much stronger inference is:

No crack of any kind or size exists anywhere in the object.

The second claim may require a different method or additional evidence.

Worked Case 1: The Hidden Side

Every metal bracket passes in front of a camera, but the camera views only one side. A crack on the rear face is later found.

The line had 100% unit coverage but incomplete surface coverage. “Every bracket passed the camera” is not the same as “every surface of every bracket was imaged”.

Worked Case 2: The Defect Is Too Small

An inspection method is qualified using scratches 2 mm and larger. A later laboratory microscope finds a 0.3 mm scratch.

The microscope result does not automatically show the original inspection was performed incorrectly. The small scratch may simply have been below the practical detection capability of the visual method.

Worked Case 3: The Glare Problem

A shiny surface reflects the lamp directly into the inspector’s eyes at one angle. A faint dent becomes difficult to see.

Rotating the item or changing the lighting can improve detection. The example shows why viewing conditions belong in the method.

Worked Case 4: The Wrong Kind of Test

Every cable is visually checked for cuts and missing insulation. The company then claims, “100% inspection proves every cable carries the required electrical current.”

The evidence and claim are misaligned. Visual inspection can support claims about visible features it was designed to detect. Electrical performance requires appropriate electrical testing.

Worked Case 5: The Perfect Challenge Set

An inspector correctly identifies every defect in a training set of 20 samples. A manager writes, “The inspector will now detect every future defect.”

The qualification result is encouraging but does not justify unlimited certainty. Future products can differ in defect size, location, contrast and viewing condition.

Worked Case 6: The Camera Threshold

A vision system rejects any dark region above a chosen score. Lowering the threshold finds more faint marks but begins rejecting acceptable printed patterns.

This is a detection trade-off. “More sensitive” does not automatically mean “better in every way”. The threshold must suit the defined quality job.

Worked Case 7: Two Inspections, Different Jobs

All bottles undergo a visual crack check. A sample of bottles also undergoes a destructive pressure test.

It would be wrong to merge those statements into “all bottles were pressure tested”. One quality programme can contain both 100% inspection and sampled testing, each with a different evidence job.

What Evidence Strengthens a “100% Inspected” Claim?

A clear record showing every unit entered the inspection step.
A precise definition of the defects or characteristics being inspected.
A method appropriate for those defects.
Defined lighting, viewing, speed or instrument settings where relevant.
Qualified inspectors or validated automated systems.
Challenge samples representing realistic defect types and sizes.
Monitoring of missed defects and false rejections.
Transparent follow-up when unexpected defects escape detection.
A conclusion limited to the scope of the actual inspection.

What Weakens It?

Using “100%” without stating what was 100%.
No definition of the inspected defect.
An inspector who cannot reliably see the relevant feature.
A camera with insufficient resolution or poor lighting.
Inspecting only one side while implying the whole item was examined.
Claiming hidden chemical or internal quality from a surface look.
Ignoring known escape defects.
Treating one successful qualification exercise as permanent perfection.

Tempting Reasoning That Fails

“100% inspected means 100% defect-free.” Inspection coverage and product condition are different.
“Every unit passed, so no defect could have been missed.” The method may have limited detection capability.
“A human can miss defects, but a camera cannot.” Automated systems have their own detection limits and thresholds.
“If one escaped defect is found, the inspection claim must have been fraudulent.” Not necessarily. The coverage claim can be true while detection is imperfect.
“If the inspector passed training, future performance is guaranteed.” Qualification supports confidence; it does not remove all future uncertainty.
“Visual inspection proves every property of the product.” A method supports only the characteristics it can validly inspect.

How Far Can the Conclusion Travel?

A careful statement can travel this far:

Every unit was subjected to the stated inspection process for the specified defect types under the defined conditions.

It does not automatically travel to:

every defect was detected;
no defective unit escaped;
every product property was tested;
all invisible defects are absent;
the product can never fail later;
the inspection process has zero false accepts or false rejects.

PSLE-Style Transfer Case

A manufacturer visually inspects all 2,000 plastic covers under the same lamp for scratches at least 2 mm long. All covers pass through the inspection. Later, a microscope finds a 0.4 mm scratch on one cover.

A student says, “The 100% inspection was impossible, because one scratch was found later.”

Explain why the statement is not necessarily correct.

Answer: 100% inspection can mean every cover was subjected to the stated visual check. The later 0.4 mm scratch may have been below the size that the visual method was designed or capable of detecting. The coverage of the inspection and the detection capability of the method must be evaluated separately.

What additional evidence would help?

Useful information includes the defined defect size, lighting conditions, viewing method, inspector or camera qualification, and results from challenge samples containing known scratches of different sizes and contrasts.

Explained Practice

Practice A: Every unit passed through a camera. Does that prove every surface was visible to the camera? No.

Practice B: The method was designed for visible scratches. Does passing prove correct chemical composition? No.

Practice C: An inspector detects all defects in a training set. Is future detection guaranteed? No.

Practice D: Why can changing lighting matter? Contrast and visibility can change even when the physical defect stays the same.

Practice E: A company checks every unit for one defined feature. Can “100% inspected for that feature” be a meaningful claim? Yes, if the process records support it and the scope stays clear.

Delayed Independent Return: C-A-P-E

When you meet “100% inspected”, ask four quiet questions:

C — Coverage: What exactly received the inspection?
A — Attribute: What defect or characteristic was the method looking for?
P — Performance: How well can the method detect that attribute under realistic conditions?
E — Extrapolation: Is the claim being stretched to properties the inspection never tested?

This is a thinking aid, not an official examination template.

Parent and Tutor Teaching Guide

Prepare twenty cards. Put a faint pencil dot on three, a bold marker dot on three and leave the rest blank. Ask the learner to inspect all twenty quickly under ordinary room light. Record what they detect.

Then improve the lighting and allow more viewing time. The learner may find additional faint dots. Every card was inspected in both rounds, yet detection performance changed.

Next hide one mark on the back of a card while allowing only the front to be viewed. This demonstrates a different issue: complete unit coverage can coexist with incomplete surface coverage.

The lesson is not that inspection is useless. It is that the strength of an inspection claim comes from the match between the defect, the method and the conditions.

Authoritative Sources

The FDA material is used here only as an authoritative example of a general evidence principle: visual detection depends on method capability and qualified inspection conditions. This article gives no medical or product-safety advice. SEAB’s 2026 PSLE Science objectives require learners to interpret and analyse information and evaluate observations, information and methods. MOE’s syllabus advocates healthy scepticism: questioning observations, methods, processes and data.

The Quiet Return

“100% inspected” can be a useful and meaningful statement.

Read it precisely.

Ask what received the inspection, what the method was built to detect and how well it could perform that job.

Complete coverage is valuable evidence. It is not the same thing as perfect detection.

Small Group Tutorials