WONDER — The Mystery
DOK 1–2 — ObserveEvery Measurement Has an Uncertainty
You and your sibling are checking a rover’s measuring tape. Tape A gives almost the same reading every time; tape B gives more scattered readings. Your sibling immediately trusts A. Could a very repeatable tape still disagree with a reference length?
The Mystery:
Does taking more readings fix every kind of error? How can we describe the uncertainty in a result calculated from several measurements?
Today's Mission: distinguish precision from accuracy, investigate systematic bias and random scatter, and calculate uncertainty in derived results using the AS rules.
PREDICT — Commit to Your Answer
DOK 2 — PredictScenario: A student measures the same length 5 times: 2.19, 2.20, 2.19, 2.21, 2.20 cm. The TRUE value is 2.0 cm.
How would you describe these measurements?
Step 1: Pick your answer.
STOP! Before revealing — why did you pick that?
Correct Answer: A) Precise but not accurate (clustered together, but off from true value).
Did you pick the right WHY? You should have selected: "The measurements are close TOGETHER (precise) but all ABOVE the true value of 2.0 (not accurate). Precision = closeness of repeated measurements to each other; accuracy = closeness to the true value."
The Simple Truth: the readings span just 0.02 cm and their mean is 2.198 cm, about 0.20 cm above the reference. They show high repeatability but poor closeness to this reference. This suggests a bias worth investigating by calibration; five readings alone do not establish its cause.
PLAY — The Error & Uncertainty Lab
DOK 2 — InvestigateWhat you will DO and OBSERVE
- • DO: adjust fixed bias, random standard deviation and number of readings; use Play, Pause, Step and Reset.
- • OBSERVE: bias moves the cluster; random effects scatter individual readings. The sample mean varies from run to run.
- • DISCOVER: more independent readings reduce the model’s random uncertainty in the mean. They do not reduce the scatter of individual readings or remove bias; a particular mean need not move closer on every run.
Throw darts at the bullseye. Adjust the systematic offset (shifts the cluster), random spread (scatters the darts), and number of measurements. The true value is the bullseye center.
Reading the diagram: horizontal displacement is reading error in cm. Vertical scatter is only the dartboard analogy. The purple cross marks the model bias; the green cross marks the sample mean.
Readings & mean uncertainty (model extension)
Reading the diagram: cyan dots are individual horizontal errors. The gold interval is mean ± σ/√n, using the selected model σ. This statistical extension assumes independent readings; it does not include bias or guarantee that the true value lies inside.
Live Readouts
BULLSEYE!
Ideal model: zero bias and zero random scatter.
Live AS uncertainty calculator
x and y are lengths in m. Adjust their values and absolute uncertainties. These are the syllabus simple-addition estimates; ×/÷ use a first-order approximation.
DISCOVER — The Golden Rules
DOK 1 — RecallYou've played with the Error Lab. Now flip each card to reveal the four golden rules for Errors, Uncertainties, Precision, and Accuracy.
Rule #1
Systematic Errors
Click to flip
Stable or Predictable Bias
Systematic error: a bias that remains constant or changes predictably under repeat conditions. A zero error is a fixed offset example.
Causes: calibration, zero error, or a consistent parallax/viewpoint bias. Effect: reduces ACCURACY. Fix: calibrate, zero, or correct for offset. CANNOT be reduced by averaging!
Rule #2
Random Errors
Click to flip
Unpredictable Scatter
RANDOM errors cause measurements to SCATTER unpredictably (sometimes high, sometimes low).
Causes: fluctuations, reading errors, noise. Effect: reduces PRECISION. Fix: take repeated measurements and average them — this reduces the effect of random error on the mean.
Rule #3
Precision vs Accuracy
Click to flip
The Dartboard Analogy
PRECISION = how close repeated measurements are to EACH OTHER. Tight cluster = high precision.
ACCURACY = closeness to the TRUE value. Hitting bullseye = high accuracy. You can have one without the other!
Rule #4
Uncertainty Propagation
Click to flip
AS uncertainty estimates
+ / −: add absolute uncertainties: Δz = Δx + Δy.
× / ÷: add fractional or percentage uncertainties: Δz/|z| ≈ Δx/|x| + Δy/|y|.
Multiply/divide rules are first-order estimates, best for small fractional uncertainties. A half-range estimate describes scatter and does not detect a fixed bias.
Examiner trap: subtraction still adds absolute uncertainties in this syllabus method. Do not subtract them.
All four rules unlocked!
MISCONCEPTION ATTACK — Destroy the Myths
DOK 2 — AnalyzePick TRUE or FALSE for each myth. The truth will be revealed instantly.
💀 Myth #1: "Taking more measurements reduces systematic errors."
Correct — the answer is FALSE.
Averaging reduces the effect of RANDOM error on the mean, NOT systematic error! A systematic bias persists under repeated conditions. In the fixed zero-error example below, the mean retains the same offset.
If a length gauge reads +0.2 cm when the true separation is zero, repeating the same measurement leaves that fixed offset in the mean. Zero the gauge, calibrate it, or subtract the known offset.
💀 Myth #2: "A precise measurement is always accurate."
Correct — the answer is FALSE.
Precision and accuracy describe different qualities. Closely grouped readings can still be displaced from a reference by a systematic bias.
The classic example: a scale that reads 2 kg high every time — very precise (same reading every time) but very inaccurate (2 kg off). Precision = closeness of repeated values to each other. Accuracy = closeness to the true value. You need BOTH for good measurements. A systematic bias can give high precision but poor accuracy. Random scatter can give low precision even when the mean is close to the true value.
Myths destroyed! You're inoculated against the two biggest error-and-uncertainty traps.
THE SECOND 'AHA!' — Instructor Drill
DOK 2 — ApplicationYour instructor has 4 rapid-fire questions about errors, uncertainties, and propagation. The live scene on the left updates as you answer. Get all 4 right to unlock the practice round.
Live Error Scene
Reading the diagram: a tile lights up only when its drill answer is correct.
Current Question Visualization
Waiting for Q1...
Instructor Q1 — Spot the Error Type
A length gauge displays +0.2 cm with its jaws closed, when the true separation is zero. This is a:
Instructor Q2 — Averaging
For independent repeated readings under stable conditions, averaging reduces the influence of which type of error on the mean?
Instructor Q3 — Absolute Uncertainty (Addition)
x = 10 ± 1 m, y = 5 ± 0.5 m. Find the absolute uncertainty in (x + y).
Instructor Q4 — Percentage Uncertainty (Multiplication)
x = 10 ± 1 m, y = 5 ± 0.5 m. Find the percentage uncertainty in (x × y).
4/4 — Errors & Uncertainties Mastered!
You now know: zero errors are systematic, repeats + averaging reduce random influence on the mean, addition adds absolutes, multiplication adds percentages. Time to practice!
PRACTICE — Calculate the Uncertainty
DOK 2 — Practice🎯 Challenge 1: Percentage Uncertainty
A length is measured as 20 ± 1 cm. Find the percentage uncertainty (in %).
Step-by-Step Solution (click to expand)
Step 1 — Identify: Percentage uncertainty = Δ xx × 100%
Step 2 — Substitute: = 120 × 100%
Step 3 — Solve: = 0.05 × 100% = 5%
🎯 Challenge 2: Adding Uncertainties
x = 4.0 ± 0.2 m, y = 2.0 ± 0.1 m. Find the absolute uncertainty in (x + y) (in m).
Step-by-Step Solution (click to expand)
Step 1 — Identify: For addition/subtraction, ADD ABSOLUTE uncertainties.
Step 2 — Substitute: Δ(x+y) = Δ x + Δ y
Step 3 — Solve: = 0.2 + 0.1 = 0.3 m
Note: The sum itself is x + y = 4.0 + 2.0 = 6.0 m. So the answer is 6.0 ± 0.3 m.
🎯 Challenge 3: Evaluate the Measurement Plan
During calibration, a balance gives the readings shown below. A student proposes weighing the same object 100 times and using the mean to remove the discrepancy. Evaluate this plan and state a better method.
Step-by-Step Solution
Step 1 — Identify: the differences are 0.5 − 0, 1.5 − 1 and 2.5 − 2 kg.
Step 2 — Substitute: each difference is +0.5 kg, indicating a fixed systematic offset in the tested range. The mean retains the same offset.
Step 3 — Solve: the plan does not remove this bias. Zero/calibrate the balance or subtract the known 0.5 kg offset, then retain an appropriate uncertainty estimate.
CHALLENGE — The Einstein Sequence
DOK 2 — SequencingBuild a Defensible Speed Report
A rover travels x = (5.0 ± 0.1) m in t = (2.0 ± 0.1) s. A student reports “2.5000 m/s” without uncertainty. Arrange six linked operations to produce a sensible speed report. Each step uses the previous result. Select a card, then a slot; select a filled slot with no card selected to return it.
Build the workflow in order:
Available steps (shuffled — pick one, then click a slot):
Real Physics — The Higgs Boson & Statistical Significance (extension)
In 2012, CERN experiments reported evidence for a new particle consistent with the Higgs boson. A key idea was 5σ statistical significance: the observed excess was extremely unlikely to arise from a background-only fluctuation under the statistical model.
The Physics: A 5σ result means that, under the usual background-only Gaussian model, a fluctuation at least this extreme would be extraordinarily unlikely (about 1 in 3.5 million using the standard one-sided convention). That is evidence against the background-only explanation — not a literal probability that the discovery is "wrong."
From planets to particles, scientists need uncertainty information to judge how strong a result really is. In experimental physics, a value is much more useful when its uncertainty and measurement method are also known.
TEACH — The Feynman Technique
DOK 3 — Explain & ConnectYour Final Mission:
Your 10-year-old younger sibling is helping you measure the length of a room with a tape measure. They ask: "Why do you write '4.2 ± 0.1 m' instead of just '4.2 m'? And what's the difference between precision and accuracy?"
Explain to your sibling — in plain words — what an uncertainty is, the difference between systematic and random errors, and why precision and accuracy are not the same thing!
Type Your Explanation (≥ 3 sentences)
Draw a measurement interval or a precision/accuracy comparison for your sibling. Add at least three purposeful strokes. This checks effort; it does not mark the physics.
Cyan pen selected.
10 · THE CREST VAULT DOK 1 — Consolidate
Final reward below
Hidden exam secrets inside — Crest's free notebook.
📓 PHYSICS CREST — Secret Notebook
🎯Mission 1.3 — Errors & Uncertainties
"Every Measurement Has an Uncertainty" 🎯📏
🌍 OVERVIEW: Errors & Uncertainties
Percentage uncertainty = (Δx/|x|) × 100%.
z = x ± y: Δz = Δx + Δy (simple conservative estimate).
z = xy or x/y: Δz/|z| ≈ Δx/|x| + Δy/|y|.
z = xⁿ: Δz/|z| ≈ |n| Δx/|x|. Exact numerical factors do not carry measurement uncertainty.
Use small-uncertainty approximations where appropriate. Statistical standard uncertainties use a different combination rule; they are not the same as these conservative syllabus estimates.
| Issue | What it affects | Response |
|---|---|---|
| Fixed bias | Closeness to reference | Zero, calibrate or correct |
| Random scatter | Repeatability | Repeat and average |
| Derived result | Uncertainty estimate | Use the correct propagation rule |
No measurement is exact. An uncertainty estimate describes the range associated with the measured value. We often write it as value ± uncertainty (e.g., 1.50 ± 0.02 m). Error is measured value minus reference value; uncertainty estimates how well the value is known. It is not a known error.
1. Systematic Errors (FIX BY CALIBRATION)
- Systematic error: a bias that remains constant or changes predictably under repeat conditions. A zero error is a fixed offset example.
- Causes: zero error, calibration error, or a consistent parallax error from viewing a scale at the same wrong angle.
- Effect: Reduces ACCURACY (closeness to true value).
- Fix: Calibrate the instrument, correct for known offset, or zero before use.
- CANNOT be reduced by averaging!
2. Random Errors (REDUCE THEIR EFFECT WITH REPEATS + MEAN)
- Definition: Cause measurements to SCATTER unpredictably (sometimes high, sometimes low).
- Causes: Fluctuating conditions, reading fluctuations, environmental noise.
- Effect: Reduces PRECISION (readings are less closely grouped).
- Fix: Take repeated measurements and average them — this reduces the effect of random error on the mean.
3. Precision vs Accuracy (The Dartboard Analogy)
- PRECISION: How close repeated measurements are to EACH OTHER. Tight cluster = high precision.
- ACCURACY: How close measurements are to the TRUE value. Hitting bullseye = high accuracy.
- Precise but NOT accurate → tight cluster off-center (SYSTEMATIC error).
- Mean close to true value but LOW precision → scattered around the bullseye (random error dominates).
- BOTH → tight cluster on bullseye.
- NEITHER → scattered and off-center.
4. Estimating & Converting Uncertainty
- Repeated readings: where appropriate, absolute uncertainty = half the range = (maximum − minimum)/2. This is a scatter estimate; it does not include an undetected systematic bias.
- Absolute → percentage: (absolute uncertainty / measured value) × 100%.
- Percentage → absolute: (percentage uncertainty / 100) × measured value.
5. Uncertainty Propagation
When you calculate a derived quantity, uncertainties propagate.
- Addition / Subtraction: ADD ABSOLUTE uncertainties. Δ(x+y) = Δx + Δy
- Multiplication / Division: add fractional or percentage uncertainties using the first-order AS estimate: Δz/|z| ≈ Δx/|x| + Δy/|y|.
Worked Example: x = 5.0 ± 0.1 m, t = 2.0 ± 0.1 s. Speed = x/t = 2.5 m/s. % uncertainties: (0.1/5.0) × 100% = 2%; (0.1/2.0) × 100% = 5%. Total = 7%. Raw uncertainty estimate = 0.175 m/s, so report sensibly as 2.5 ± 0.2 m/s.
🚨 CREST EXAM SECRETS
🔑 EXAM TIP: Report uncertainty sensibly
Keep extra digits during working. Round the final uncertainty suitably (usually one significant figure; sometimes two) and give the measured value to the same decimal place. A small scatter does not prove accuracy.
⚠️ COMMON MISTAKE #1: "Averaging fixes systematic errors"
If a length gauge reads +0.2 cm when the true separation is zero, repeating the same measurement leaves that fixed offset in the mean. Zero the gauge, calibrate it, or subtract the known offset.
- To fix systematic: calibrate, zero, or correct for offset.
- To reduce random influence: repeat measurements and calculate a mean.
⚠️ COMMON MISTAKE #2: "Precise = Accurate"
Precision does not guarantee accuracy. A balance with a stable +2 kg offset can give closely grouped readings that miss the reference mass. Random scatter can also make individual readings less accurate, even when their mean is close to the reference.
🧠 MEMORY TRICKS
- "Plus/Minus → Plus the absolutes. Times/Divide → Plus the percents."
- "Stable or predictable bias = SYSTEMATIC. Unpredictable scatter = RANDOM."
- "Systematic = Stable bias. Random = Scatter."
- "Calibrate for Systematic. Average for Random."
🔬 BONUS FACT: The Higgs Boson Discovery (extension)
In 2012, CERN experiments reported a new particle consistent with the Higgs boson. The result reached the particle-physics discovery threshold of about 5σ significance, meaning a background-only fluctuation this extreme would be extraordinarily unlikely under the statistical model. This is why careful treatment of uncertainty and statistical significance matters.
✅ CREST CHECKLIST (memorize!)
- Every measurement = value ± uncertainty.
- Systematic errors: stable or predictable bias → investigate and correct/calibrate.
- Random errors: scatter → reduce their effect with repeated measurements and a mean.
- Precision = closeness of repeated values to one another. Accuracy = closeness to the true value. Accuracy is not established from repeatability alone.
- Addition/subtraction → ADD ABSOLUTE uncertainties.
- Multiplication/division → ADD PERCENTAGE uncertainties.
- Averaging does NOT fix systematic errors.
- Precision alone does not guarantee accuracy.
- % uncertainty = (Δx / |x|) × 100%, for non-zero x.
- Repeated readings: where appropriate, absolute uncertainty ≈ half the range. This is a scatter estimate; it does not include an undetected systematic bias.
— Crest ✨
"If you don't know your uncertainty, you don't know your measurement!"
Mission 1.3 Complete! · All 10 levels cleared · Physics Crest
You've earned the rank of Master of Errors & Uncertainties.
Errors & Uncertainties — mastered. You now know what every measurement is hiding!