WHY MATHEMATICS?

Why Mathematics? | Chen–Shapiro Test: Normalized Spacings, Expected Normal Scores, Local Gaps and Normality Evidence

How adjacent sample gaps can be compared with their expected normal-score gaps, turning the geometry of an ordered plot into a simple, auditable statistic.

Why mathematics is important becomes easier to see when a real decision depends on it. This guide follows the mechanism from definitions and dimensions to a worked calculation, validation, practical learning and responsible use. It keeps factual evidence separate from interpretation and treats every illustrative scenario as an example rather than a promise of results.

Choose a reading route

Order the observations and compute adjacent gaps
Generate expected normal scores from plotting positions
Normalize each observed spacing by its expected gap
Scale the aggregate and calibrate by sample size

Chapter map

1. Order
2. Observed gaps
3. Plot positions
4. Normal scores
5. Expected gaps
6. First normalized gap
7. Second gap
8. Symmetry
9. Aggregate
10. Sample scale
11. Denominator
12. CS statistic
13. Rejection rule
14. Ties
15. Comparison
16. Claim

Visit the How Mathematics Works hub for the wider series and its connections to education, reasoning, science, computing, design and careers.

1. Order

The central mathematical idea is concrete. Sort all observations.

A concrete situation

1,2,3,4,5.

What the mathematics does not license

Do not rank within subgroups. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Keep one list. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

2. Observed gaps

Here the geometry becomes operational. Subtract adjacent values.

A concrete situation

Every gap equals one.

What the mathematics does not license

Ties create zero gaps. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Record resolution. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

3. Plot positions

This step deserves a careful pause. Compute (i-.375)/(n+.25).

A concrete situation

Five probabilities are symmetric.

What the mathematics does not license

Other formulas yield different scores. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

State convention. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

4. Normal scores

The useful question is not only how to calculate, but what the calculation assumes. Apply inverse Phi.

A concrete situation

Scores≈-1.180,-.497,0,.497,1.180.

What the mathematics does not license

Tail quantiles need precision. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Use stable routine. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

Did You Know? In this chapter, the most important number is useful only together with its construction rule. Changing scaling, sample membership or validation design can change the numerical answer without changing the arithmetic.

5. Expected gaps

A small example makes the mechanism visible. Subtract adjacent scores.

A concrete situation

.683,.497,.497,.683.

What the mathematics does not license

All must be positive. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Audit order. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

6. First normalized gap

This is where modelling choices begin to affect evidence. Divide 1 by .683.

A concrete situation

Value≈1.464.

What the mathematics does not license

This is not a probability. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Label ratio. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

7. Second gap

The central mathematical idea is concrete. Divide 1 by .497.

A concrete situation

Value≈2.012.

What the mathematics does not license

Rounding affects sum. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Keep precision. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

8. Symmetry

Here the geometry becomes operational. Check mirrored gaps.

A concrete situation

Ratios repeat 1.464,2.012.

What the mathematics does not license

A skewed sample breaks pattern. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Plot. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

9. Aggregate

This step deserves a careful pause. Sum four ratios.

A concrete situation

Total≈6.952.

What the mathematics does not license

One tail gap can dominate. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Show contributions. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

10. Sample scale

The useful question is not only how to calculate, but what the calculation assumes. Compute S with declared divisor.

A concrete situation

S≈1.581.

What the mathematics does not license

Mixing population and sample SD changes result. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Use n-1. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

11. Denominator

A small example makes the mechanism visible. Multiply (n-1)S.

A concrete situation

Value≈6.324.

What the mathematics does not license

n rather than n-1 is wrong. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Check formula. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

12. CS statistic

This is where modelling choices begin to affect evidence. Divide aggregate by denominator.

A concrete situation

CS≈1.099.

What the mathematics does not license

A number near one is not automatic acceptance. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Use critical values. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

13. Rejection rule

The central mathematical idea is concrete. Follow the implementation’s stated tail.

A concrete situation

Small values commonly indicate departure.

What the mathematics does not license

Guessing direction reverses inference. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Verify source. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

14. Ties

Here the geometry becomes operational. Inspect grouped or rounded data.

A concrete situation

Zero spacings are informative.

What the mathematics does not license

Jitter invents precision. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Use tie-aware method. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

15. Comparison

This step deserves a careful pause. Place CS beside probability plot.

A concrete situation

Local gaps explain curvature.

What the mathematics does not license

Multiple tests need a plan. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Choose purpose. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

16. Claim

The useful question is not only how to calculate, but what the calculation assumes. State spacing-based evidence.

A concrete situation

The ordered gaps agree or disagree with normal spacing.

What the mathematics does not license

No cause is proven. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Check measurement. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

A seven-day practice lab

Day 1: define the decision, observation unit, X variables and Y or reference condition.
Day 2: reproduce the smallest worked example and check every dimension, unit and sign.
Day 3: build a transparent baseline and write what it assumes.
Day 4: fit the main method inside a training-only pipeline; record every parameter.
Day 5: challenge one assumption with a deliberately difficult dataset or operating regime.
Day 6: compare validation metrics with the practical workload created by errors.
Day 7: present one supported conclusion, one unresolved uncertainty and one safe next experiment.

Questions for a student, parent or mentor

Which part of the model is a mathematical definition, and which part is a choice?
What small calculation would catch a sign, scale or dimension error?
What evidence comes from training, validation and the untouched test?
What benign condition could look like a fault or prediction failure?
How would the answer change for another response, asset, school project or time period?
What should a human review before acting on the score?

A method-specific audit before deployment

The Chen–Shapiro test converts adjacent order-statistic gaps into normalized spacings. Each observed gap is divided by the corresponding expected-normal-score gap. If the ordered sample follows a normal pattern, the adjusted spacings are broadly compatible and their aggregate, scaled by sample standard deviation, stays near its calibrated normal range.

For 1,2,3,4,5, use plotting positions (i-0.375)/(n+0.25). The expected normal scores are approximately -1.180,-0.497,0,0.497,1.180, with adjacent gaps 0.683,0.497,0.497,0.683. Dividing the four observed gaps of one by these gives about 1.464,2.012,2.012,1.464.

Their sum is about 6.952. Dividing by (n-1)S with S≈1.581 gives a Chen–Shapiro statistic near 1.099. This hand result illustrates the calculation. It is not a p-value, and the rejection direction and critical values must follow the specific implementation and sample size.

Spacings reveal local gaps, clumps and tail stretch that a moment summary may smooth away. Rounding produces zero or repeated gaps, while one large endpoint can dominate a tail spacing. Keep the ordered table, plotting-position convention, expected score gaps and measurement resolution in the analysis record.

Frequently asked questions

What is a normalized spacing?

It is an observed adjacent order-statistic gap divided by the corresponding expected-normal-score gap.

Why use expected normal scores?

They describe where ordered observations should lie under a standard normal pattern.

What plotting positions are used?

A common Chen–Shapiro construction uses (i-0.375)/(n+0.25).

What happens with ties?

Observed zero gaps can reduce the statistic and require matching discrete-data interpretation.

Is the hand statistic a p-value?

No. Critical values depend on n and the exact implementation.

What does the test reveal visually?

A spacing table shows clumps, holes and stretched tails rank by rank.

Primary source and fact-check note

Ling Chen and Samuel S. Shapiro, “An Alternative Test for Normality Based on Normalized Spacings,” Journal of Statistical Computation and Simulation 53(3–4), 269–287 (1995), DOI 10.1080/00949659508811711. Definitions and core claims in this article were checked against this primary publication. Calculations shown here were independently recomputed. Examples are educational illustrations; they are not reported industrial performance results.

Useful next reading

Keep the options open. Mathematics education supports clearer questions, transferable problem-solving skills and informed choices across science, engineering, computing, AI, statistics, design and many other careers. It does not guarantee intelligence, income, admission or employment; those outcomes depend on far more than one subject or model.

Extended laboratory: topic-specific validation and responsible use

Design audit

Freeze independent observations, target residuals and inclusion rules.

Transformation audit

Reproduce ordering, centring, scale and probability calculations.

Statistic audit

Show every gap, moment, component and coefficient.

Calibration audit

Match sample size, parameter fitting, rejection direction and Monte Carlo procedure.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Graph audit

Preserve probability plots, spacing tables and influence displays.

Magnitude audit

Explain the practical consequence in original units.

Sensitivity audit

Use only prespecified formula and influence checks.

Final claim

Keep inference inside the model and sampling design.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Mechanism workshop: calculate, challenge and transfer

Workshop 1: Order

Begin with this mechanism: Sort all observations. Recreate the concrete case: 1,2,3,4,5. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Do not rank within subgroups. Complete the transfer step: Keep one list. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 2: Observed gaps

Begin with this mechanism: Subtract adjacent values. Recreate the concrete case: Every gap equals one. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Ties create zero gaps. Complete the transfer step: Record resolution. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 3: Plot positions

Begin with this mechanism: Compute (i-.375)/(n+.25). Recreate the concrete case: Five probabilities are symmetric. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Other formulas yield different scores. Complete the transfer step: State convention. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 4: Normal scores

Begin with this mechanism: Apply inverse Phi. Recreate the concrete case: Scores≈-1.180,-.497,0,.497,1.180. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Tail quantiles need precision. Complete the transfer step: Use stable routine. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Workshop 5: Expected gaps

Begin with this mechanism: Subtract adjacent scores. Recreate the concrete case: .683,.497,.497,.683. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: All must be positive. Complete the transfer step: Audit order. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 6: First normalized gap

Begin with this mechanism: Divide 1 by .683. Recreate the concrete case: Value≈1.464. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: This is not a probability. Complete the transfer step: Label ratio. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 7: Second gap

Begin with this mechanism: Divide 1 by .497. Recreate the concrete case: Value≈2.012. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Rounding affects sum. Complete the transfer step: Keep precision. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 8: Symmetry

Begin with this mechanism: Check mirrored gaps. Recreate the concrete case: Ratios repeat 1.464,2.012. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: A skewed sample breaks pattern. Complete the transfer step: Plot. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Workshop 9: Aggregate

Begin with this mechanism: Sum four ratios. Recreate the concrete case: Total≈6.952. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: One tail gap can dominate. Complete the transfer step: Show contributions. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 10: Sample scale

Begin with this mechanism: Compute S with declared divisor. Recreate the concrete case: S≈1.581. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Mixing population and sample SD changes result. Complete the transfer step: Use n-1. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 11: Denominator

Begin with this mechanism: Multiply (n-1)S. Recreate the concrete case: Value≈6.324. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: n rather than n-1 is wrong. Complete the transfer step: Check formula. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 12: CS statistic

Begin with this mechanism: Divide aggregate by denominator. Recreate the concrete case: CS≈1.099. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: A number near one is not automatic acceptance. Complete the transfer step: Use critical values. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Workshop 13: Rejection rule

Begin with this mechanism: Follow the implementation’s stated tail. Recreate the concrete case: Small values commonly indicate departure. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Guessing direction reverses inference. Complete the transfer step: Verify source. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 14: Ties

Begin with this mechanism: Inspect grouped or rounded data. Recreate the concrete case: Zero spacings are informative. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Jitter invents precision. Complete the transfer step: Use tie-aware method. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 15: Comparison

Begin with this mechanism: Place CS beside probability plot. Recreate the concrete case: Local gaps explain curvature. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Multiple tests need a plan. Complete the transfer step: Choose purpose. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.

Workshop 16: Claim

Begin with this mechanism: State spacing-based evidence. Recreate the concrete case: The ordered gaps agree or disagree with normal spacing. Write ordered observations, observed gaps, plotting positions, normal scores, expected gaps, normalized spacings, aggregate sums, sample scales, CS statistics and critical values with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: No cause is proven. Complete the transfer step: Check measurement. Change exactly one observation, tie, plotting position, scale divisor, tail probability, spacing, range endpoint, median, absolute deviation, moment, coefficient, rejection direction, null approximation or Monte Carlo seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep design, transformation, statistic, calibration, magnitude, interpretation and validation on separate notebook lines. Explain the result twice: first to a student using the visible ordered values or component table, then to a parent using the real decision and original units. Add a counterexample in which the same statistic arises from skewness, heavy tails, rounding, censoring or dependence. State what extra graph, design record or held-out observation would distinguish the explanations. Finish by writing one sentence the data support, one sentence they do not support and one useful next measurement.