WHY MATHEMATICS?

Why Mathematics? | Desgagné–Lafaye de Micheaux Rao Score Test: Second-Power Skewness, Second-Power Kurtosis and Asymmetric Power Distributions

How lower-power shape summaries reduce extreme-value amplification, how the asymmetric power family gives them a score-test foundation, and how the modified statistic improves finite-sample use.

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

Calculate signed-square and absolute-deviation summaries
Standardize second-power skewness and kurtosis inputs
Build the Rao score vector inside the APD family
Choose asymptotic, modified or directional calibration before results

Chapter map

1. Sample
2. Mean
3. Deviations
4. Variance
5. Signed squares
6. Second-power skew numerator
7. Skew ratio
8. Absolute deviations
9. Absolute mean
10. Scale ratio
11. Normal reference
12. Score vector
13. Quadratic score
14. Modified version
15. Directional option
16. Claim

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

1. Sample

The useful question is not only how to calculate, but what the calculation assumes. Freeze independent observations.

A concrete situation

Use 1,2,3,4,8.

What the mathematics does not license

Five points are arithmetic only. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Label scope. 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. Mean

A small example makes the mechanism visible. Calculate xbar.

A concrete situation

Mean=3.6.

What the mathematics does not license

The mean is not robust. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Show influence. 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. Deviations

This is where modelling choices begin to affect evidence. Subtract the mean.

A concrete situation

-2.6,-1.6,-.6,.4,4.4.

What the mathematics does not license

Signs matter. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Keep 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.

4. Variance

The central mathematical idea is concrete. Average squared deviations.

A concrete situation

m2=5.84.

What the mathematics does not license

Divisor conventions change ratios. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Match paper. 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. Signed squares

Here the geometry becomes operational. Compute d|d|.

A concrete situation

Values sum to 9.84.

What the mathematics does not license

This differs from d squared. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Preserve sign. 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. Second-power skew numerator

This step deserves a careful pause. Average signed squares.

A concrete situation

Value=1.968.

What the mathematics does not license

Cancellation can hide mixtures. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Plot data. 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. Skew ratio

The useful question is not only how to calculate, but what the calculation assumes. Divide by m2.

A concrete situation

Raw value≈.337.

What the mathematics does not license

It is not final score. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Label component. 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. Absolute deviations

A small example makes the mechanism visible. Compute |d|.

A concrete situation

They sum to 9.6.

What the mathematics does not license

Large values enter linearly. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Show shares. 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. Absolute mean

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

A concrete situation

Value=1.92.

What the mathematics does not license

Missingness changes n. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Freeze cohort. 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. Scale ratio

The central mathematical idea is concrete. Divide by sqrt(m2).

A concrete situation

Raw value≈.795.

What the mathematics does not license

Reference depends on definition. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Name measure. 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. Normal reference

Here the geometry becomes operational. Compare with sqrt(2/pi)≈.7979 where appropriate.

A concrete situation

The raw ratio is close.

What the mathematics does not license

One matching component is insufficient. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Keep skewness. 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. Score vector

This step deserves a careful pause. Apply published APD score formulas.

A concrete situation

Obtain asymmetry and tail coordinates.

What the mathematics does not license

Raw summaries are not final statistic. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Use validated code. 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. Quadratic score

The useful question is not only how to calculate, but what the calculation assumes. Combine with information matrix.

A concrete situation

Form Rao statistic.

What the mathematics does not license

Dropping covariance changes calibration. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Document matrix. 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. Modified version

A small example makes the mechanism visible. Name the finite-sample adjustment if used.

A concrete situation

Use its chi-square approximation.

What the mathematics does not license

Mixing versions invalidates p-value. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Report version. 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. Directional option

This is where modelling choices begin to affect evidence. Use second-power kurtosis alone only when symmetry is justified.

A concrete situation

Predefine symmetric case.

What the mathematics does not license

Post-hoc choice inflates error. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Plan test. 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 central mathematical idea is concrete. State second-power shape evidence.

A concrete situation

Asymmetry or tail power conflicts with normality.

What the mathematics does not license

Mechanism is not proven. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Inspect provenance. 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

Desgagné and Lafaye de Micheaux replace classical third-power skewness and fourth-power kurtosis with second-power shape measures. The lower powers reduce the explosive influence of extreme deviations while keeping an interpretable separation between asymmetry and tail shape. Their test is also a Rao score test inside an asymmetric power-distribution family.

For 1,2,3,4,8, the mean is 3.6 and the n-divisor variance is 5.84. The average signed squared deviation, mean[(x-xbar)|x-xbar|], is 1.968, so the raw second-power skewness is about .337 after division by 5.84. The mean absolute deviation about the mean is 1.92, giving a raw scale ratio about .795 after division by sqrt(5.84).

Those raw summaries feed the published score construction; they are not by themselves the final chi-square test. The paper derives an asymptotic statistic and a modified version whose finite-sample distribution is closely approximated by chi-square. An implementation must name which version it reports.

The asymmetric power family supplies direction: one component responds to unequal left and right shape, while another responds to exponential-power tail behaviour. A significant result still does not decide whether the cause is contamination, mixture structure, transformation error or a genuinely nonnormal process.

Frequently asked questions

What is second-power skewness?

It uses signed squared deviations rather than cubed deviations.

Why use lower powers?

They reduce how rapidly a single extreme value grows in the shape summary.

What family supports the score test?

The asymmetric power distribution family.

What is the null reference?

The joint score statistic is asymptotically chi-square under its conditions.

What does the modified version do?

It improves the finite-sample chi-square approximation.

When is the directional test appropriate?

Only when symmetry is justified before seeing the data.

Primary source and fact-check note

Alain Desgagné and Pierre Lafaye de Micheaux, “A Powerful and Interpretable Alternative to the Jarque–Bera Test of Normality Based on 2nd-Power Skewness and Kurtosis, Using the Rao’s Score Test on the APD Family,” Journal of Applied Statistics 45(13), 2307–2327 (2018), DOI 10.1080/02664763.2017.1415311. 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 observations, target residuals, dependence and inclusion rules.

Transformation audit

Reproduce ordering, leave-one-out scales, kernels, powers or probability weights.

Statistic audit

Show every coordinate, reference value, covariance term and coefficient.

Calibration audit

Match sample size, parameter estimation, dependence and statistic version.

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 rank plots, component maps, influence displays and Q–Q views.

Magnitude audit

Explain consequences in original units.

Sensitivity audit

Use only prespecified validation and resolution checks.

Final claim

Keep the conclusion inside the measured population and model.

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: Sample

Begin with this mechanism: Freeze independent observations. Recreate the concrete case: Use 1,2,3,4,8. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Five points are arithmetic only. Complete the transfer step: Label scope. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 2: Mean

Begin with this mechanism: Calculate xbar. Recreate the concrete case: Mean=3.6. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: The mean is not robust. Complete the transfer step: Show influence. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 3: Deviations

Begin with this mechanism: Subtract the mean. Recreate the concrete case: -2.6,-1.6,-.6,.4,4.4. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Signs matter. Complete the transfer step: Keep list. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 4: Variance

Begin with this mechanism: Average squared deviations. Recreate the concrete case: m2=5.84. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Divisor conventions change ratios. Complete the transfer step: Match paper. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence 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: Signed squares

Begin with this mechanism: Compute d|d|. Recreate the concrete case: Values sum to 9.84. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: This differs from d squared. Complete the transfer step: Preserve sign. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 6: Second-power skew numerator

Begin with this mechanism: Average signed squares. Recreate the concrete case: Value=1.968. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Cancellation can hide mixtures. Complete the transfer step: Plot data. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 7: Skew ratio

Begin with this mechanism: Divide by m2. Recreate the concrete case: Raw value≈.337. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: It is not final score. Complete the transfer step: Label component. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 8: Absolute deviations

Begin with this mechanism: Compute |d|. Recreate the concrete case: They sum to 9.6. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Large values enter linearly. Complete the transfer step: Show shares. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence 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: Absolute mean

Begin with this mechanism: Divide by n. Recreate the concrete case: Value=1.92. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Missingness changes n. Complete the transfer step: Freeze cohort. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 10: Scale ratio

Begin with this mechanism: Divide by sqrt(m2). Recreate the concrete case: Raw value≈.795. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Reference depends on definition. Complete the transfer step: Name measure. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 11: Normal reference

Begin with this mechanism: Compare with sqrt(2/pi)≈.7979 where appropriate. Recreate the concrete case: The raw ratio is close. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: One matching component is insufficient. Complete the transfer step: Keep skewness. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 12: Score vector

Begin with this mechanism: Apply published APD score formulas. Recreate the concrete case: Obtain asymmetry and tail coordinates. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Raw summaries are not final statistic. Complete the transfer step: Use validated code. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence 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: Quadratic score

Begin with this mechanism: Combine with information matrix. Recreate the concrete case: Form Rao statistic. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Dropping covariance changes calibration. Complete the transfer step: Document matrix. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 14: Modified version

Begin with this mechanism: Name the finite-sample adjustment if used. Recreate the concrete case: Use its chi-square approximation. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Mixing versions invalidates p-value. Complete the transfer step: Report version. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 15: Directional option

Begin with this mechanism: Use second-power kurtosis alone only when symmetry is justified. Recreate the concrete case: Predefine symmetric case. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Post-hoc choice inflates error. Complete the transfer step: Plan test. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 16: Claim

Begin with this mechanism: State second-power shape evidence. Recreate the concrete case: Asymmetry or tail power conflicts with normality. Write deviations, variances, signed squares, absolute deviations, second-power ratios, score coordinates, information matrices and calibrated statistics with units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Mechanism is not proven. Complete the transfer step: Inspect provenance. Change exactly one observation, leave-one-out index, median convention, kernel value, moment power, order-statistic weight, covariance entry, rejection direction, finite-sample correction 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 once to a student with the component table and once to a parent with the real decision and original units. Add a counterexample in which the same statistic comes from rounding, censoring, mixture structure, dependence or contamination. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.