WHY MATHEMATICS?

Why Mathematics? | Siegel Repeated Median Regression: Nested Medians, 50% Breakdown and Direct Robust Line Fitting

How a median slope is calculated around every point before a second median combines them, and why high breakdown still needs efficiency, leverage and uncertainty checks.

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 all slopes around each anchor point
Take one local median for every observation
Take the median of those local medians
Add a disclosed intercept and full-pipeline validation

Chapter map

1. Model
2. Anchor point
3. Local slopes
4. First median
5. Repeat
6. Remote anchor
7. Second median
8. Intercepts
9. Fitted line
10. Breakdown
11. Efficiency
12. Ties
13. Computation
14. Uncertainty
15. Transfer
16. Claim

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

1. Model

The useful question is not only how to calculate, but what the calculation assumes. Specify bivariate line, units and distinct-x rule.

A concrete situation

Use the five-point example.

What the mathematics does not license

Axis choice changes slopes. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Freeze response. 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. Anchor point

A small example makes the mechanism visible. Choose observation i and pair it with every j≠i.

A concrete situation

Each anchor has four slopes.

What the mathematics does not license

An anchor is not automatically trustworthy. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Keep IDs. 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. Local slopes

This is where modelling choices begin to affect evidence. Compute signed slopes around anchor zero.

A concrete situation

Get 1.1,.95,1.0667,4.75.

What the mathematics does not license

Zero denominators need handling. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Show fractions. 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. First median

The central mathematical idea is concrete. Take the median within anchor zero.

A concrete situation

Local median≈1.0833.

What the mathematics does not license

Even-sample convention matters. The same reasoning helps students move from an exercise to a real dataset.

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.

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. Repeat

Here the geometry becomes operational. Calculate a local median for every anchor.

A concrete situation

Obtain five local summaries.

What the mathematics does not license

Skipping an awkward anchor changes protection. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Complete table. 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. Remote anchor

This step deserves a careful pause. Inspect the fifth local median.

A concrete situation

It is about 7.2583.

What the mathematics does not license

A large local summary is not proof of error. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Investigate context. 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 median

The useful question is not only how to calculate, but what the calculation assumes. Take the median of local medians.

A concrete situation

Slope=1.125.

What the mathematics does not license

Nested medians are not the median of all slopes. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Keep levels separate. 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. Intercepts

A small example makes the mechanism visible. Compute y_i−1.125x_i.

A concrete situation

Take their median.

What the mathematics does not license

Other intercept conventions exist. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

State choice. 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. Fitted line

This is where modelling choices begin to affect evidence. Write y=.975+1.125x.

A concrete situation

Compute predictions in original units.

What the mathematics does not license

The line need not pass through any observed point. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Plot it. 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. Breakdown

The central mathematical idea is concrete. Relate nested medians to near-half contamination resistance.

A concrete situation

One extreme anchor cannot control the second median.

What the mathematics does not license

Finite-sample setup matters. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

State conditions. 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. Efficiency

Here the geometry becomes operational. Compare sampling variability under clean models.

A concrete situation

Use simulation or theory.

What the mathematics does not license

Breakdown alone does not rank estimators. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Report trade-off. 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. Ties

This step deserves a careful pause. Define local and outer medians for even counts and duplicates.

A concrete situation

Keep all equal slopes.

What the mathematics does not license

Silent tie breaking harms reproducibility. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Document rule. 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. Computation

The useful question is not only how to calculate, but what the calculation assumes. Use direct calculation or a verified fast algorithm.

A concrete situation

Store local medians.

What the mathematics does not license

Software defaults may choose another intercept. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Read documentation. 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. Uncertainty

A small example makes the mechanism visible. Bootstrap the full nested-median procedure if justified.

A concrete situation

Resample sampling units.

What the mathematics does not license

Pairwise slopes are not independent. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Mirror design. 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. Transfer

This is where modelling choices begin to affect evidence. Use as a resistant start for later M or MM fitting.

A concrete situation

Compare with Theil–Sen and LMS.

What the mathematics does not license

A start is not the final inferential model. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Separate stages. 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 both median levels and limitations.

A concrete situation

Teaching slope=1.125.

What the mathematics does not license

Do not label minority points false. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Preserve 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

Siegel’s repeated median regression uses two nested medians. For each observation i, calculate every slope joining i to another observation and take that local median. Then take the median of those n local medians. The first median prevents one remote partner from controlling a point’s local slope; the second prevents a minority of contaminated points from controlling the final slope.

For the five-point example, the local median slopes are approximately 1.0833,1.0750,1.1250,1.1833 and 7.2583. Their median is 1.125. With the disclosed intercept convention median(y_i−1.125x_i), the intercept is .975, giving y=.975+1.125x. The large local slope at the final point remains visible, yet it is only one of five local summaries.

Siegel showed that the repeated median can maintain a 50% breakdown value in the bivariate linear-regression setting described in the paper. That resistance is stronger than the usual Theil–Sen slope breakdown, but breakdown is only one property. Efficiency, discrete ties, predictor configuration and uncertainty still matter. A high-breakdown line is not automatically the best predictive model for every clean dataset.

The estimator is direct rather than an iterative reweighting procedure, making it useful as a robust starting value. Its pairwise work is computationally heavier than ordinary least squares, although faster algorithms exist. Repeated x-values, duplicate points and even-number medians require explicit conventions. Keep local medians so a reader can see whether one region of predictor space behaves differently.

Frequently asked questions

What is repeated?

A median is taken within each anchor point’s slopes, then again across anchors.

What is the teaching slope?

1.125 for the five-point example.

Why is breakdown high?

Nested medians prevent a minority of anchors and partners from controlling the result.

Is it iterative?

The defining estimator is direct, though fast algorithms may implement it.

How are ties handled?

By explicit local and outer median conventions.

What must be reported?

Eligible pairs, local medians, outer median, intercept, duplicates and uncertainty method.

Primary source and fact-check note

Andrew F. Siegel, “Robust Regression Using Repeated Medians,” Biometrika 69(1), 242–244 (1982), DOI 10.1093/biomet/69.1.242. 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 sampling units, response, predictor orientation, dependence, ties, duplicates and missingness.

Calculation audit

Retain pair IDs, local medians, residual signs, cut counts, scores, weights and coefficient updates.

Robustness audit

State the exact breakdown, depth or influence property and the assumptions supporting it.

Solver audit

Record direct algorithm or IRLS start, scale, tuning, candidate set, tolerance and stopping criterion.

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.

Leverage audit

Inspect predictor-space geometry separately from vertical residual magnitude.

Uncertainty audit

Use rank intervals, depth regions, robust covariance or resampling matched to the full estimator.

Sensitivity audit

Change one tie rule, intercept convention, candidate line, scale, tuning constant or leverage point at a time.

Final claim

Translate the line into original units while preserving minority observations and causal uncertainty.

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: Model

Begin with this mechanism: Specify bivariate line, units and distinct-x rule. Recreate the concrete case: Use the five-point example. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Axis choice changes slopes. Complete the transfer step: Freeze response. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Anchor point

Begin with this mechanism: Choose observation i and pair it with every j≠i. Recreate the concrete case: Each anchor has four slopes. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: An anchor is not automatically trustworthy. Complete the transfer step: Keep IDs. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Local slopes

Begin with this mechanism: Compute signed slopes around anchor zero. Recreate the concrete case: Get 1.1,.95,1.0667,4.75. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Zero denominators need handling. Complete the transfer step: Show fractions. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: First median

Begin with this mechanism: Take the median within anchor zero. Recreate the concrete case: Local median≈1.0833. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Even-sample convention matters. Complete the transfer step: Keep precision. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Repeat

Begin with this mechanism: Calculate a local median for every anchor. Recreate the concrete case: Obtain five local summaries. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Skipping an awkward anchor changes protection. Complete the transfer step: Complete table. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Remote anchor

Begin with this mechanism: Inspect the fifth local median. Recreate the concrete case: It is about 7.2583. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: A large local summary is not proof of error. Complete the transfer step: Investigate context. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Second median

Begin with this mechanism: Take the median of local medians. Recreate the concrete case: Slope=1.125. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Nested medians are not the median of all slopes. Complete the transfer step: Keep levels separate. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Intercepts

Begin with this mechanism: Compute y_i−1.125x_i. Recreate the concrete case: Take their median. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Other intercept conventions exist. Complete the transfer step: State choice. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Fitted line

Begin with this mechanism: Write y=.975+1.125x. Recreate the concrete case: Compute predictions in original units. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: The line need not pass through any observed point. Complete the transfer step: Plot it. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Breakdown

Begin with this mechanism: Relate nested medians to near-half contamination resistance. Recreate the concrete case: One extreme anchor cannot control the second median. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Finite-sample setup matters. Complete the transfer step: State conditions. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Efficiency

Begin with this mechanism: Compare sampling variability under clean models. Recreate the concrete case: Use simulation or theory. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Breakdown alone does not rank estimators. Complete the transfer step: Report trade-off. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Ties

Begin with this mechanism: Define local and outer medians for even counts and duplicates. Recreate the concrete case: Keep all equal slopes. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Silent tie breaking harms reproducibility. Complete the transfer step: Document rule. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Computation

Begin with this mechanism: Use direct calculation or a verified fast algorithm. Recreate the concrete case: Store local medians. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Software defaults may choose another intercept. Complete the transfer step: Read documentation. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Uncertainty

Begin with this mechanism: Bootstrap the full nested-median procedure if justified. Recreate the concrete case: Resample sampling units. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Pairwise slopes are not independent. Complete the transfer step: Mirror design. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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: Transfer

Begin with this mechanism: Use as a resistant start for later M or MM fitting. Recreate the concrete case: Compare with Theil–Sen and LMS. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: A start is not the final inferential model. Complete the transfer step: Separate stages. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. 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 both median levels and limitations. Recreate the concrete case: Teaching slope=1.125. Write predictors, responses, anchor-specific slopes, local medians, outer medians, intercept residuals, coefficients, uncertainty and original measurement units with units, signs, pair or point identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Do not label minority points false. Complete the transfer step: Preserve provenance. Change exactly one observation, predictor, response, pair rule, median convention, intercept rule, tie handling, candidate line, residual sign, cut position, robust scale, tuning constant, starting coefficient, leverage value, tolerance, interval or covariance formula. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, calculation, geometry, robustness property, optimisation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the full table and once to a parent with the decision in original units. Add a counterexample involving rounding, repeated x-values, dependence, a step change, a real minority group, high leverage or measurement error. State what extra graph or measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.