WHY MATHEMATICS?

Why Mathematics? | Instrumental-Variable Quantile Regression: Structural Quantiles, Rank Conditions, Moment Equations and Weak Instruments

How instruments can identify structural quantile effects under explicit assumptions, why a zero moment is not proof, and how weak identification changes uncertainty.

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

Define the structural quantile effect and source of endogeneity
State instrument relevance, exclusion and rank restrictions
Build and solve quantile moment conditions
Use weak-instrument-robust inference and careful interpretation

Chapter map

1. Question
2. Endogeneity
3. Instrument
4. Centred instrument
5. Score
6. Candidate A
7. Moment A
8. Candidate B
9. Moment B
10. Relevance
11. Exclusion
12. Ranks
13. Optimisation
14. Weak-instrument inference
15. Interpretation
16. Claim

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

1. Question

This is where modelling choices begin to affect evidence. Define the treatment, outcome and structural quantile.

A concrete situation

Use tau=.5.

What the mathematics does not license

Changing tau changes the causal estimand. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Record target. 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. Endogeneity

The central mathematical idea is concrete. Explain why treatment can share unobserved causes with outcome.

A concrete situation

Ordinary conditional QR may be inconsistent.

What the mathematics does not license

A rich covariate list does not guarantee exogeneity. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Draw a causal graph. 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. Instrument

Here the geometry becomes operational. Choose z that shifts treatment under a stated design.

A concrete situation

Use z=0,0,1,1.

What the mathematics does not license

A label is not automatically a valid instrument. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Defend relevance and exclusion. 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. Centred instrument

This step deserves a careful pause. Subtract the sample mean .5.

A concrete situation

Get -.5,-.5,.5,.5.

What the mathematics does not license

Centering is part of this teaching moment only. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Keep 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.

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. Score

The useful question is not only how to calculate, but what the calculation assumes. Use psi=.5-I(u<0).

A concrete situation

Negative residual gives -.5; positive gives .5.

What the mathematics does not license

Zero residual needs a declared convention. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

State 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.

6. Candidate A

A small example makes the mechanism visible. Use signs negative, positive, negative, positive.

A concrete situation

Scores alternate -.5,.5,-.5,.5.

What the mathematics does not license

Signs alone omit residual magnitudes. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Label illustration. 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. Moment A

This is where modelling choices begin to affect evidence. Multiply centred z by scores.

A concrete situation

Products sum to zero.

What the mathematics does not license

One zero sample moment is not identification. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Check all moments. 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. Candidate B

The central mathematical idea is concrete. Use signs negative, negative, positive, positive.

A concrete situation

Scores are -.5,-.5,.5,.5.

What the mathematics does not license

A different parameter can sort signs differently. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Compare objective. 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. Moment B

Here the geometry becomes operational. Multiply and add.

A concrete situation

Products sum to one.

What the mathematics does not license

Sample moments fluctuate. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Use uncertainty. 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. Relevance

This step deserves a careful pause. Measure how z changes treatment.

A concrete situation

Weak first stage weakens information.

What the mathematics does not license

Statistical significance alone is insufficient. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Show strength. 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. Exclusion

The useful question is not only how to calculate, but what the calculation assumes. State why z affects outcome only through treatment in the model.

A concrete situation

Use domain evidence.

What the mathematics does not license

Exclusion is not directly testable from one outcome. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Bound claim. 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. Ranks

A small example makes the mechanism visible. Distinguish rank invariance from rank similarity.

A concrete situation

Use the weaker justified condition.

What the mathematics does not license

Rank language can be over-personalised. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Define population. 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. Optimisation

This is where modelling choices begin to affect evidence. Search the structural coefficient and nuisance parameters.

A concrete situation

Profile a quantile-regression criterion.

What the mathematics does not license

Non-convexity and flat regions can appear. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Record solver. 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. Weak-instrument inference

The central mathematical idea is concrete. Use procedures valid under weak identification.

A concrete situation

Confidence sets may be wide or disconnected.

What the mathematics does not license

A narrow Wald interval can mislead. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Stress 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.

15. Interpretation

Here the geometry becomes operational. Translate a structural quantile effect.

A concrete situation

Keep it distinct from individual effect.

What the mathematics does not license

Percentile labels do not track fixed people automatically. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Use careful prose. 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

This step deserves a careful pause. State assumptions, moment, strength and uncertainty.

A concrete situation

Teaching moments are zero and one.

What the mathematics does not license

Do not call the zero moment proof. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Bound conclusion. 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

Instrumental-variable quantile regression asks a causal question at a chosen part of an outcome distribution when the treatment or exposure may be endogenous. Ordinary quantile regression conditions on observed predictors, but a treatment chosen in response to unobserved circumstances can remain correlated with the structural disturbance. Chernozhukov and Hansen show how an instrument, structural quantiles and restrictions on ranks can identify quantile treatment effects under stated assumptions rather than by ordinary residual fitting alone.

At tau=.5, write the score as psi(u)=.5-I(u<0). Let the centred instrument be z minus .5 for z=0,0,1,1. Candidate A has residual signs negative, positive, negative, positive, giving scores -.5,.5,-.5,.5. The products with centred z sum to zero. Candidate B has signs negative, negative, positive, positive; the products sum to one. Candidate A satisfies this one sample orthogonality check, but a zero sample moment is not proof that its structural coefficient is identified or correct.

An instrument must be relevant to treatment and valid for the structural outcome equation under the model. Exclusion, independence and rank conditions are substantive assumptions, not properties that a computer can certify from the observed table. Weak instruments can make an objective nearly flat, estimates unstable and confidence sets wide or disconnected. Rank invariance is stronger than rank similarity, and either restriction needs an interpretation tied to the application.

Report the structural quantile, treatment, instrument, covariates, moment conditions, identification assumptions, first-stage evidence, optimisation method, weak-instrument procedure and uncertainty set. A structural quantile effect is not the same object as an individual treatment effect, an ordinary conditional association or a claim that everyone at the same observed percentile moves by the coefficient.

Frequently asked questions

Why is ordinary quantile regression insufficient?

Endogenous treatment can remain correlated with the structural disturbance.

What makes an instrument useful?

It must shift treatment and satisfy the model’s validity conditions.

What is a structural quantile effect?

A model-based effect on a structural outcome quantile under identification assumptions.

Why do rank conditions matter?

They connect potential outcomes across treatment states strongly enough for identification.

What happens with weak instruments?

Objectives can be flat and confidence sets wide or disconnected.

What must be reported?

Tau, instrument, moments, identification assumptions, strength, solver and robust uncertainty.

Primary source and fact-check note

Victor Chernozhukov and Christian Hansen, “An IV Model of Quantile Treatment Effects,” Econometrica 73(1), 245–261 (2005), DOI 10.1111/j.1468-0262.2005.00570.x. 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 observational unit, treatment or predictors, panel IDs, time order, response orientation, tail direction, missingness and dependence.

Objective audit

Write structural moments, check losses, individual-effect penalties, tail normalisations or multi-parameter likelihoods explicitly.

Assumption audit

List exclusion, relevance, rank, panel, tail-equivalence, response-family and link assumptions that the calculation needs.

Tuning audit

Record tau, penalty, threshold, tail sequence, family, smoothness and solver settings before interpreting output.

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.

Computation audit

Check moment balance, identification rank, optimisation ties, convergence, effective degrees of freedom and numerical sensitivity.

Uncertainty audit

Use weak-identification, cluster, extreme-value, simulation or likelihood methods matched to the actual design.

Validation audit

Use untouched units or future observations for loss, centile coverage, tail calibration and subgroup checks.

Final claim

Translate the conditional result without turning an assumption, percentile, fitted distribution or association into a guarantee.

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: Question

Begin with this mechanism: Define the treatment, outcome and structural quantile. Recreate the concrete case: Use tau=.5. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Changing tau changes the causal estimand. Complete the transfer step: Record target. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 2: Endogeneity

Begin with this mechanism: Explain why treatment can share unobserved causes with outcome. Recreate the concrete case: Ordinary conditional QR may be inconsistent. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: A rich covariate list does not guarantee exogeneity. Complete the transfer step: Draw a causal graph. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 3: Instrument

Begin with this mechanism: Choose z that shifts treatment under a stated design. Recreate the concrete case: Use z=0,0,1,1. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: A label is not automatically a valid instrument. Complete the transfer step: Defend relevance and exclusion. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 4: Centred instrument

Begin with this mechanism: Subtract the sample mean .5. Recreate the concrete case: Get -.5,-.5,.5,.5. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Centering is part of this teaching moment only. Complete the transfer step: Keep formula. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra 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: Score

Begin with this mechanism: Use psi=.5-I(u<0). Recreate the concrete case: Negative residual gives -.5; positive gives .5. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Zero residual needs a declared convention. Complete the transfer step: State rule. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 6: Candidate A

Begin with this mechanism: Use signs negative, positive, negative, positive. Recreate the concrete case: Scores alternate -.5,.5,-.5,.5. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Signs alone omit residual magnitudes. Complete the transfer step: Label illustration. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 7: Moment A

Begin with this mechanism: Multiply centred z by scores. Recreate the concrete case: Products sum to zero. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: One zero sample moment is not identification. Complete the transfer step: Check all moments. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 8: Candidate B

Begin with this mechanism: Use signs negative, negative, positive, positive. Recreate the concrete case: Scores are -.5,-.5,.5,.5. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: A different parameter can sort signs differently. Complete the transfer step: Compare objective. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra 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: Moment B

Begin with this mechanism: Multiply and add. Recreate the concrete case: Products sum to one. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Sample moments fluctuate. Complete the transfer step: Use uncertainty. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 10: Relevance

Begin with this mechanism: Measure how z changes treatment. Recreate the concrete case: Weak first stage weakens information. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Statistical significance alone is insufficient. Complete the transfer step: Show strength. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 11: Exclusion

Begin with this mechanism: State why z affects outcome only through treatment in the model. Recreate the concrete case: Use domain evidence. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Exclusion is not directly testable from one outcome. Complete the transfer step: Bound claim. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 12: Ranks

Begin with this mechanism: Distinguish rank invariance from rank similarity. Recreate the concrete case: Use the weaker justified condition. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Rank language can be over-personalised. Complete the transfer step: Define population. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra 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: Optimisation

Begin with this mechanism: Search the structural coefficient and nuisance parameters. Recreate the concrete case: Profile a quantile-regression criterion. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Non-convexity and flat regions can appear. Complete the transfer step: Record solver. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 14: Weak-instrument inference

Begin with this mechanism: Use procedures valid under weak identification. Recreate the concrete case: Confidence sets may be wide or disconnected. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: A narrow Wald interval can mislead. Complete the transfer step: Stress test. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 15: Interpretation

Begin with this mechanism: Translate a structural quantile effect. Recreate the concrete case: Keep it distinct from individual effect. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Percentile labels do not track fixed people automatically. Complete the transfer step: Use careful prose. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra 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 assumptions, moment, strength and uncertainty. Recreate the concrete case: Teaching moments are zero and one. Write outcomes, treatments, instruments, covariates, quantile levels, structural coefficients, residual signs, score values, moments, first-stage measures and uncertainty sets with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Do not call the zero moment proof. Complete the transfer step: Bound conclusion. Change exactly one observation, treatment assignment, instrument, rank restriction, panel visit, individual effect, penalty, tau, sample size, threshold, tail index, family, link, scale, shape parameter, smoother, solver tolerance, bootstrap rule or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, objective, assumptions, tuning, computation, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving weak identification, informative dropout, sparse tails, family misspecification, subgroup miscalibration, measurement error or high leverage. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.