Choose your reading route
eduKateSG · Why Mathematics? · Hall–Wellner Confidence Bands for Reliability Curves: Simultaneous Coverage, Brownian Bridges, Time Ranges and Interpretation
Start with the question, audit the worked example, or jump to the assumptions and practice. Every chapter stays open for uninterrupted reading.
Understand the question
Define the quantity and follow the mechanism.Route 2 · Chapters 6–10
Reproduce the example
Audit the numbers one step at a time.Route 3 · Chapters 11–15
Check assumptions
See where design and calibration matter.Route 4 · Chapters 16–20
Explore applications
Connect the method to a concrete investigation.Route 5 · Chapters 21–25
Study and verify
Practise, inspect failure cases and read further.
Full chapter index · Practice and checking · Mathematics Learning Hub
Why mathematics is important becomes practical when a reliability claim concerns a whole time range rather than one preselected hour. Hall–Wellner confidence bands use the large-sample Gaussian-process behaviour of the Kaplan–Meier estimator, transform it toward a Brownian bridge, and choose a critical boundary for simultaneous coverage across a declared interval. The band answers a different question from many separate pointwise intervals.
This guide develops simultaneous coverage, variance time, Brownian bridges, endpoint choice, transformations, equal-precision alternatives and interpretation. A band can be wide, especially where follow-up is weak. Its coverage is tied to the stated range and assumptions; extending the graph after analysis or ignoring tiny risk sets changes the claim.
Full chapter index
Chapters 1–5 · Understand the question
Chapters 6–10 · Reproduce the example
Chapters 11–15 · Check assumptions
Chapters 16–20 · Explore applications
Chapters 21–25 · Study and verify
CHAPTER 1 OF 25 · Understand the question
1. Why pointwise intervals are not a curve band
A 95% interval at each time does not imply a 95% chance that every interval covers simultaneously. Scanning 30 milestones creates many opportunities for at least one miss. This is the central mathematical mechanism in why pointwise intervals are not a curve band. Write the quantities and assumptions before using software, then reproduce the calculation or geometric argument in a small example. Pointwise intervals answer local questions. Choose local or whole-range inference. The useful habit is to keep the physical question, the probability model and the numerical method in separate columns so that a correct computation is not mistaken for complete evidence.
Did You Know? A result can be numerically stable and still answer the wrong question if the failure event, denominator or transformation changes. Test one nearby case and predict the direction of change before recalculating. Save inputs, intermediate values and units so a second reader can audit the path.
For practice, turn the example into a sensitivity check. Change one defensible input while holding the others fixed, compare the new result with the original and explain why it moved. Then construct a counterexample in which the shortcut fails. End with two sentences: what the mathematics supports, and what remains uncertain.
Start with the operational question behind simultaneous coverage target. A 1-alpha band aims to contain the true survival curve for every t in a declared interval [a,b] with approximate probability 1-alpha. In the illustrative case, A 95% band over 0–1000 hours supports one whole-range statement. Draw the relevant graph, density, boundary or timeline before calculating; the picture should use the same variables and sign convention as the equations. Coverage outside the interval is not promised. Print [a,b] in the caption.
A student should be able to rebuild the result without hidden settings. Record the sampling rule or optimisation criterion, the exact failure definition, and the evidence used for every distribution or parameter. Compare a transparent baseline with the advanced method. If they disagree, investigate the mechanism before deciding which answer is more credible.
This chapter belongs to a larger decision, so translate the number carefully. State one action the evidence permits and one action it does not. A parent or educator can ask which school ideas are doing the work: algebra organises relationships, probability represents uncertainty, calculus describes local change, and computing repeats the documented rules at scale.
Treat kaplan–meier centre curve as a model-checking problem rather than a vocabulary exercise. Hall–Wellner bands are centred on the product-limit survival estimator for right-censored data. The staircase remains the central estimate. Recompute the example from the raw values and label every unit. A confidence band cannot correct a biased centre from informative censoring. Validate the observation process first.
The strongest check uses evidence the model did not choose for itself. Hold out later observations, an independent simulation set or a known-answer benchmark. Report errors where decisions are sensitive, especially near the failure boundary, instead of relying on one global accuracy score. A method that performs well on easy safe cases can still fail at the rare cases that motivated the analysis.
Now change one assumption and make a directional prediction before running the calculation. If the result moves the opposite way, inspect the sign convention, transformation, density ratio or constraint. This prediction-first routine develops transferable problem-solving skills because it connects symbolic work to a testable expectation.
The mathematics of gaussian-process limit becomes useful only when its mechanism is explicit. After scaling by sqrt(n), the estimation error across time converges to a correlated Gaussian process under regularity conditions. Consider the worked illustration: Errors at 500 and 600 hours are related because they share earlier events. Preserve the original inputs, calculate step by step and compare with a second method whenever an analytic or high-budget reference exists. Treating time points as independent is wrong. Preserve curve covariance.
There are three uncertainties to keep apart. Sampling or numerical uncertainty comes from finite computation. Parameter uncertainty comes from estimating inputs. Model-form uncertainty comes from simplified physics, distributions or dependence. Reducing the first does not automatically reduce the other two. A credible report names all three and shows which one drives the decision.
For a practical investigation, create a small reproducible notebook or spreadsheet with a frozen data sheet, formula sheet, diagnostics sheet and decision note. Another student should be able to change one input and obtain the same update without guessing what the model meant.
Primary source: https://doi.org/10.1093/biomet/67.1.133. Read it for the method and its assumptions; do not transfer its performance claims automatically to another model.
Use brownian bridge idea to practise sensitivity analysis. A time and space transformation maps the limiting error process to a Brownian bridge over a variance-time interval. The bridge is pinned by its construction and has known boundary distributions. Before accepting the result, identify the quantity that would change the conclusion most. This is an asymptotic mathematical representation, not physical degradation. Separate stochastic-process tool from mechanism.
A simple baseline is essential. It reveals whether the advanced method is improving accuracy or only adding complexity. Compare model calls, wall time, uncertainty and failure-mode coverage under one declared budget. Repeat stochastic calculations with several seeds; a favourable single run is not evidence of dependable efficiency.
Students can extend the example by adding correlation, a second failure mode or a modest measurement error. Explain which part of the formula changes and which remains fixed. This exercise shows how mathematics transfers from a clean textbook case to engineering and computing without pretending that the classroom model is a certified design tool.
Build a small counterexample for variance-time transformation. Greenwood-type cumulative variance determines where each reliability time lies on the transformed scale. The initial example is: Sparse late risk sets move rapidly through variance time. Then create a nearby case where the usual shortcut becomes unreliable. Calendar spacing and information spacing differ. Plot numbers at risk beside time.
Counterexamples are positive learning tools. They reveal the conditions that make a method work and help students recognise when a software output deserves a pause. Keep the counterexample small enough to plot or enumerate. Check sign, support, bounds and conservation of probability before adding complexity.
When writing the conclusion, separate a documented fact from an interpretation and from an illustrative scenario. Cite the primary method paper for the factual mechanism, use the local calculation for the illustration, and label any recommendation as conditional on the stated model. This keeps career and engineering advice age-appropriate and honest.
Connect hall–wellner band shape to a real decision while protecting the boundary between calculation and judgement. The band width depends on survival level, cumulative variance and a simultaneous critical value. Widths adapt over follow-up rather than staying constant. First verify the arithmetic, then translate it into an allowed action and an action the evidence does not support. A single critical value does not imply equal absolute width. Inspect the formula and endpoints.
The audit should include the original question, input provenance, failure convention, algorithm settings, convergence or stopping rule, and a sensitivity result. For an adaptive method, also preserve the sequence of choices. For an optimisation, preserve every constraint. For Bayesian updating, preserve the prior and observation model. These records turn a result into something another person can challenge and improve.
A final transfer question is useful: where else would the same mathematical idea appear? Density ratios appear in survey correction and machine learning, probabilistic constraints in planning, and conditional probability in medical testing. The context changes, but transparent assumptions and reproducible checks remain valuable.
If the band multiplier is k=1.3, n=100 and the scale factor is .4, the half-width is about 1.3·.4/10=.052. The illustration isolates the square-root-n scaling. This is the central mathematical mechanism in a simplified width calculation. Write the quantities and assumptions before using software, then reproduce the calculation or geometric argument in a small example. Real Hall–Wellner scale factors depend on the variance transformation. Use software only after verifying inputs. The useful habit is to keep the physical question, the probability model and the numerical method in separate columns so that a correct computation is not mistaken for complete evidence.
Did You Know? A result can be numerically stable and still answer the wrong question if the failure event, denominator or transformation changes. Test one nearby case and predict the direction of change before recalculating. Save inputs, intermediate values and units so a second reader can audit the path.
For practice, turn the example into a sensitivity check. Change one defensible input while holding the others fixed, compare the new result with the original and explain why it moved. Then construct a counterexample in which the shortcut fails. End with two sentences: what the mathematics supports, and what remains uncertain.
Start with the operational question behind choosing the lower endpoint. Hall–Wellner construction can often include time zero, where survival is one, under its standard formulation. In the illustrative case, Set a=0 for a mission beginning at installation if appropriate. Draw the relevant graph, density, boundary or timeline before calculating; the picture should use the same variables and sign convention as the equations. An early endpoint before any meaningful exposure may add little. Tie a to the decision.
A student should be able to rebuild the result without hidden settings. Record the sampling rule or optimisation criterion, the exact failure definition, and the evidence used for every distribution or parameter. Compare a transparent baseline with the advanced method. If they disagree, investigate the mechanism before deciding which answer is more credible.
This chapter belongs to a larger decision, so translate the number carefully. State one action the evidence permits and one action it does not. A parent or educator can ask which school ideas are doing the work: algebra organises relationships, probability represents uncertainty, calculus describes local change, and computing repeats the documented rules at scale.
Treat choosing the upper endpoint as a model-checking problem rather than a vocabulary exercise. Select b before inspecting favourable curve regions and keep it below unsupported late follow-up. Use 1000 hours if both designs retain adequate risk sets there. Recompute the example from the raw values and label every unit. Choosing b after seeing results changes error control. Pre-specify b and a risk threshold.
The strongest check uses evidence the model did not choose for itself. Hold out later observations, an independent simulation set or a known-answer benchmark. Report errors where decisions are sensitive, especially near the failure boundary, instead of relying on one global accuracy score. A method that performs well on easy safe cases can still fail at the rare cases that motivated the analysis.
Now change one assumption and make a directional prediction before running the calculation. If the result moves the opposite way, inspect the sign convention, transformation, density ratio or constraint. This prediction-first routine develops transferable problem-solving skills because it connects symbolic work to a testable expectation.
The mathematics of critical values becomes useful only when its mechanism is explicit. The simultaneous critical constant comes from the distribution of a transformed Brownian bridge, not an ordinary 1.96 pointwise cutoff. Consider the worked illustration: A whole-range band needs a larger boundary than a local interval. Preserve the original inputs, calculate step by step and compare with a second method whenever an analytic or high-budget reference exists. Using 1.96 everywhere usually undercovers the curve. Document the critical-value source.
There are three uncertainties to keep apart. Sampling or numerical uncertainty comes from finite computation. Parameter uncertainty comes from estimating inputs. Model-form uncertainty comes from simplified physics, distributions or dependence. Reducing the first does not automatically reduce the other two. A credible report names all three and shows which one drives the decision.
For a practical investigation, create a small reproducible notebook or spreadsheet with a frozen data sheet, formula sheet, diagnostics sheet and decision note. Another student should be able to change one input and obtain the same update without guessing what the model meant.
Use transformation for bounded bands to practise sensitivity analysis. Log-minus-log or related transformations can keep back-transformed limits in [0,1] and improve behaviour. Asymmetric bounds are natural near survival one. Before accepting the result, identify the quantity that would change the conclusion most. Back-transformation order can reverse. Test monotonicity with known values.
A simple baseline is essential. It reveals whether the advanced method is improving accuracy or only adding complexity. Compare model calls, wall time, uncertainty and failure-mode coverage under one declared budget. Repeat stochastic calculations with several seeds; a favourable single run is not evidence of dependable efficiency.
Students can extend the example by adding correlation, a second failure mode or a modest measurement error. Explain which part of the formula changes and which remains fixed. This exercise shows how mathematics transfers from a clean textbook case to engineering and computing without pretending that the classroom model is a certified design tool.
A second primary source is https://doi.org/10.1080/00401706.1984.10487964. Compare the problem definition, estimator and validation rather than quoting one result in isolation.
Build a small counterexample for equal-precision bands. Nair’s equal-precision bands scale width with the pointwise standard deviation and use a different simultaneous critical value. The initial example is: They can offer narrower relative precision in selected variance-time ranges. Then create a nearby case where the usual shortcut becomes unreliable. They may exclude endpoints where variance-time restrictions fail. Compare bands on the declared interval.
Counterexamples are positive learning tools. They reveal the conditions that make a method work and help students recognise when a software output deserves a pause. Keep the counterexample small enough to plot or enumerate. Check sign, support, bounds and conservation of probability before adding complexity.
When writing the conclusion, separate a documented fact from an interpretation and from an illustrative scenario. Cite the primary method paper for the factual mechanism, use the local calculation for the illustration, and label any recommendation as conditional on the stated model. This keeps career and engineering advice age-appropriate and honest.
Connect hall–wellner versus equal precision to a real decision while protecting the boundary between calculation and judgement. The methods allocate width differently across follow-up while targeting simultaneous coverage. One may be narrower early and another in the middle. First verify the arithmetic, then translate it into an allowed action and an action the evidence does not support. Choosing the narrowest after seeing data is selection. Set the comparison rule in advance.
The audit should include the original question, input provenance, failure convention, algorithm settings, convergence or stopping rule, and a sensitivity result. For an adaptive method, also preserve the sequence of choices. For an optimisation, preserve every constraint. For Bayesian updating, preserve the prior and observation model. These records turn a result into something another person can challenge and improve.
A final transfer question is useful: where else would the same mathematical idea appear? Density ratios appear in survey correction and machine learning, probabilistic constraints in planning, and conditional probability in medical testing. The context changes, but transparent assumptions and reproducible checks remain valuable.
Heavy censoring shrinks later risk sets, inflates cumulative variance and widens or destabilises bands. A fleet with many contract endings before 1000 hours has weak late support. This is the central mathematical mechanism in effect of censoring. Write the quantities and assumptions before using software, then reproduce the calculation or geometric argument in a small example. Band machinery does not make missing follow-up reappear. Report censoring and risk counts. The useful habit is to keep the physical question, the probability model and the numerical method in separate columns so that a correct computation is not mistaken for complete evidence.
Did You Know? A result can be numerically stable and still answer the wrong question if the failure event, denominator or transformation changes. Test one nearby case and predict the direction of change before recalculating. Save inputs, intermediate values and units so a second reader can audit the path.
For practice, turn the example into a sensitivity check. Change one defensible input while holding the others fixed, compare the new result with the original and explain why it moved. Then construct a counterexample in which the shortcut fails. End with two sentences: what the mathematics supports, and what remains uncertain.
Start with the operational question behind effect of tied events. Tied failures enter the Kaplan–Meier and variance calculations through event counts at shared times. In the illustrative case, Coarse hourly records can create many ties. Draw the relevant graph, density, boundary or timeline before calculating; the picture should use the same variables and sign convention as the equations. Approximations may be sensitive with very heavy ties. Preserve measurement resolution.
A student should be able to rebuild the result without hidden settings. Record the sampling rule or optimisation criterion, the exact failure definition, and the evidence used for every distribution or parameter. Compare a transparent baseline with the advanced method. If they disagree, investigate the mechanism before deciding which answer is more credible.
This chapter belongs to a larger decision, so translate the number carefully. State one action the evidence permits and one action it does not. A parent or educator can ask which school ideas are doing the work: algebra organises relationships, probability represents uncertainty, calculus describes local change, and computing repeats the documented rules at scale.
Treat comparing two curves as a model-checking problem rather than a vocabulary exercise. Separate bands are not automatically a valid simultaneous band for their difference. Overlapping group bands do not provide a formal equality test. Recompute the example from the raw values and label every unit. Visual overlap rules can mislead. Use a direct difference band or planned test.
The strongest check uses evidence the model did not choose for itself. Hold out later observations, an independent simulation set or a known-answer benchmark. Report errors where decisions are sensitive, especially near the failure boundary, instead of relying on one global accuracy score. A method that performs well on easy safe cases can still fail at the rare cases that motivated the analysis.
Now change one assumption and make a directional prediction before running the calculation. If the result moves the opposite way, inspect the sign convention, transformation, density ratio or constraint. This prediction-first routine develops transferable problem-solving skills because it connects symbolic work to a testable expectation.
The mathematics of band inversion for requirements becomes useful only when its mechanism is explicit. A lower simultaneous band can assess whether survival stays above a requirement throughout a range. Consider the worked illustration: Check whether the lower band remains above .90 from 0 to 800 hours. Preserve the original inputs, calculate step by step and compare with a second method whenever an analytic or high-budget reference exists. This is stronger than a one-time milestone claim. Predefine the requirement and range.
There are three uncertainties to keep apart. Sampling or numerical uncertainty comes from finite computation. Parameter uncertainty comes from estimating inputs. Model-form uncertainty comes from simplified physics, distributions or dependence. Reducing the first does not automatically reduce the other two. A credible report names all three and shows which one drives the decision.
For a practical investigation, create a small reproducible notebook or spreadsheet with a frozen data sheet, formula sheet, diagnostics sheet and decision note. Another student should be able to change one input and obtain the same update without guessing what the model meant.
Use bootstrap bands to practise sensitivity analysis. Resampling whole systems can estimate simultaneous critical values from maximum curve deviations. Use the maximum standardised error across the planned grid in each bootstrap sample. Before accepting the result, identify the quantity that would change the conclusion most. Bootstrap validity still depends on design and censoring. Match resampling to clusters.
A simple baseline is essential. It reveals whether the advanced method is improving accuracy or only adding complexity. Compare model calls, wall time, uncertainty and failure-mode coverage under one declared budget. Repeat stochastic calculations with several seeds; a favourable single run is not evidence of dependable efficiency.
Students can extend the example by adding correlation, a second failure mode or a modest measurement error. Explain which part of the formula changes and which remains fixed. This exercise shows how mathematics transfers from a clean textbook case to engineering and computing without pretending that the classroom model is a certified design tool.
Build a small counterexample for small-sample limitations. Asymptotic bands can miscover with few events, extreme censoring or a short variance-time range. The initial example is: Ten failures cannot support a detailed late band. Then create a nearby case where the usual shortcut becomes unreliable. A valid-looking plot may rely on weak approximation. Use simulation or bootstrap checks.
Counterexamples are positive learning tools. They reveal the conditions that make a method work and help students recognise when a software output deserves a pause. Keep the counterexample small enough to plot or enumerate. Check sign, support, bounds and conservation of probability before adding complexity.
When writing the conclusion, separate a documented fact from an interpretation and from an illustrative scenario. Cite the primary method paper for the factual mechanism, use the local calculation for the illustration, and label any recommendation as conditional on the stated model. This keeps career and engineering advice age-appropriate and honest.
Connect decision interpretation to a real decision while protecting the boundary between calculation and judgement. A whole-range lower bound can support mission-window assurance when combined with consequences and requirements. A band dipping below threshold identifies where evidence is weak. First verify the arithmetic, then translate it into an allowed action and an action the evidence does not support. The band does not certify safety by itself. Link inference to an external standard.
The audit should include the original question, input provenance, failure convention, algorithm settings, convergence or stopping rule, and a sensitivity result. For an adaptive method, also preserve the sequence of choices. For an optimisation, preserve every constraint. For Bayesian updating, preserve the prior and observation model. These records turn a result into something another person can challenge and improve.
A final transfer question is useful: where else would the same mathematical idea appear? Density ratios appear in survey correction and machine learning, probabilistic constraints in planning, and conditional probability in medical testing. The context changes, but transparent assumptions and reproducible checks remain valuable.
Construct pointwise intervals at five times, then simulate why all-five coverage is below the nominal level and compare a simultaneous band. Change b and predict the critical-width effect. This is the central mathematical mechanism in practice activity. Write the quantities and assumptions before using software, then reproduce the calculation or geometric argument in a small example. Toy simulations are not certification. Keep the range fixed. The useful habit is to keep the physical question, the probability model and the numerical method in separate columns so that a correct computation is not mistaken for complete evidence.
Did You Know? A result can be numerically stable and still answer the wrong question if the failure event, denominator or transformation changes. Test one nearby case and predict the direction of change before recalculating. Save inputs, intermediate values and units so a second reader can audit the path.
For practice, turn the example into a sensitivity check. Change one defensible input while holding the others fixed, compare the new result with the original and explain why it moved. Then construct a counterexample in which the shortcut fails. End with two sentences: what the mathematics supports, and what remains uncertain.
Start with the operational question behind parent and educator guidance. Confidence bands connect graphs, probability, maxima, transformations and fair comparisons. In the illustrative case, Ask why protecting an entire curve needs a wider boundary. Draw the relevant graph, density, boundary or timeline before calculating; the picture should use the same variables and sign convention as the equations. Use several covered dots before a continuous band. Show local versus global questions.
A student should be able to rebuild the result without hidden settings. Record the sampling rule or optimisation criterion, the exact failure definition, and the evidence used for every distribution or parameter. Compare a transparent baseline with the advanced method. If they disagree, investigate the mechanism before deciding which answer is more credible.
This chapter belongs to a larger decision, so translate the number carefully. State one action the evidence permits and one action it does not. A parent or educator can ask which school ideas are doing the work: algebra organises relationships, probability represents uncertainty, calculus describes local change, and computing repeats the documented rules at scale.
Treat common misconceptions as a model-checking problem rather than a vocabulary exercise. A Hall–Wellner band is not many independent 95% intervals, overlap is not a two-sample test and coverage does not extend beyond [a,b]. A wide band communicates limited information, not method failure. Recompute the example from the raw values and label every unit. Late tails need risk-set support. State range and confidence level.
The strongest check uses evidence the model did not choose for itself. Hold out later observations, an independent simulation set or a known-answer benchmark. Report errors where decisions are sensitive, especially near the failure boundary, instead of relying on one global accuracy score. A method that performs well on easy safe cases can still fail at the rare cases that motivated the analysis.
Now change one assumption and make a directional prediction before running the calculation. If the result moves the opposite way, inspect the sign convention, transformation, density ratio or constraint. This prediction-first routine develops transferable problem-solving skills because it connects symbolic work to a testable expectation.
Continue with the Mathematics Learning Hub, the eduKate Mathematics Learning System, the approach to learning mathematics, the Singapore Mathematics curriculum overview, the cross-subject learning approach and the previous reliability article. Each destination has its own role: subject navigation, learning method, school-to-advanced progression, or continuity with the latest reliability mechanism.
The mathematics of research and next reading becomes useful only when its mechanism is explicit. Hall and Wellner derived censored-data survival bands through Gaussian-process transformations, while Nair compared simultaneous alternatives including equal-precision bands. Consider the worked illustration: Reproduce a simplified square-root-n width before software. Preserve the original inputs, calculate step by step and compare with a second method whenever an analytic or high-budget reference exists. Method papers do not validate local censoring assumptions. Read band geometry beside the risk table.
There are three uncertainties to keep apart. Sampling or numerical uncertainty comes from finite computation. Parameter uncertainty comes from estimating inputs. Model-form uncertainty comes from simplified physics, distributions or dependence. Reducing the first does not automatically reduce the other two. A credible report names all three and shows which one drives the decision.
For a practical investigation, create a small reproducible notebook or spreadsheet with a frozen data sheet, formula sheet, diagnostics sheet and decision note. Another student should be able to change one input and obtain the same update without guessing what the model meant.
Quick answers
What should I calculate first? Start with the estimand or failure event, then a simple baseline.
Does the advanced method guarantee accuracy? No. It needs input, numerical and model checks.
Which school mathematics matters? Algebra, graphs, probability, functions and careful units all contribute.
How can a student practise safely? Use fictional or open benchmark data and preserve every intermediate value.
What makes a career claim responsible? Describe how the mathematics is used without promising admission, salary or employment.
Extended practice laboratory
Practice case 1: Why pointwise intervals are not a curve band
Setup: Use this chapter’s illustrative case: Scanning 30 milestones creates many opportunities for at least one miss.
Task: Choose local or whole-range inference.
Checking target: Pointwise intervals answer local questions.
Practice case 2: Simultaneous coverage target
Setup: Use this chapter’s illustrative case: A 95% band over 0–1000 hours supports one whole-range statement.
Task: Print [a,b] in the caption.
Checking target: Coverage outside the interval is not promised.
Practice case 3: Kaplan–Meier centre curve
Setup: Use this chapter’s illustrative case: The staircase remains the central estimate.
Task: Validate the observation process first.
Checking target: A confidence band cannot correct a biased centre from informative censoring.
Practice case 4: Gaussian-process limit
Setup: Use this chapter’s illustrative case: Errors at 500 and 600 hours are related because they share earlier events.
Task: Preserve curve covariance.
Checking target: Treating time points as independent is wrong.
Practice case 5: Brownian bridge idea
Setup: Use this chapter’s illustrative case: The bridge is pinned by its construction and has known boundary distributions.
Task: Separate stochastic-process tool from mechanism.
Checking target: This is an asymptotic mathematical representation, not physical degradation.
Practice case 6: Variance-time transformation
Setup: Use this chapter’s illustrative case: Sparse late risk sets move rapidly through variance time.
Task: Plot numbers at risk beside time.
Checking target: Calendar spacing and information spacing differ.
Practice case 7: Hall–Wellner band shape
Setup: Use this chapter’s illustrative case: Widths adapt over follow-up rather than staying constant.
Task: Inspect the formula and endpoints.
Checking target: A single critical value does not imply equal absolute width.
Practice case 8: A simplified width calculation
Setup: Use this chapter’s illustrative case: The illustration isolates the square-root-n scaling.
Task: Use software only after verifying inputs.
Checking target: Real Hall–Wellner scale factors depend on the variance transformation.
Practice case 9: Choosing the lower endpoint
Setup: Use this chapter’s illustrative case: Set a=0 for a mission beginning at installation if appropriate.
Task: Tie a to the decision.
Checking target: An early endpoint before any meaningful exposure may add little.
Practice case 10: Choosing the upper endpoint
Setup: Use this chapter’s illustrative case: Use 1000 hours if both designs retain adequate risk sets there.
Task: Pre-specify b and a risk threshold.
Checking target: Choosing b after seeing results changes error control.