In a frailty model, an individual's unobserved susceptibility to an event is represented by a nonnegative frailty random variable multiplying that individual's hazard function. Conditional on its value, subjects with the same measured covariates can therefore have different event rates. The frailty is a persistent latent characteristic of the individual, not newly sampled noise at each time.
Here the conditional hazard function is , with baseline hazard . A multiplier raises susceptibility and lowers it. Normalization fixes the frailty scale: without it, multiplying every frailty by a constant and dividing the baseline hazard by that constant would give the same individual hazards. The variance of measures unobserved heterogeneity. An atom at , if allowed, would give a permanently zero conditional hazard.
The hazard ratio between groups is for individuals compared at the same frailty. It need not equal the population hazard ratio, because survival selection changes the frailty distribution among those still alive or event-free.
Differentiating the log-likelihood again gives
Thus the observed Fisher information for the rate is proportional to the total number of observed events. At a positive interior maximum-likelihood estimator, the large-sample variance estimate and standard error are
Precision is therefore driven by the number of observed events rather than by the number recruited alone. Heavy right censoring can leave a large study with little information about its hazard function. Under the regularity conditions for the Fisher information, the expected information is .
A right-censored observation is not wholly uninformative: its exposure contributes to , changes the maximum-likelihood estimator, and affects the expected number of events under a sampling design. Its contribution happens to have zero second derivative in this rate parameterization. For example, if no events occur, the likelihood still declines as and disfavors large rates, despite its zero curvature. The usual interior variance approximation is then invalid.
For an exponential distribution of rate , the probability density function and survivor function are and . Set
where is the observed event count and is total person-time at risk. The log-likelihood, up to a constant, becomes
When and , its score function is , which vanishes at
Its second derivative is negative, so this is the unique maximum-likelihood estimator. This is exponential-rate estimation from censored exposure: the estimate is observed events divided by all observed exposure, including exposure from right-censored individuals.
If and , decreases strictly. There is no maximizer in the open parameter space ; its supremum occurs as , or the extended boundary estimate is if a zero rate is permitted. The formula should not be interpreted as an interior regular fit in this case.
Assume independent individuals and independent censoring: conditional on any modelled covariates, the censoring mechanism gives no additional information about the latent event time and has no parameter shared with its survival distribution. An observed event at contributes the probability density function . A right-censored observation contributes the survivor function , since it establishes only .
More formally, let be an independent censoring time with density and survivor function . The observed pair , has a parameter-dependent factor if and if . The factors involving only the censoring distribution can be omitted from the event-time likelihood. Thus
Using also gives the useful survival likelihood form
Every individual therefore contributes survival exposure, whether or not an event is observed. These formulas do not justify ignoring informative censoring; that situation requires a model for the observation mechanism as well.
The paper uses for the survivor function, rather than for the cumulative distribution function. For a nonnegative survival time with an absolutely continuous probability distribution, define
Where , the hazard function is the instantaneous event rate conditional on survival:
The integrated hazard is . Since , integration gives . With the usual time origin and no event at zero, , so
Conversely, differentiating gives , so these relationships determine each of the survivor function, hazard function and integrated hazard from either of the others. A positive limiting survivor function is possible if the integrated hazard remains finite; a proper finite event time instead has .
Insert the shape-two gamma prior into the integral from part (ii). The gamma function integral gives
Therefore has a negative binomial distribution with shape two and success probability , using the failures-before-two-successes convention. This is the Poisson-gamma mixture. The identity verifies that the probabilities sum to one.
Its adjacent probability ratio is
Substituting in the posterior formula gives
exactly the estimate from part (i). The equality is required by sufficiency, and the direct computation confirms that the two negative-binomial and gamma parameter conventions have been used consistently.
Given , the sum of the independent yearly counts is : their probability generating functions multiply to . Averaging this conditional law over the prior gives the mixed Poisson distribution
The likelihood for the full vector depends on only through , so is a sufficient statistic; conditioning on the total or on the full observed vector gives the same posterior for .
For an event of positive probability , the posterior density is proportional to . Put . Its mean is . The future count has conditional mean , so
This is the posterior mean from adjacent mixed Poisson probabilities. Both probabilities refer to the same -year exposure; the numerator is not a probability for . For a proper prior and , the integrals defining this posterior mean are finite because is bounded on .
Let . The conditional joint likelihood function of the observed counts is
The prior is a gamma distribution with shape two and rate . Multiplication by its density gives the unnormalized posterior density . Normalizing with the gamma function yields
Thus Poisson-gamma conjugacy gives . Conditional on and the past data, the next count has mean , so the law of total expectation gives
With and prior mean , this is exactly
Here , , and the credibility factor increases to one as the number of observed years grows. The denominator uses the prior rate and total exposure ; it does not add two to the exposure when adding the prior shape.
For a regular iid model with a fixed-dimensional interior parameter , identifiability, sufficient differentiability and moment domination, and nonsingular per-observation Fisher information , a consistent maximum-likelihood estimator solves the score equation. Expanding it gives
where . The score has mean zero and covariance under the regularity assumptions. This proves the usual large-sample information bound and asymptotic efficiency within regular parametric inference. The observed curvature and the empirical score covariance consistently estimate when the model is correctly specified.
For smooth constraints with full-row-rank derivative , the Wald test uses
The likelihood-ratio test and score test, evaluated using the constrained fit and nuisance-adjusted efficient information, have the same first-order chi-square law. A quadratic local likelihood expansion explains their asymptotic equivalence. Profile likelihood supplies inference about a parameter of interest with fitted nuisance parameters, and the corresponding chi-square cutoffs give approximate confidence regions.
These conclusions depend on the stated regularity: parameters at a boundary, nonidentifiability, singular information, changing dimension, or parameter-dependent support can give different rates and limits. Under misspecification, the fitted parameter generally approaches a pseudo-true value and its variance is a sandwich covariance matrix, not simply inverse Fisher information.
For a fully specified continuous null distribution , compare the empirical distribution function with . The probability integral transform makes iid uniform under the null, giving distribution-free calibration. Three common discrepancies are
These define the Kolmogorov-Smirnov test, Cramér–von Mises criterion and Anderson–Darling test, respectively. Reject for large values. The supremum detects the largest discrepancy, the unweighted integral combines discrepancies throughout the distribution, and the reciprocal weight emphasizes the tails.
The Donsker theorem for empirical distribution functions gives for a Brownian bridge , with covariance . Consequently the standard continuous-null limits are , , and for , respectively; the weighted last limit also requires control near the endpoints. Exact finite-sample or asymptotic null quantiles provide critical values.
If parameters in are estimated from the same observations, these fully specified-null critical values generally fail. A first-order expansion is
so the estimated parameter changes the limiting process. One must use the fitted-model null law, adjusted asymptotic theory, or a parametric bootstrap that re-estimates the parameters in each replicate.
Population influence functions describe infinitesimal contamination; finite-sample analogues examine actual changes to an empirical sample. For its empirical distribution , add an observation and define the sensitivity curve
Under suitable differentiability it approximates the influence function. A replacement version is , approximately . Their supremum magnitudes quantify finite one-observation sensitivity.
The empirical influence function instead substitutes directly in . For a differentiable M-estimator this gives . For quantile functionals, a discrete unsmoothed empirical law may not admit the required density derivative; a sensitivity curve or a smoothed empirical law is then preferable.
Global contamination is measured by finite-sample maximum bias , where counts replacements. Under the convention of the first fraction that causes unbounded bias,
This is the finite-sample replacement breakdown point. The largest replacement fraction still guaranteed bounded is one grid step smaller, so the conventions must be specified. One replacement sends a sample mean arbitrarily far, giving , while the ordinary scalar sample median has first-breaking fraction with the usual midpoint convention for even samples. Its limiting value is . Local influence and global breakdown answer different questions; a bounded influence function is not by itself a proof of a high breakdown point.
In statistical decision theory, the statistical invariance principle says that a procedure should respect transformations that leave the inferential problem unchanged. If a group action preserves the model and carries each of the null and alternative parameter sets into itself, an invariant test has rejection probability function . A maximal invariant identifies its orbits, so every invariant test factors through that statistic: define its value at an orbit using any representative; invariance makes the definition independent of the representative. If the group is transitive on the null parameter set, the null law of any invariant statistic is free of that nuisance parameter.
For example, testing a normal mean with unknown scale, positive transformations preserve a one-sided mean alternative. The statistic , with the usual unbiased sample-variance scale , is invariant and has the Student t-distribution with degrees of freedom under the null. For estimation the corresponding requirement is an equivariant estimator, together with a loss function invariant under simultaneous transformation of decision and parameter. Invariance alone does not prove unrestricted optimality or supply a unique best procedure.
There is a distinct probabilistic use of the term: the Donsker invariance principle states that centred iid partial sums, normalized by , converge as processes on to Brownian motion. Its empirical-process version gives a Brownian bridge, used by goodness-of-fit tests in the next note. These functional limit theorems should be distinguished from the group-based inferential principle.
Let be a regular likelihood-ratio test statistic for smooth restrictions, so its leading null law is . Suppose a higher-order calculation gives . The Bartlett correction rescales the statistic:
Its mean is , eliminating the first-order mean error. In regular likelihood models satisfying the usual Bartlett expansion, the same scaling removes the leading distributional error against the chi-square reference law, leaving an error for fixed tail probabilities. Matching a mean alone would not establish this distributional conclusion for an arbitrary statistic.
The coefficient depends on the model, null restrictions and nuisance parameters; it can be obtained from higher derivatives and cumulants of the log-likelihood, or estimated by null simulation when an analytic expression is inconvenient. Nuisance parameters require their appropriate null values or estimates. A correction improves finite-sample calibration within regular asymptotic theory; it does not turn a boundary-parameter or unidentifiable problem into a regular chi-square problem.
For , suppose has a unique interior maximum , is positive definite, the amplitude is smooth there, and contributions away from that maximum are negligible. Set . The quadratic expansion yields the Laplace approximation
The determinant comes from diagonalizing the Gaussian integral; it is an inverse square root, not an inverse determinant. Under standard higher smoothness and domination, the relative error is .
For a regular model and a fixed smooth positive proper prior , apply this to the marginal likelihood . With ,
As is of order , this produces the complexity term in the Bayesian information criterion after multiplying log marginal likelihood by . A posterior mode can also be used with the corresponding combined log-density curvature. Boundary modes, singular information and multiple dominant modes invalidate the single interior Gaussian formula; distinct nondegenerate dominant modes must have their contributions added.
Parameters and are orthogonal parameters if the cross block of the expected Fisher information matrix is zero throughout the relevant parameter region:
This is a statement about expected information, not a requirement that every observed cross derivative vanish. For regular maximum-likelihood estimators, has limiting covariance equal to the inverse per-observation Fisher information matrix. Block diagonality therefore gives asymptotically independent fitted interest and nuisance parameters. The efficient information reduces to at an orthogonal parametrization, although nuisance estimation can still matter at higher orders.
For scalar interest and nuisance, retain and write the old nuisance as . The transformed cross information is . Hence an orthogonal nuisance coordinate can locally be found from
subject to local solvability and a nonsingular coordinate change. Global coordinates need not follow from this local construction. For the normal distribution with separately varying mean and variance parameters , the scores are and ; their product has zero expectation by symmetry. Thus the per-observation information is diagonal with entries and .
An Edgeworth expansion refines the central limit theorem by incorporating higher cumulants. For iid centred observations with variance , standardized sum , standardized third cumulant and fourth cumulant , its central-region density expansion is
under sufficient moment and smoothness/nonlattice conditions. Here , , are probabilists' Hermite polynomials. Expanding the characteristic exponent gives the cubic and quartic terms, while squaring the cubic term in the exponential gives the correction. Integrating the first correction, using , gives
Thus skewness determines the first departure from the normal approximation; inversion gives a Cornish-Fisher expansion for quantiles. These are asymptotic expansions, not guaranteed nonnegative densities in extreme tails. Lattice data require continuity/lattice corrections, and moment assumptions alone do not justify a uniform smooth-density expansion.
The root calculation for the Huber location estimator gives
Since and equality is attained for , its gross-error sensitivity is
Thus it is a B-robust estimator at the normal reference distribution. At a general centred distribution the same conclusion requires , in which case the bound is . Boundedness of the score alone does not rescue a zero sensitivity denominator.
At the standard normal distribution, symmetry and the odd Huber score give population root zero. For , the derivative of the clipped score is except at its two corners, which have zero normal probability. Put and write for the standard normal density. The sandwich variance of an M-estimator has numerator
Using and integration by parts,
Therefore
As checks, it tends to as , the sample mean limit, and to as , the sample median limiting variance. The estimator variance for a sample of size is approximately this quantity divided by .
One presentation is the cost-effectiveness plane, with horizontally and vertically. Plot the point estimate with a joint confidence region or paired bootstrap cloud. Its position shows the tradeoff quadrant, while the cloud shows uncertainty, dependence and possible dominance. A willingness-to-pay threshold is the line through the origin. Points below that line have positive incremental net monetary benefit, including points in quadrants where a ratio rule would fail. The point-estimated switch occurs at pounds per day.
A second presentation is a cost-effectiveness acceptability curve, giving an uncertainty-based measure of cost-effectiveness for a range of thresholds. For paired bootstrap increments , plot
This counts positive net benefits, not merely ratios below the threshold, so it handles all quadrants. Using Bayesian inference the curve is the posterior probability that incremental net monetary benefit is positive; a frequentist bootstrap curve is a resampling summary, not literally a probability that a fixed parameter is positive.
Here the estimated incremental net monetary benefit and its standard error are
Thus a normal approximation uncertainty curve is
It passes through one half at £40 per day. Alternatively display the incremental net monetary benefit curve itself with approximate pointwise bands . The plane preserves the joint uncertainty; the acceptability curve shows how the threshold changes the cost-effectiveness assessment. Neither presentation chooses society's willingness-to-pay threshold.
The incremental cost-effectiveness ratio must be read with the signs of both increments. A negative ratio can mean either greater effectiveness at lower cost, which is favorable, or lower effectiveness at greater cost, which is unfavorable. A positive ratio can likewise lie in either the more-costly/more-effective or cheaper/less-effective quadrant. Thus ordering ratio values without locating the cost-effectiveness plane quadrant can reverse a decision.
The incremental cost-effectiveness ratio is undefined at a zero effect increment and extremely unstable near it. Its sampling distribution can be highly skewed and its mean may fail to exist; a usual symmetric confidence interval or an interval spanning positive and negative ratios can conceal that instability. Cost-effect correlation must be retained, as the Fieller confidence set calculation illustrates.
Interpretation also requires a specified comparator, effect unit, costing perspective, follow-up horizon and willingness-to-pay threshold. Different comparator choices give different incremental ratios, and lower ratios alone do not settle decisions across differently scaled outcomes. Longer survival need not mean a proportional quality-of-life gain; extrapolation, censored outcomes and omitted costs can affect the increments. Point estimates conceal both sampling uncertainty and uncertainty in these assumptions.
For a positive effect increment, comparison with a threshold is equivalent to . For negative effect increments that inequality reverses when multiplied by , so a rule that blindly compares all ratios with is wrong. Use the signs and joint cost/effect uncertainty, or the incremental net monetary benefit , rather than the ratio alone. No particular threshold is given in the paper, so the point estimate does not establish whether reimbursement should occur.

Pinned article: Introduction to the OurBigBook Project

Welcome to the OurBigBook Project! Our goal is to create the perfect publishing platform for STEM subjects, and get university-level students to write the best free STEM tutorials ever.
Everyone is welcome to create an account and play with the site: ourbigbook.com/go/register. We belive that students themselves can write amazing tutorials, but teachers are welcome too. You can write about anything you want, it doesn't have to be STEM or even educational. Silly test content is very welcome and you won't be penalized in any way. Just keep it legal!
We have two killer features:
  1. topics: topics group articles by different users with the same title, e.g. here is the topic for the "Fundamental Theorem of Calculus" ourbigbook.com/go/topic/fundamental-theorem-of-calculus
    Articles of different users are sorted by upvote within each article page. This feature is a bit like:
    • a Wikipedia where each user can have their own version of each article
    • a Q&A website like Stack Overflow, where multiple people can give their views on a given topic, and the best ones are sorted by upvote. Except you don't need to wait for someone to ask first, and any topic goes, no matter how narrow or broad
    This feature makes it possible for readers to find better explanations of any topic created by other writers. And it allows writers to create an explanation in a place that readers might actually find it.
    Figure 1.
    Screenshot of the "Derivative" topic page
    . View it live at: ourbigbook.com/go/topic/derivative
  2. local editing: you can store all your personal knowledge base content locally in a plaintext markup format that can be edited locally and published either:
    This way you can be sure that even if OurBigBook.com were to go down one day (which we have no plans to do as it is quite cheap to host!), your content will still be perfectly readable as a static site.
    Figure 5. . You can also edit articles on the Web editor without installing anything locally.
    Video 3.
    Edit locally and publish demo
    . Source. This shows editing OurBigBook Markup and publishing it using the Visual Studio Code extension.
  3. https://raw.githubusercontent.com/ourbigbook/ourbigbook-media/master/feature/x/hilbert-space-arrow.png
  4. Infinitely deep tables of contents:
    Figure 6.
    Dynamic article tree with infinitely deep table of contents
    .
    Descendant pages can also show up as toplevel e.g.: ourbigbook.com/cirosantilli/chordate-subclade
All our software is open source and hosted at: github.com/ourbigbook/ourbigbook
Further documentation can be found at: docs.ourbigbook.com
Feel free to reach our to us for any help or suggestions: docs.ourbigbook.com/#contact