Normal Distribution
nothing is called 95% until it has been counted
Concept

Hierarchical model — where it appears

A model in which group-level parameters are themselves drawn from a distribution whose parameters are estimated. It sits between fitting each group separately and pooling them all, and where it sits is decided by the data rather than by the analyst.

Named by 30 essays across 4 fields — each of them below, with the objects they name alongside it.

Eight groups, τ = 1 against a within-group spread of 3. Each row is a group. The hollow circle is the group's own mean, the filled one is the estimate after pooling, and the small mark is the truth the data was generated from. The group of 3 moves 75% of the way to the population mean of 0.10; the group of 40 moves 18%.

Eight groups, one population

Eight hospitals are neither one hospital nor eight unrelated problems. The two obvious answers cost 2.23 and 1.15 in squared error; the estimate between them costs 0.88, and the weight it uses is not a matter of taste.

hierarchical · Pooling Six groups of 10, each fitting its own slope, then borrowing. Each faint line is one group's own least-squares slope through its own centre; each solid line is that slope after pooling towards the population slope of 0.79. Every group has the same 10 observations. The group whose x values span 0.4 has a slope standard error of 1.86 and moves 91% of the way in; the group spanning 2.0 has a standard error of 0.37 and moves 28%.

The slope that borrows

Pooling a mean makes it look as though how much a group borrows depends on how much data it has. Pool a slope instead and the illusion breaks — ten groups with ten observations each can borrow anything from 28% to 91%, decided entirely by where those ten observations were placed.

multilevel · Levels What eight groups say about τ, when the truth is 1. The posterior density for the population spread after eight groups whose standard errors run from 0.5 to 2.1. The shaded band is the central 95% interval, from 1.02 to 4.44; the posterior median is 1.99 and the mean 2.18. The vertical mark at 1.07 is the moment estimate that empirical Bayes substitutes and then treats as known.

What the plug-in forgets

The shrinkage weight needs a population spread, and the population spread has to be estimated from eight numbers. Empirical Bayes estimates it, substitutes it, and proceeds as though it were known — and the interval that comes out covers 79% rather than the 95% it claims.

fullbayes · Shrinkage Which groups partial pooling serves, standard error 1 population width. Pooling's expected squared error for a group, divided by its own mean's, against how far the group truly sits from the centre. It is ×0.25 at the centre and crosses ×1 at 1.732 population widths, beyond which 8.33% of a normal population lies; capping the shift at one standard error holds every group under ×2.

A group from the population's own tail

Partial pooling halves the total squared error when a group's own standard error equals the spread between groups. Every group whose true effect sits more than 1.73 population widths from the centre — 8.33% of a perfectly normal population — does worse than it would have with its own mean, and its loss grows without bound. Among eight groups with the spread estimated, the most extreme is worse off in 61.6% of datasets. Capping the shift at one standard error keeps the total at 0.528 of the unpooled error and holds every group under twice it.

borrowed · Shrinkage Three priors on the spread, at a true τ of 0.5. The posterior for τ under a flat prior (mean 1.66), a half-Cauchy of scale 1 (1.32) and one of scale 0.25 (1.15). The three answers differ by 30% of the widest. The prior does visible work when eight groups cannot separate a small spread from none, and almost none when they can.

A prior on the spread

Integrating over the population spread means putting a prior on it, which sounds like the objection rather than the repair. The prior's effect is measurable, it is invisible where the groups are clearly different, and the reflex choice for a scale parameter turns out not to have a posterior at all.

fullbayes · Prior Ten groups of 10, pooled on the log-odds scale. Each row is a group. The hollow circle is its own proportion, the filled one is the estimate after pooling, and the vertical rule is the pooled population proportion of 31.3%. One group saw no events at all, and its raw proportion of zero becomes 21.2% — an estimate the group's own data cannot produce and the population's can. The arrows are not the same length, and none of the groups differs in size.

Pooling a proportion

A proportion cannot be shrunk on its own scale — an estimate would leave the interval, and how much information a count carries depends on where it sits. Move to log-odds and the approximation works, at the price of a group that saw nothing having no estimate at all until the correction supplies one.

multilevel · Levels How the truth, the raw means, the posterior means and the constrained estimates spread, standard error 1. Beyond two population widths above the centre lie 2.28% of the true values, 7.86% of the raw means, 0.234% of the posterior means, and 2.28% of the constrained estimates.

Estimates that are too alike

Posterior means give each group its least-error estimate, and as a set they are too alike: with each group's standard error equal to the population's spread, they spread 0.707 as widely as the truth. Beyond two population widths lie 2.28% of the true effects, 7.86% of the groups' own means, and 0.234% of the posterior means — a tenth of the truth. Rescaling the estimates to the right spread counts the tail exactly and costs 17% more squared error; summing each group's posterior chance of being beyond the line counts it without changing any estimate.

borrowed · Shrinkage 16 groups shrunk towards a fitted line, at γ = 1.2. Hollow circles are the groups' own values, filled ones the estimates after pooling, and the diagonal is the line fitted through them with each group weighted by how well it is measured — slope 1.21, intercept 0.20. The horizontal rule is where the same 16 groups would have been shrunk to with no covariate. The spread left to borrow against is 0.64 with the covariate against 1.28 without, so every group is pulled further in than it would otherwise have been.

Borrowing towards a line

A group shrunk towards the average of all groups is being compared with groups it has nothing in common with. Fit a group-level predictor and it is shrunk towards what the predictor says a group like it should be — which halves the spread left to borrow against and takes a quarter off the squared error.

multilevel · Levels How often "there is no spread between the groups" is reported about data that has one. Every dataset here was generated with a real population spread of 1. The moment estimator is the difference between the observed spread and what noise alone would produce, clamped at zero, and the difference comes out negative often: at eight groups it reports exactly zero on 32.6% of datasets, which is an instruction to pool completely and give all eight groups the same estimate. The rate falls to 4.2% at 48 groups.

When the spread estimates to zero

The usual estimate of a population spread is a difference of two positive quantities, clamped at zero. On a third of eight-group datasets with a real spread in them the difference comes out negative, the estimate is exactly zero, and every group is pooled completely on data that said no such thing.

fullbayes · Pooling How many of a league table's top ten are small groups, ranked four ways. A hundred groups with sizes from 4 to 400, of which 36% have twenty units or fewer. Small groups make up 36.3% of the true top ten, 62.1% of the top ten by raw means, 13.4% by posterior means and 22.9% by the posterior chance of being in the top ten. The three rankings recover 4.43, 5.38 and 5.47 of the true top ten.

A league table of a hundred

A hundred groups with sizes from 4 to 400, and a top ten to publish. Ranked by their own means, small groups fill 62.1% of the top ten against their 36.3% share of the true top ten. Ranked by posterior means they fill 13.4%. The ranking built from each group's chance of being in the top ten recovers 5.47 of the true ten, the best of three and barely half; and the group ranked first could hold any rank from 1 to 31.

borrowed · Shrinkage Counted coverage of two 95% intervals, over 2,000 datasets. Each point is one of the eight groups, at its own standard error. The integrated interval covers 95.2% overall against its stated 95%; the plug-in covers 78.8%, and its shortfall grows with the group's standard error — from 86.1% at se 0.5 to 77.0% at se 2.1. The mean widths are 3.48 and 2.66.

The interval that integrates

A credible interval for one group in a hierarchy has to average over every value the population spread might take. That averaging is what makes it cover — 95.2% against the plug-in's 78.8% — and it costs 31% more width, a heavier tail, and a mixture rather than a normal.

fullbayes · Credible What 20 clusters of 20 correlated observations do to a 95% interval. Each study has 400 observations arranged as 20 clusters of 20. The lower points are the counted coverage of the usual interval, which treats them as 400 independent observations; the curve through them is 2Φ(1.96/√deff) − 1 with deff = 1 + 19ρ, computed before any data was drawn. At ρ = 0.81 the interval covers 36% rather than 95%. The upper points treat the cluster as the unit and need no variance components at all.

Two levels at once

A third level of grouping adds no new arithmetic and produces one number — the design effect — that decides how many independent observations a clustered study is worth. It is the same quantity the time-series field computes for autocorrelated data, arrived at from a completely different picture.

multilevel · Levels One group 6 population widths from the rest. Squared error for each group under partial pooling, with each group's own mean beside it. Seven of the eight are estimated better by pooling. The eighth, which was never from the population, is estimated 6.1 times worse — 13.9 against 2.3.

When borrowing goes wrong

Partial pooling wins on the total and can lose badly on one group. Placed six population widths out, the group that was never from the population is estimated six times worse than by its own mean — and nothing in the output says so.

hierarchical · Pooling The bottom ten of a league table of a hundred when larger groups do better, size–effect slope 0.6. Groups of 4 to 400, true effects rising with log size at a slope of 0.6 population widths. The true bottom ten is 80.6% small groups. Ranked by their own means the bottom ten holds 4.28 of the true ten; pooled to one centre, 3.19, with 42.6% small groups; pooled to a centre that rises with size, 4.35, with 83.0%.

The small groups one centre protects

In a league table of a hundred groups where larger groups do better on average, pooling every group towards one centre overstates the small groups by half a population width — 0.501 — because it pulls them towards an average they are not part of. The top ten barely notices. The bottom ten, the list that flags poor performers, holds 3.19 of the true bottom ten when pooled to one centre and 4.28 when the groups are ranked by their own noisy means. Pooling towards a centre that rises with size restores 4.35 and puts the small groups back at the bottom where 80.6% of the true bottom ten sit.

borrowed · Shrinkage What each estimator costs, against how many groups there are. Each point is 4,000 datasets, with the population spread estimated from the data rather than supplied. Partial pooling first beats BOTH of the estimators it sits between at 5 groups; below that, complete pooling — which estimates nothing at all — is the better answer. Its own cost falls from 1.751 at 2 groups to 0.795 at 40.

The fewest groups that can borrow

At three groups the estimator that shrinks towards its own data's mean returns the group means untouched, on every dataset, because its constant is J − 3. At two it expands instead of shrinking. And the number of groups at which partial pooling starts to be worth doing is five, or two, or never — it depends on how far apart the groups are.

hierarchical · Pooling What 240 observations are worth, by which question is asked. 8 rows and 10 columns with 3 observations in each cell — 240 in all, each belonging to one row and one column, neither nested in the other. the overall mean: variance 0.1703 against a naive 0.0060, a design effect of 28.4 and 8.5 effective observations; a difference between two rows: variance 2.0682 against a naive 0.0960, a design effect of 21.5 and 11.1 effective observations; a difference between two columns: variance 1.0845 against a naive 0.1200, a design effect of 9.0 and 26.6 effective observations.

Two groupings that cross

Pupils belong to a school and to a neighbourhood, and neither is nested in the other. There is then no design effect: the overall mean is worth 8.5 independent observations out of 240, a row difference 11.1 and a column difference 26.6, and which grouping matters depends on the question rather than on the study.

multilevel · Levels How often a trial that borrows its control from an earlier one declares a treatment with no effect a success, by the earlier control's drift. Fifty patients an arm, an earlier control of two hundred, a one-sided 2.5% threshold. Without borrowing the rate is 2.5% at every drift. Pooling outright reaches 73.1% at a drift of half a standard deviation. A half-normal prior of scale 0.05 on the between-trial spread peaks at 35.2% and is still at 31.8% at a drift of 1.5; a half-Cauchy of the same scale peaks at 11.0% near a drift of 0.4 and falls back to 3.4%.

A control borrowed from the last trial

A trial of fifty patients an arm can borrow its control group from an earlier trial of two hundred by treating the two control means as draws from one population with a spread between trials. With two trials there is one difference to learn that spread from, so the prior on it decides how much is borrowed — and its tail decides whether the borrowing stops when the two trials disagree. A half-normal prior of scale 0.05 buys 17.4 points of power and, at a drift of 1.5 standard deviations, still declares a treatment with no effect a success 31.8% of the time. A half-Cauchy of the same scale buys 13.6 points, peaks at 11.0% and falls back to 3.4%.

fullbayes · Prior How far each estimate of the small groups sits from their truth, when the size relation is flattening below the median size. The small groups' average true effect is −0.259. Their estimates are off on average by −0.012 (own means), +0.276 (one fitted centre), −0.173 (fitted line in log size), −0.025 (fitted quadratic in log size), −0.003 (the true relation).

The relation the table has to estimate

Pooling a league table's groups towards a centre that rises with size repaired the small groups' bias — when the size relation was known. Fitted from the table itself, a straight line in log size loses almost nothing: the small groups' error is 0.750 against 0.719 with the true relation. When the relation bends below the median size, the fitted line extrapolates the large groups' slope into the region where it is barely observed and misplaces the small groups by −0.173 if the relation flattens and +0.264 if it steepens — smaller than one centre's error, of either sign. A quadratic in log size removes nearly all of it, −0.025 and +0.014, and costs 0.025 in error when the relation was straight.

borrowed · Shrinkage What each group gains from being pooled, τ = 1. Eight groups whose sizes span a factor of 13.3. The smallest gains 2.076 of squared error, which is 36.8% of the total reduction; the largest gains 0.030, which is 0.5%. 2 of the eight account for half of everything pooling buys.

Where the borrowing goes

Pooling cuts the total squared error across eight groups by 56%. Two of the eight take 61% of that reduction, the four best-measured groups share 11% between them, and the largest group gets 1.5% of what the smallest does. The headline is a fact about the groups nobody was asking about.

hierarchical · Pooling What a variance estimated from K units is worth. The between-unit mean square is a scaled chi-square on K − 1 degrees of freedom, so the estimator's whole distribution is decided by the number of units. At two units its interquartile range spans a factor of 13.03 and its ten-to-ninety range a factor of 171.3, and it comes out exactly zero on 26.7% of studies. The closed form and 3,000 simulated studies agree to 0.051 at every quantile.

A level with two units

A variance estimated from two units is a scaled chi-square on one degree of freedom. Its interquartile range spans a factor of thirteen, its ten-to-ninety range a factor of a hundred and seventy-one, and it comes out exactly zero on 26.7% of studies — so the design effect it decides runs from 1.00 to 7.01 against a truth of 4.69.

multilevel · Levels How often a trial borrowing from 16 earlier trials that agree declares a treatment with no effect a success, by its own control's drift. Fifty patients an arm, each earlier control of two hundred, a one-sided 2.5% threshold. With 16 earlier trials, the half-normal prior peaks at 88.1%, the half-Cauchy at 80.7% near a drift of 0.6, and the half-Cauchy with a fifth of the prior on a vague component at 16.9%. At a drift of 1.5 the three read 73.3%, 25.5% and 3.7%.

A history that agrees with itself

Borrowing a control from sixteen earlier trials that agree with each other should be safer than borrowing from one, and it is the opposite. Sixteen agreeing trials estimate the spread between trials as small, and a small spread estimated confidently is a licence to pool: under a half-Cauchy prior that let go of a single disagreeing trial, a current control 0.6 standard deviations from a sixteen-trial history turns a treatment with no effect into a success 80.7% of the time, against 10.5% at worst with one earlier trial. Moving a fifth of the prior onto a vague component brings the worst case to 16.9% and keeps power at 91.0%, against 70.5% without borrowing.

fullbayes · Prior How often seven 95% intervals cover the small groups of a league table whose size relation is steepening below the median size. Groups of twenty patients or fewer, 600 tables of a hundred groups. Coverage: one centre, plugged in 58.0% (average width 2.81); one centre, its uncertainty added 58.5% (average width 2.83); a fitted line, plugged in 83.1% (average width 1.58); a fitted line, its uncertainty added 88.4% (average width 1.77); a fitted quadratic, plugged in 85.5% (average width 1.53); a fitted quadratic, its uncertainty added 93.7% (average width 1.92); the true relation and spread 94.8% (average width 1.55).

The interval a small group is given

A league table's 95% interval for a small group, pooled towards a size relation fitted from the table, covers 92.8% of the time when the relation is straight and 83.1% when it steepens below the median — and 77.4% for the groups of eight patients or fewer. Adding the fitted centre's own uncertainty, weighted by how much the group borrows from it, brings the line to 88.4% and a fitted quadratic to 93.7%, and the quadratic's interval covers 94.6% of the smallest groups at every shape tried. One centre, plugged in, covers the smallest groups 35.2% of the time when the relation steepens, and no added variance repairs a centre in the wrong place.

borrowed · Shrinkage Twelve groups from two clusters, τ = 1. Every group's truth is in one of two clusters, and the population's total spread is exactly τ = 1, so the analysis recovers τ̂ = 1.515 and every shrinkage weight is what it would be for a single normal population. The estimates are pulled towards the grand mean, which is the middle of the gap — a place 0 of the 12 truths are and 6 of the estimates end up.

One population, or two

Group effects from two clusters rather than one bell, with the same total spread. The analysis recovers the same population spread, uses the same weight for every group, and reports nothing unusual — while 46% of its estimates land in a region holding 6.6% of the truths.

hierarchical · Pooling Each interval covers one question and not the other. Coverage of each interval for the overall mean, scored against both estimands, over 20,000 two-site studies of 10 observations apiece. The fixed-effect interval covers the mean of the two sites in hand 96.37% of the time and the population mean 54.77%. The random-effects interval covers the population mean 94.96% — exactly its level, from one degree of freedom — and over-covers the two sites in hand at 98.25%. Both are correct; they are answers to different questions printed in the same place.

What a two-unit study should report

The fixed-effect interval covers the mean of the two sites in hand 96.37% of the time and the population mean 54.77%. The random-effects interval covers the population mean 94.96% — exactly its level, from one degree of freedom — and is 11.6 times wider.

multilevel · Levels How often a trial borrowing from 8 earlier trials on a trend declares a treatment with no effect a success. Fifty patients an arm, each earlier control of two hundred, the current trial one step after the last. At slopes 0, 0.01, 0.02, 0.03, 0.05, 0.075, 0.1: exchangeable 2.7%, 4.4%, 6.6%, 8.3%, 9.9%, 8.3%, 6.7%; with a fifth of the prior vague 2.4%, 3.6%, 5.5%, 6.9%, 7.6%, 6.5%, 5.7%; with a line in time 2.9% at every slope.

A history that has been moving

Earlier control arms borrowed as exchangeable draws are pooled towards the history's middle, and when the control mean has been rising the current trial sits at its end. A drift of 0.02 standard deviations a trial — a step between neighbouring trials of under a third of either one's standard error — turns a treatment with no effect into a success 9.9% of the time from sixteen earlier trials, and a drift of 0.05 overstates the effect by 0.130 standard deviations from eight. A line in time removes the bias exactly, at every slope, and pays for it in borrowing: from eight earlier trials it borrows 175 control patients where the exchangeable model borrows 439.

fullbayes · Prior Every group's 90% rank interval in one league table whose size relation is steepening below the median size, with its true rank. A hundred groups from four to four hundred patients, ranked from their posterior under a fitted quadratic size relation whose uncertainty is drawn once per table. The groups of twenty patients or fewer have rank intervals 30.1 places wide on average, and 34 of their 36 true ranks fall inside; the groups of over a hundred, 32.1 places, with 27 of 30 inside.

A ranking with its uncertainty

A league table's reader wants a rank, and a rank interval inherits a fitted size relation's errors differently from an effect interval. When the relation steepens below the median size, a fitted line's 95% interval for a small group's effect covers 83.1% of the time; its 90% interval for the same group's rank covers 95.6%, because the line's error lifts the small groups into the crowded middle of the table, where rank intervals are wide. The error surfaces instead in the large groups' ranks, 87.6%, and in width: 36.9 places against the 27.3 the true relation needs. A quadratic's uncertainty has to be drawn once per table, since it moves every small group together — added group by group it over-covers at 92.8%.

borrowed · Shrinkage The bias each model puts into the treatment effect when sixteen earlier controls stepped once. Sixteen earlier trials of two hundred, the first eight a step below the last eight, which sit at the current control's level; the current trial has fifty patients an arm and no treatment effect; 1,500 trials at each step. Exchangeable trials: -0.003, 0.020, 0.041, 0.056, 0.063, 0.065, 0.054. A line in time: -0.003, -0.016, -0.029, -0.041, -0.052, -0.067, -0.075. A change point: -0.003, 0.003, 0.003, -0.001, -0.002, -0.003, -0.003 — at steps of 0, 0.05, 0.1, 0.15, 0.2, 0.3, 0.5.

A history that stepped

When sixteen earlier control arms stepped once — the first eight a fifth of a standard deviation below the last — borrowing them as exchangeable flatters a treatment with no effect by 0.063 and stops borrowing, and fitting a line through them penalises it by 0.052. A model with a change point is unbiased at every step size, borrows more the larger the step, 512 patients at a fifth, and puts two thirds of its belief on the right place for the change. Against a steady drift it is the one that fails, so each of the three is safe only against the shape of history it assumes.

fullbayes · Prior Sixteen groups from two clusters: the true effects, the observed means and the pooled estimates, with the kurtosis of each. One dataset of 16 groups of four observations each, effects drawn from two clusters with total spread 2, noise 3. The fitted spread is 1.701, so every estimate is pulled 43.7% of the way to the centre -0.213. Kurtosis: true effects 1.203, observed means 2.348, pooled estimates 2.348 — the last two identical, because one is an affine image of the other.

The shape the estimates keep

Partial pooling was suspected of destroying the evidence of its own misfit: compressing a two-cluster population of group effects towards one centre before any test could see the clusters. With groups of equal size it destroys nothing, because every estimate moves the same fraction of the way to the same place and the estimates keep the means' shape exactly — except when the fit pools completely, as it does on 29.9% of eight-group datasets at a spread of one. The evidence was never there: a shape check has 6.5% power at eight groups. With groups of unequal size the pooled estimates see the clusters better than the raw means do.

hierarchical · Pooling Coverage of the population mean from two units, by the true spread between units. 4,000 two-unit studies at each true spread, ten observations a unit, within-unit standard deviation 1.2. exact, on one degree of freedom: 94.97%, 95.50%, 94.95%, 94.83%, 95.15%, 95.20%, 94.67%. spread supplied as 1: 100.00%, 100.00%, 99.92%, 95.53%, 81.85%, 70.03%, 51.55%. half-normal, scale 1: 100.00%, 100.00%, 99.78%, 94.63%, 86.25%, 79.70%, 68.38%. half-Cauchy, scale 1: 100.00%, 100.00%, 100.00%, 98.98%, 95.50%, 92.38%, 88.00%. inverse-gamma(0.001, 0.001) on the variance: 100.00%, 100.00%, 100.00%, 99.15%, 97.05%, 95.28%, 92.65% — at spreads 0, 0.25, 0.5, 1, 1.5, 2, 3. The exact interval holds 95% at every spread; every integrated one over-covers when the spread is small and under-covers when it is large, and where it crosses 95% is set by the prior's tail.

A prior on the spread between two sites

Integrate the between-site spread of a two-site study under a half-normal prior of scale 1 and the interval for the overall mean is 3.01 wide and covers 94.63% when the true spread is 1 — the supplied interval again, under another name. At a true spread of 3 it covers 68.38%. Every prior tried covers 95% averaged over its own draws and none covers it at every spread; each holds up to about its own scale, because two sites move the posterior median of the spread by less than a third of what the truth moves.

multilevel · Levels How often a league table's own data choose a bent centre over a straight one. Over 400 tables of a hundred groups each. The quadratic's curvature coefficient is significant at 5% on 5.3% of tables whose relation is straight, 14.5% of tables whose relation is flattening below the median, 29.3% of tables whose relation is steepening below the median; its average AIC weight is 0.374, 0.456, 0.558.

A centre chosen by its own table

A league table that picks its centre's shape from its own data — a line unless a curvature test says otherwise — picks the line on 85.5% of tables whose relation flattens below the median, because the bend lives where the groups are smallest and noisiest. Its bottom ten then flags like the line's: 2.61 innocent small groups a table against 2.19 for a quadratic told the shape. Always fitting the quadratic costs 0.07 of a correctly flagged group when the relation is straight and matches the told answer when it bends, and a flag set by probability carries the choice through a mixture of the two.

borrowed · Shrinkage

Named alongside it

The objects these essays reach for when they reach for this one.

Partial poolingShrinkageEmpirical BayesVariance componentsPriorRandom-effectsLeague tablePrior sensitivitySample sizeStudy designCoverageThe half-Cauchy prior

All concepts