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

Experimental design — where it appears

The decisions taken before any data exists — which settings to measure at, how many runs, which unit gets what. They are the decisions with the largest effect on what an experiment can conclude and the only ones that cannot be revisited afterwards.

Named by 83 essays across 22 fields — each of them below, with the objects they name alongside it.

What a median split can see. A standard normal covariate with its median marked, and the mean of each category as a vertical rule: -0.7979, 0.7979. A rule that balances the categories is balancing those numbers and nothing else, so the part of the covariate it can act on is the variance between them — 0.6366 of the total, which at two categories is exactly 2/π because the two half-normal means are ±√(2/π). The rest, 0.3634, is variation inside the categories that the rule cannot see and does not touch: the assignment within a category is still a coin. Everything the next figure measures is a consequence of this one, and it is available before any unit has arrived.

A covariate with no levels

Every balancing rule on this site reads a level. Age and blood pressure have none, so somebody cuts them into categories — and a median split can see exactly 2/π of a normal covariate, whatever the rule does with the halves.

continuous · Assignment The expansion that never terminates. The Hermite coefficients of a median split, in magnitude, against the reference j to the power −3/4, anchored at the first one. Every even order is exactly zero because sign is an odd function, and every odd order is not, so no truncation is exact — where a polynomial of degree d is exact at any order past d. Summed, the tail past J falls like 1/√J: sixty orders still leave 6.6% of the variance outside. That statement is what made a cut dictionary's geometry unavailable in closed form, and it is a statement about the function against itself. What it is not is the accuracy of an inner product between two correlated variables, where every term past J carries a factor of ρ^m as well.

A cut is not a polynomial, and it does not have to be

A threshold's expansion never terminates, which is why a balancing dictionary's geometry was closed for powers and taken to draws for cut points. Conditioning on the second variable closes it for both.

splits · Blocking Where a D-optimal design puts its runs. The D-optimal measure over 121 candidate settings on a square region. It keeps 9 of them and discards the rest, and the 9 it keeps are the settings a catalogue would have offered without any of this arithmetic. What the search adds is the weights: 0.1458, 0.0962, 0.0802, which nine equal runs cannot express.

A design is a number

A standard design is taken from a catalogue and then measured. Turn the arithmetic round and a design becomes the answer to an optimisation — and over 121 candidate settings the search keeps nine of them, which are exactly the nine a catalogue would have offered, at weights nine equal runs cannot express.

optimality · Criterion The rule is parity, and it runs both ways. At a correlation of 0.5, four combinations of a dictionary and an outcome shape. The joint sign flip (X, Y) → (−X, −Y) leaves the bivariate normal alone at every correlation, so a function that changes sign under it is orthogonal to one that does not. A product of two odd functions is even; a product of an odd and an even one is odd. So an odd dictionary removes exactly none of the first and something of the second, and an even dictionary does the reverse — which it does, to machine precision, in both of the two rows that should be zero. This is one rule where there had been two: that a median split's square is constant, and that a polynomial dictionary contains the products a correlation generates.

A dictionary that is neither

A rule handed two median splits removes none of their interaction; a rule handed two covariates removes none of their product. Those were two results with two explanations, and they are one result with one — and finding it corrected the number underneath both.

dict · Criterion A proposal that moves more, refused more often. The two halves of the trade, both exact, on the 410 admissible assignments of twelve units. The integrated autocorrelation time of an imbalance the rule was never handed falls from 7.30 at one swap to 3.97 at three, and the acceptance rate falls with it, from 58.8% to 40.8%. A rejected proposal costs one evaluation and leaves the chain where it was, so acceptance is not the price of anything and the ranking by acceptance is the reverse of the ranking by cost. Past three the family folds: exchanging k of six from each arm is the complement of exchanging six − k, so k = 5 has the same 36 proposals as k = 1 and k = 6 has 1.

A proposal that moves more than two units

The walk's autocorrelation is a fact about its step size and not about its acceptance rate. Exchanging three units from each arm mixes nearly twice as fast as exchanging one, and is refused a third more often.

blocks · Randomisation Two designs for one model, and the weights are not equal. Where the runs go, for the same two-parameter model at K = 1 and a ceiling of T = 10. The D-optimal design for both parameters is the familiar one: half the runs at 0.8333 and half at the ceiling. The design for the half-saturation constant alone moves the lower setting down to 0.6040 — where the response curve is still bending, which is where K is visible — and, unlike every design in the two fields before this one, it does not split the runs evenly: the weights are exactly 1/√2 and 1 − 1/√2, 0.7071 and 0.2929, at every K, V and T. The equal weights of D-optimality were a consequence of asking about both parameters at once, and nobody had to notice while that was the only question being asked.

An efficiency that is a ratio

A design chosen for a model is not a design chosen for the parameter somebody wanted. Asking for one of two parameters moves the runs, unbalances the weights, and costs the other question exactly 15.07% — at every setting, because it is algebra.

guarantee · Criterion Every split of 100 units, σ = 1 against 3. Each point is one integer split, with its variance computed exactly rather than simulated. The minimum is at 25:75, which is the ratio of the spreads 25:75, and equal allocation costs 25% more variance — the same as throwing away 20 of the 100 units. The shaded band is every split within 5% of the best, and it runs from 17% to 35%: sharp to state, flat to sit on.

Not half and half

The same units, the same measurements, the same analysis — and a different variance, decided before anything is measured. When the two arms have different spreads the best split is σ₁ : σ₂, equal allocation costs 2(σ₁²+σ₂²)/(σ₁+σ₂)², and at three to one that is a quarter of the experiment.

allocation · Allocation A design that is right once, and one that is never wrong by much. Three designs for the Michaelis–Menten model, scored at every true value of K across a 16-fold range. The peaked curve is the two-point local design built at K = 1: 100% there and 66.7% at the worst point of the range. The curve just under it is the design that averages the criterion over a uniform prior on the same range, which is barely different — 67.9% at worst — because averaging is dominated by the middle of the range where the local design is already good. The flat line is the maximin design: 3 settings, never above 80.8% and never below 78.8%. Its worst case is 12.0 points better, and what it gives up is the 21.2 points at the one value the local design was built for.

The design for the worst case

A design for a non-linear model is optimal at a guess about the answer. Averaging over a prior repairs that on average; protecting the worst value in a range is a different problem, with a different answer, and it needs a third setting to reach it.

robust · Local design Four runs, and the term they cannot reach. Every run sits at a corner, so x₁² and x₂² are 1 at every run and both columns are copies of the intercept. The normal matrix is singular: the design has no information about curvature at all, and no analysis can recover it.

The design that cannot see a curve

A two-level factorial has every run at a corner, where every squared term equals one — so the column that would estimate curvature is a copy of the intercept, and the design has no information about it at all. A few runs at the centre buy one number back, and only one.

surface · Factorial The allocations the rule could have made, from these exact patients. One 200-patient trial allocated by response-adaptive randomisation, re-randomised 999 times. No outcome is redrawn anywhere in this figure: each re-randomisation runs the same rule over the same patients in the same order, so what is drawn is the set of experiments that could have happened rather than a sampling distribution. The observed |z| is 1.417, 258 of the 999 re-randomisations reach it, and the p-value is (1 + 258)/(1 + 999) = 0.2590. The curve is the half-normal the ordinary analysis reads the same statistic against; its 5% point is 1.96 and this distribution's is 2.101.

The experiments that could have happened

An adaptive trial's allocation is a function of the outcomes it will later be compared against, so the ordinary analysis rejects a true null 9.2% of the time. Hold the outcomes fixed, re-run the rule that assigned them, and count — the same statistic against a reference distribution the trial could actually have drawn from is back at 4.0%.

exact · Reference The end of the family is the member the algorithm cannot reach. Every member of the Φₚ family optimised on the same 11² candidates, and two eigenvalues of each answer. The upper curve is the smallest eigenvalue of the information matrix — the quantity E-optimality maximises — which rises from 0.0993 at the D end to 0.1994 at p = 64. The lower curve is the gap between that eigenvalue and the next one up, which falls from 0.0611 to 0.0011. A smallest eigenvalue is not differentiable where it is repeated, and the family is driving the gap to zero: the one criterion here whose meaning fits in a sentence is the one whose optimum sits on a corner of its own surface. The candidate grid is on a slider and it answers a narrower question than it looks. On this square region, refining an odd grid from seven to eleven moves nothing at all — the optimum's support is the corners, the edge midpoints and the centre, and every odd grid from five up contains all of them. An even grid has no centre point and cannot reach the answer at any member of the family. On a disc, where the boundary passes through no grid point, refinement does move it.

The family behind the letters

A, D and E are not three ideas. They are three points of one family with a single dial, and running the dial from one end to the other doubles the smallest eigenvalue of the information matrix while closing the gap above it fifty-three-fold — which is the family driving its own last member to the place where it stops being differentiable.

criteria · Criterion What is left of a probe after the rule has had it. The share of each dictionary function a rule balancing x, x2, x3, cut0 has already taken, on trials of 14 units, averaged over 100 designs. Four of the eight functions are the basis, so their share is exactly one: a randomisation test run on one of them is asking about a quantity the rule forced to zero, and one of them is the default probe of the field this measurement comes from. The four that are not still read 0.919, 0.873, 0.903, 0.832 — between 0.832 and 0.919 of them is inside the span — against closed-form removed shares of 0.000, 0.692, 0.590, 0.692. At 14 units a rule with four functions in it takes most of anything it is shown.

The part the rule already took

A diagnostic that reports on what a balancing rule was not handed is run through a column that is 92% inside the span the rule balanced — because orthogonality in the population is not orthogonality on fourteen units.

aimed · Randomisation The same 40 units, arranged two ways. Both designs estimate the same effect of 0.5 and both are unbiased — 0.488 and 0.497. The blocked design's estimate has standard deviation 0.318 against 0.692, a variance ratio of 0.21 where the model predicts 0.20.

The variance removed before the data

Arranging forty units in pairs rather than assigning them at random cuts the variance of the estimated effect to a fifth — and the fifth is knowable in advance, because it is exactly the share of the variance the pairs do not carry.

design · Blocking Where one rule becomes three. Every arrival in 200 simulated trials is put to all three scores, and the picture is how often they would send that patient to different arms. The range and the pairwise sum are the same rule at two arms and at three — for sorted counts the pairwise sum is twice the range, so the arm that minimises one minimises the other — and they part company at four, where the pairwise sum is 3(d − a) + (c − b) and the range still sees only d − a. The variance disagrees with both from two arms onwards, on 5.4% of arrivals at two and 27.0% at five, because the scores are summed over 3 factors and a sum of squares does not order the candidates the way a sum of absolute values does. All three are called minimisation.

Three arms and three scores

Minimisation balances a trial by keeping the arms' counts even inside every prognostic factor. With two arms there is one way to measure how uneven two counts are. With three there are several, they are all called minimisation, and they send different patients to different arms.

multiarm · Assignment What a mean split leaves, with both halves varying. The share of a mean split's interaction that survives the rule balancing it, at every copula and every marginal, matched at a Spearman correlation of 0.40. The three radially symmetric copulas leave exactly nothing with a symmetric covariate and rise steeply with the skew. The two asymmetric ones start at 7.707% and go opposite ways: the lower-tail copula falls to 0.002% at a skewness of 0.95 — the two failures cancel almost exactly, and a guarantee both fields report as broken is restored — while the upper-tail one climbs to 40.288%. And the heavy-tailed symmetric covariate, which leaks exactly nothing on its own, doubles what the asymmetric copulas leak: 14.229% against 7.707%.

Two failures that cancel

A mildly skewed covariate under a lower-tail copula leaks 0.002% of an interaction where each failure alone leaks eight and seven per cent. Turn the copula over and the same pair compounds.

compound · Adjustment What the rule blocks is not where it splits. How much of the separating direction each probe carries, over 100 designs of 14 units whose admissible set is enumerated and split into two pieces. The deferral this field answers proposed the constraint's active set — which exchanges the tolerance box actually blocks — as a better probe than the design's own leverage, on the ground that leverage is a heuristic and the active set is the quantity. Modelled from the design and the tolerance, it reads 0.4272 against leverage's 0.5395, at 4.43 paired standard errors the wrong way. Counted exactly over the enumerated set — at a cost no trial can pay — it reads 0.3854, worse again. Both beat a random direction at 0.2622, so they are probes; neither beats the two the earlier field already had.

What the rule blocks

A balancing rule breaks the admissible set into pieces by refusing exchanges. Which exchanges it refuses is computable from the design and the tolerance alone, before any assignment exists — and it makes a probe.

blocked · Randomisation One relationship at five designs, residual spread 1.00. Every panel has the same slope of 1, the same intercept of 0 and the same residual standard deviation of 1.00. Only the range of x differs. R-squared runs from 0.021 to 0.849, and the estimated residual spread is 0.9932 in all five.

R² is a property of the design

One line, one residual spread, five studies that differ only in how far apart they placed their x values. R² runs from 0.021 to 0.849 and the estimated residual spread is 0.993 in every one of them. Nothing about the relationship changed.

spread · Summary A model of the active set, and the active set. Each of the 14 units of one design, at the share of its exchanges the tolerance box blocks — computed from the design's columns and the tolerance under a uniform position in the box, against counted over all 116 admissible assignments. The diagonal is where the two would agree. Over 192 designs they agree about the ordering of the units at a correlation of 0.8141 ± 0.0112, negative on 0.5% of them, and disagree about the level: 0.8442 counted against 0.8170 modelled, a gap of 0.0272 ± 0.0051. An admissible assignment does not sit uniformly in its box, and this is the size of that.

A model and a count

The share of a unit's exchanges a tolerance box refuses can be modelled from the design or counted over the admissible set. They order the units the same way at a correlation of 0.81 and disagree about the level by 0.027.

blocked · Randomisation How far apart the two components are, on each probe. The median separation between the two components of the admissible set — the difference in their mean probe values, over the spread inside a component — over the 100 of 200 designs whose set is enumerated and found split. The separating direction carries 10.565 and needs the enumeration. The fourth power as the earlier fields use it carries 1.543; projected off the span the rule balances, 5.080. The design's own leverage, which uses no dictionary and no outcome, carries 3.836. A random direction in the same subspace carries 0.942, and a direction chosen by looking for concentrated structure carries 0.543 — below random, and the one heuristic here that is worse than not choosing at all.

A probe chosen from the design

The design's own leverage aligns with the separating direction four times better than a random direction in the same subspace. The concentrated direction the argument invites is worse than random.

aimed · Randomisation The same copula, turned over. A Clayton copula and its reflection, at the same Spearman correlation of 0.40 and the same Kendall tau of 0.275, against the covariate's marginal. With a symmetric covariate the two are the same number to nine decimals — 7.707% apiece — because the leak then depends on how much asymmetry the copula has and not on which way it points. Skew the covariate and they come apart: at a skewness of 2.26 the lower-tail copula leaves 3.431% and the upper-tail one 36.213%, a factor of 10.6. Both halves of the dependence are asymmetries and an asymmetry has a direction; a lower-tail copula concentrates the dependence where a right-skewed marginal is compressed and the two distortions partly undo each other, and an upper-tail one concentrates it where the marginal is stretched.

A symmetry that was not enough

A heavy-tailed symmetric covariate has a skewness of zero and leaks exactly nothing under three copulas. Under the two asymmetric ones it doubles the leak, from 7.707% to 14.229%.

compound · Adjustment A trial designed 2:1:1, and what two scores deliver. 500 trials of 180 patients, three arms, a target of 2:1:1. The shaded bars are a minimisation score that divides each arm's count by the share that arm is supposed to receive before measuring the spread; it delivers 49.9% : 25.1% : 25.1%. The others are the same rule with the counts left raw, which delivers 33.4% : 33.3% : 33.3% — the balance it enforces inside every factor level is equality, and equality is what it gets. The marks are the shares that were asked for.

Balancing towards unequal targets

A three-arm trial allocating two to one to one is the ordinary case, and a balancing rule built from raw counts does not know it. It balances the arms towards equality inside every factor level, delivers a third to each arm, and reports that it minimised imbalance.

multiarm · Allocation Four designs, scored on the one parameter that was wanted. Every design scored by its Ds-efficiency for K at 13 true values across a 16-fold range. The peaked curve is the subset design built at the guess K = 1: 100% there and 42.1% at the worst point of the range. The flat curve is the maximin-Ds design, never above 64.8% and never below 61.2%. Between them is the maximin design for the pair — a robust design, protecting something else, and worth 41.9% at worst here. The lowest curve is the D-optimal design at the guess, which is what an experimenter who wanted K and looked up a design for the model would actually run: 29.1% at the worst point, against 61.2% available.

Protecting one parameter over a range

A design for a non-linear model is optimal at a guess. A design for one of its parameters over a range of guesses is a worst case of a ratio of two determinants, and it is not a special case of either problem it is made of.

guarantee · Local design Stationary is not the same as convergent. How far each k-swap walk is from uniform after t steps, started at the least balanced admissible assignment of 410. Every one of these chains has a symmetric proposal and rejects by standing still, so every one of them is doubly stochastic and every one preserves the uniform distribution exactly. Only five of the six get there. Exchanging all six units of each arm is a single proposal — the complement — and the admissible set is closed under complement, so the walk takes it every time and oscillates between two assignments for ever: after 160 steps it has visited 1 state and sits 0.9976 from uniform. Its stationary distribution is a fact about the matrix; its limit does not exist.

Stationary is not convergent

A walk that exchanges every unit in each arm preserves the uniform distribution exactly and never gets near it. Every doubly stochastic matrix has the same stationary distribution; only some of them have a limit.

blocks · Randomisation Sheppard's arcsine, by two routes. Corr(sign X, sign Y) as the covariates' correlation runs from zero to one, drawn twice. One route is a sixty-four-node quadrature of the orthant probability over the correlation — the general construction, which works at any pair of cut points; the other is (2/π) arcsin ρ, which is elementary and works only at the median. They agree to 3.3e-16 at every one of 81 correlations, which is what licenses the quadrature everywhere else. The curve is above the diagonal at small ρ and below it at large: two signs agree with probability ½ + arcsin(ρ)/π, so a correlation of 0.5 gives exactly ⅓ and a correlation of 0.8 gives 0.5903.

The arcsine that closes it, and the error that was overstated

Two median splits of a correlated pair agree with probability ½ + arcsin(ρ)/π, exactly. And the truncation the field was avoiding falls geometrically in the correlation, not algebraically in the order.

splits · Routes A budget of 4,000, at 1 and 20 a unit. Every affordable pair, enumerated. The best is 280 cheap units and 186 expensive ones — a ratio of 1.51, against the σᵢ/√cᵢ rule's 1.49. The unit rule, which says buy in the ratio of the spreads, lands at 66:197 and costs 17% more variance for the same money. Both rules are right about their own constraint; only one of them was asked.

The cost of a unit

Change the constraint from units to money and the allocation rule changes with it — from σᵢ to σᵢ/√cᵢ, which can point the other way. An arm that is noisy and expensive gets fewer units than the same arm would if the money were not the thing running out.

allocation · Allocation Two rates, not a factor. The standard deviation of the covariate imbalance under three rules, at five trial sizes, 260 trials each, on log axes. The upper line is a coin: its slope is -0.489, against a closed form of exactly −½. The middle line is minimisation on a median split; its slope is -0.519 — the same rate — because inside a category the assignment is still a coin, and what it buys is the constant, 0.654 of a coin's at n = 200. The lower line is the rule that reads x and maximises the information about the treatment effect: slope -0.987, nearly twice as steep. Its advantage is therefore not a number that can be quoted — it is 0.258 of a coin's at n = 50 and 0.065 at n = 800, and it keeps going.

The rule that reads the number

Stop categorising and let the rule read the covariate itself. What it should minimise is not an invented distance but the variance of the effect being estimated — and what comes back is not a better constant but a different rate.

continuous · Assignment The allocations the rule could have made, from these exact patients. One 200-patient trial allocated by response-adaptive randomisation, re-randomised 999 times. No outcome is redrawn anywhere in this figure: each re-randomisation runs a fair coin rule over the same patients in the same order, so what is drawn is the set of experiments that could have happened rather than a sampling distribution. The observed |z| is 1.417, 155 of the 999 re-randomisations reach it, and the p-value is (1 + 155)/(1 + 999) = 0.1560. The curve is the half-normal the ordinary analysis reads the same statistic against; its 5% point is 1.96 and this distribution's is 1.946.

The test that needs the rule

A randomisation test assumes almost nothing about the data and one thing about the experiment. Tell it a fair coin produced an allocation that an adaptive rule produced — which is what every off-the-shelf permutation routine does — and it rejects 8.0% of true nulls where knowing the rule gives 4.0%.

exact · Reference The variance touches p and never crosses it. d(x) = f(x)′M⁻¹f(x) along the diagonal of a square region, for the D-optimal measure. The line at 6 is the number of parameters in the model. Kiefer and Wolfowitz's theorem says a design is D-optimal exactly when the largest d anywhere in the region is p — not approximately, equals — so the optimal curve is tangent to that line at its support points and below it everywhere else. Here the largest value anywhere on a 41×41 grid is 6.000000000.

The theorem that says when to stop

A search that maximises the volume of the information has no way of knowing it has finished, because nothing tells it what the maximum is. Kiefer and Wolfowitz's equality does — a design is D-optimal exactly when the worst prediction anywhere in the region equals the number of parameters, which is 6.000000000059 here, gated at machine precision.

optimality · Equivalence Twenty walks up the same hill, σ = 2. Each walk fits a plane to the same four-corner factorial, takes its gradient as a direction, and steps along it until a run comes in below the one before. The true optimum is the cross. 80% of the walks stop before the best point on their own path — not because the direction was wrong, but because one noisy run is enough to stop them, and the direction error costs only 3.9% of the available gain.

Walking up the gradient

The fitted gradient is wrong by an angle with a closed form, σ/(|β|√N), and what that angle costs is its squared cosine — twelve per cent at twenty degrees. What costs a third of the gain is not the direction at all. It is deciding where to stop.

surface · Optimum What the worst case is worth, one function at a time. The smallest share each dictionary removes, over seven outcome shapes, at a correlation of 0.5. A rule balancing the mean of each covariate has a worst case of exactly zero — against the square, and against both products. Adding a median split to it, which is the second thing every trial balances, leaves the worst case at exactly zero, because a median split is odd and so is a mean. Adding the square instead moves it to 6.8%, and the extra functions after that move it to 7.4%. The worst case is decided by which parities the dictionary contains rather than by how many functions are in it.

What the extra function buys

A rule balancing the mean of each covariate has a worst case of exactly zero. Adding the median split — the other thing every trial balances — leaves it at exactly zero, and one square moves it.

dict · Criterion Three copulas that break nothing, and a factor of two between them. The three radially symmetric copulas, at a matched Spearman correlation of 0.40, against the covariate's marginal. All three leave exactly nothing with a symmetric covariate — that is the guarantee, and it holds to twenty decimal places. What they do to a skewed covariate is not the same at all: at a skewness of 2.26 a Frank copula leaves 12.118% where a Gaussian leaves 21.539% and a t on four degrees of freedom leaves 23.640%. A factor of 2.0 between two copulas that are both symmetric, both matched on rank correlation, and both harmless on their own. So the copula matters to the marginal's leak without breaking any symmetry of its own, which is a milder version of the same finding and applies to every trial rather than to the asymmetric ones.

A copula that halves a marginal

Three copulas break nothing on their own and put a factor of two between the same skewed covariate's leaks — 12.118% under a Frank against 23.640% under a t, at the same rank correlation.

compound · Adjustment What each probe can see. How far apart the two components of the admissible set are on each probe, over the spread inside a component, on 100 designs whose set is enumerated and split. It is the population quantity a chain is trying to report. The separating direction itself reads 10.5646; the projected fourth power 5.0800, the design's own leverage 3.8362, the modelled active set 1.9529, the counted active set 2.0170 and a random direction in the same subspace 0.9422. The two active-set probes beat the random direction and lose to both of the earlier field's, which is the field's answer to the question that opened it.

A quantity that loses to a heuristic

Leverage is a heuristic about which units a balancing rule has most to say about. The constraint's active set is the thing the rule actually does. As a probe, the heuristic wins by 4.4 paired standard errors.

blocked · Randomisation The best of 8 arms, tested as though it were the only one. 8,000 trials with no effect in any arm. Stage one runs 8 arms at 60 each, the best is carried forward, and stage two adds 30 more to it and to the control. The histogram is where the final statistic lands and the curve is the standard normal it is being read against — shifted right, because the arm was chosen for being ahead. 10.5% of these trials clear 1.96 against a claimed 5%, and the value that actually holds the rate for this design is 2.313.

Dropping the losers

Carrying the best of eight arms forward and testing it at 1.96 rejects a true null 10.3% of the time — the hypothesis was chosen by looking at the data, so the statistic is a maximum wearing a single comparison's clothes. The value that holds the rate is 2.313, and it has to be solved for.

adaptive · Stopping Four criteria on 5 designs, on a square region. Each design scored as an efficiency — its value over the best attainable — so four criteria in as many different units sit on one scale where 1 is the optimum. Rows are ordered by D. D picks 13-run exchange; A picks face-centred composite; G picks 13-run exchange; I picks face-centred composite. Every design has been scaled to just fit the region first, because a design run at settings the region does not contain is not a competitor on it. The disagreement is the point: the letter is a choice, and it is almost never reported as one.

Four letters and two camps

D, A, G and I are four ways of turning one matrix into one number, and they do not agree. The design that wins on D is the worst thing here on I. And the same two designs swap places entirely when the region changes from a square to a disc — on all four criteria at once.

optimality · Criterion 3 arms against one control, 360 units in all. Every control size, enumerated. The best is 132 on the control and 76 on each arm — a ratio of 1.74, against √3 = 1.73. Splitting the units evenly over all 4 groups costs 7.2%, which is small; what the larger control also does is lower the correlation between the comparisons, from 0.50 to 0.37, and that changes which multiplicity correction is right.

One control, many arms

The control appears in every comparison, so it is worth √k treatment arms — and the same sharing makes the k tests correlated at n/(n+n₀), which is the quantity Bonferroni ignores. Both facts come out of one design decision, and it is the size of the control.

allocation · Multiplicity One rule keeps its promise and the other keeps its budget. Both stopping rules at five requirements, 1,500 experiments each, with a first stage of 5. The upper curve is the two-stage rule: 97.1%, 96.2%, 95.9%, 96.1%, 96.0% — at or above 95% at every point, which is a theorem rather than a tendency, because its interval is built from a spread estimated before the stopping point was chosen. It pays 2.06×, 2.01×, 2.00×, 1.99×, 1.99× the observations that knowing σ would need. The lower curve is the rule that re-estimates after every observation: 94.5%, 89.3%, 91.0%, 91.5%, 94.3%, on 0.98×, 0.87×, 0.88×, 0.93×, 0.96×. The second rule is the one anybody would run and the first is the one whose claim is true.

Stopping when it is precise enough

An experiment that runs until its estimate is precise enough is the natural design and the one with a theorem against it. Its two-stage cousin keeps its promise exactly, for every unknown spread, and pays twice the observations for it.

guarantee · Stopping Where to look depends on the answer. The information a single run at time t carries about the rate of an exponential decay, (∂η/∂θ)² = t²·exp(−2θt), at three values of θ. Each curve has one maximum and it is at t = 1/θ exactly — marked, and found by a search over 8,001 settings that was never told the formula. Nothing in a linear model behaves this way: there the information matrix is X′X and the parameters are not in it, so a design can be chosen once and used whatever the answer turns out to be. Here the design is optimal at a guess, and the three curves are three different experiments for one model.

The design that needs the answer

Every design this site has computed is optimal whatever the experiment turns out to say, because X′X does not contain the parameters. For a non-linear model it does, so the best place to take a measurement is a function of the number the measurement exists to find — and guessing it three times too low costs two and a half times more than guessing it three times too high.

criteria · Local design One experiment finding out where to look. A single run of the fully sequential design: 40 runs, the first 8 placed at the guess K = 1, then the model refitted and the design revised after every 2. The marks are the settings the runs were made at. The horizontal lines are where a design built at the truth K = 3 would have put them — 1.875 and 10.00 — and the rule walks onto them without being told: its estimate of K after the first eight runs was 2.694, and by the end 2.765 against a truth of 3. The whole experiment is 96.5% as efficient as the design that knew the answer, where running all 40 at the guess would have been 81.1%.

The design that stops guessing

Every repair so far protects a guess. The alternative is to run part of the experiment, estimate the parameter from it, and design the rest at the estimate — which recovers most of what a threefold wrong guess costs, and has a best moment to stop guessing that is earlier than anyone expects.

robust · Local design Whichever dial made the set thin, the crossing is at the same thinness. Each curve is one dictionary, swept over eight tolerances at two hundred units; a point above the line is a set thin enough that walking beats hunting. The curves lie nearly on top of one another, which is the answer to whether the crossing is a fact about the tolerance or about the thinness it produces: the crossings sit between one admissible assignment in 176 and one in 268 for dictionaries of 3 to 6 functions. The mechanism is that a hunt costs exactly 1/p and a walk costs almost the same everywhere — between 82 and 394 evaluations per usable draw across the whole table — so the crossing is wherever 1/p reaches a number that does not move.

The set a dictionary leaves

A rule constrained on six functions at a loose tolerance leaves a set as thin as one constrained on three at a tight one. Both sampling methods cross over at the same thinness, and the tolerance where that happens moves by a factor of three.

dict · Assignment The guarantee that survives a correlation, and the one that does not. What a balancing rule handed both main effects removes of the pure interaction between them, as the covariates become dependent. For median splits it is exactly zero at every correlation, because sign(x)² = 1: the interaction sign(X)sign(Y) is orthogonal to sign(X) and to sign(Y) whatever ρ is. For the product of the raw covariates it is 4ρ²/(1+ρ²)² — 64.00% by ρ = 0.5, rising to all of it at perfect correlation. A cut away from the median sits between them and is not small: 23.01% at a cut of one. The zero is not a fact about interactions. It is a fact about a dictionary whose functions square to a constant, which a polynomial one does not.

The zero that survives a cut

A rule holding both main effects removes half of a pure interaction between correlated powers and exactly none between correlated median splits. The guarantee that a correlation destroyed was never about interactions.

splits · Criterion A central composite design, 13 runs. Adding 4 axial runs at ±√2 gives every factor three levels, which is the least that can estimate a squared term. The normal matrix now inverts, so each βᵢᵢ has an estimate of its own — and at exactly this axial distance the design is rotatable, which the next figure measures.

Three levels, and the ring where the design says the same thing

A central composite design puts its axial runs at ±α, and α is not a matter of taste. At F to the quarter the prediction variance depends only on how far a point is from the centre and not at all on which direction it lies in — a property with no simulation in it, exact or absent.

surface · Factorial How often each probe finds a split that is there. The share of 34 designs — every one of them enumerated to be in two components — on which a two-chain test of 800 draws declares the split, by probe. The fourth power as the earlier fields use it finds it on 55.9%, so it misses 44.1% of the sets that have one. The same column projected off the rule's span finds it on 88.2%, and the separating direction itself on 91.2%. The design's own leverage, chosen without any dictionary, gets 79.4%. A random direction in the same subspace gets 44.1%, and the direction chosen for being concentrated gets 38.2% — worse than random, which is what a heuristic that finds the wrong structure looks like from the outside.

What a chosen probe finds

On a chain of eight hundred draws the probe the earlier fields use misses 44% of the sets that are split. Its own residual off the rule's span misses 12%, for one least-squares fit.

aimed · Randomisation Three analyses of the same trials, none of them wrong about the data. 320 trials at n = 60 with no treatment effect at all, so every rejection counted is a false one, and a covariate that drives the outcome with coefficient 1. The unadjusted comparison is at 5.94% after a coin — its level — and at 0.00% after the rule that reads the covariate: the design removed the imbalance and the analysis is still pricing it. Adjusting for the covariate gives 4.06%, and the rule's own reference distribution — hold the outcomes, re-run the rule 199 times, count — gives 3.13% against the 4.5% that 199 draws can deliver. The last of the three has to be told the assignment rule and nothing else, which is the one thing the experimenter certainly knows.

What the balanced trial is worth

A rule that reads the covariate removes three quarters of the imbalance. An analysis that does not know it happened prices the imbalance anyway, rejects one true null in two hundred instead of one in twenty, and finds a real effect less often than a coin-tossed trial does.

continuous · Randomisation The crossing barely moves. Both methods' costs in one unit — assignments evaluated per usable draw — as the tolerance tightens. A hunt costs 1/p and rises without limit: from 2.22 at a tolerance of 1.2 to 357.14 at 0.18. A walk costs its autocorrelation time and barely moves. The two cross at a tolerance of 0.190 at one swap and 0.195 at eight — the whole family of proposal sizes crosses inside a band of about two hundredths, because where the crossing is, the large proposal has already lost its advantage. A multi-swap proposal is worth a factor of 5.65 in the regime where the walk should not be used at all.

Where the gain is, and where the decision is

A bigger proposal is worth a factor of six at a loose tolerance and nothing at a tight one. The tolerances where it helps are the ones where a hunt costs two evaluations a draw, and the crossing barely moves.

blocks · Assignment What a design does to the residuals of a correct model. Every residual has standard deviation sigma times the square root of one minus its leverage. On this design the leverages run from 0.045 to 0.663, so the residual spreads differ by a factor of 1.68 — and the model is exactly right. The high-leverage point's residual averages 0.46 of the fitted spread where a typical point's averages 0.79.

Residuals are not the errors

A residual's standard deviation is σ√(1 − hᵢᵢ), so a design whose leverages run from 0.045 to 0.663 produces residuals whose spreads differ by a factor of 1.68 with the model exactly right. On the samples where the high-leverage point really did have the largest error, a raw residual plot shows it as the largest on 0.0% of them.

lineup · Qq A wrong weight costs width; a random weight costs level. Five weightings on a trial whose variance ratio drifts by a factor of 20.1 between the first block and the last, over 4000 runs. The rule that knows every λ_b covers at 95.1% and sets the width. One ratio for the whole trial is wrong for every block and costs nothing in level — 94.8% — while being 20% wider; equal weights are calibrated by an identity and 22% wider. The ratio estimated inside each block is the only rule aimed at the quantity that actually varies, and it is the only one that misses the level, at 92.0%: a weight computed from a handful of degrees of freedom is mostly noise, and noise in a weight is not a wrong weight. Modelling the drift across blocks recovers the oracle's width at 94.8%.

A ratio that changes between blocks

A wrong weight costs width and a random weight costs level. The rule aimed at the quantity that actually varies is the only one that misses its own coverage, and the rule that models it across blocks recovers the whole of what knowing it is worth.

blocks · Nuisance What a pilot buys, σ = 1 against 3. Each point is 6,000 two-stage experiments of 100 units: a pilot of m per arm, then the rest split by the pilot's own estimate of the two spreads. Above the line the pilot has made the experiment worse than not bothering. The best pilot here is 8 per arm at 0.809, against 0.800 for a designer who knew the spreads — so the rule recovers 96% of what knowing them is worth. A larger pilot estimates the ratio better and has less left to apply it to, which is why the curve turns.

Allocating on a guess

Every allocation rule in this field is a function of quantities the experiment is being run to find out. Fed a pilot's estimate of them, the rule that minimises the variance makes the experiment worse than not bothering — until the arms differ by about a factor of two, which is further than anyone would guess.

allocation · Allocation What balancing several numbers at once costs each of them. The criterion generalises without a word changing — the covariate imbalance becomes a vector and the correction a quadratic form — so the question is what it is worth rather than whether it can be done. At n = 200 with 200 trials per point, a rule balancing one covariate leaves 12.7% of a coin's imbalance in it; balancing eight leaves 23.2% in each. The assignment has a fixed amount of freedom and every covariate added takes a share of it. The rule degrades rather than failing: at eight covariates it is still four times better balanced than a coin, and the eight are being held simultaneously rather than in turn.

Balancing more than one number

The criterion generalises to several covariates without a word changing, which makes the question what it is worth rather than whether it can be done. Each one added takes a share of the assignment's freedom, and the imbalance left in every one of them rises.

continuous · Blocking One of them is mostly leverage. How much of the design's own leverage direction each active-set probe carries, once both are standardised and projected off the rule's span — which is what a probe is, so it is the comparison that matters. Over 189 designs the modelled active set agrees with leverage at |r| = 0.8359 ± 0.0114 and the counted one at 0.4239 ± 0.0216. So the modelled probe is largely leverage under another name and the counted one is genuinely a different direction — and the counted one is the worse probe, at 0.3854 of alignment against 0.4272. What the active set contains beyond leverage points away from where the set splits.

Counting it exactly does not help

If a modelled active set lost because the model was crude, the exact one would win. It is computed at a cost no trial can pay, and it is worse — so the approximation was never what was costing the probe.

blocked · Randomisation What a guesser gets, and what a guesser gets for nothing. 600 trials of 150 patients under a fully deterministic rule, with an investigator who knows the rule, the factors and every assignment so far. The guess rate falls with the number of arms — 87.8% at 2, 86.2% at 3, 81.0% at 4 — which reads like a trial getting safer and is not: what a guesser can trade on is the excess over the 50%, 33%, 25% they would get by naming an arm at random, and that goes the other way, from 1.76× chance at two arms to 3.24× at 4. The gap a guesser manufactures between the best and worst arm under a true null is 0.759, 0.733, 0.711 standard deviations — nearly unchanged.

Guessing one arm in three

A balancing rule is guessable because it is balancing. With three arms the next assignment is worked out less often than with two — and by more, relative to what a guesser gets for nothing, and the damage they can do is almost unchanged.

multiarm · Assignment Adding a run can make the design worse. The D-efficiency of the best N-run design at each size, against the optimal measure. It is not a rising curve. 13 runs reaches 99.77% and 14 falls to 99.44%, because the optimal weights are real numbers and N runs is an integer approximation to them, so how good a design can be depends on how well N divides. A Wald interval behaves the same way: a larger sample sometimes makes its coverage worse, for exactly this reason.

The design that has to be integers

The optimal design is a set of real weights and an experiment is a set of runs, so the theory's answer is never available. Thirteen runs reach 99.77% of it and fourteen reach 99.44% — adding a run makes the design worse per run, and the search that finds it does not always find the same one.

optimality · Criterion A design that is right once, against one that is never wrong by much. Two designs for the same two-parameter model, scored at every true value of K across a range of 16-fold. The local design is the two-point optimum for a guess of K = 1: it reaches 100% there and 66.7% at the worst point of the range. The hedged design maximises the average of log|M| over a prior spanning a factor of 4 either side, uses 3 settings rather than two, and is never below 75.4%. What it costs is 10.4 points at the one value the local design was built for — which is the whole trade, and it is only available to somebody willing to say how wrong the guess might be.

The design that hedges

A locally optimal design is right at one value of the unknown and 23.9% efficient at the edge of a sixteenfold range. Averaging the criterion over a prior instead buys the worst case back to 56.3% — and buys it by adding support points, at spreads the arithmetic decides rather than the experimenter — a third setting at a factor of 3.36 and a fourth at 8.86.

criteria · Local design Four intervals at one stopping time. 3,500 experiments under the sequential rule with a first stage of 5 and a required half-width of 0.4, which is a demand that knowing σ would meet with 24.0 observations and which the rule meets with 20.8. The rule's own interval covers 90.3%. Replacing the fixed width by a t interval on the same data gives 92.0%. Keeping the rule's own random sample size and drawing a fresh sample of that size gives 89.8% at the fixed width and 95.5% for a t interval — so the sample size being random costs nothing, and the sample size being chosen by the data the interval is built from costs the rest. The spread estimated at the stopping moment is 17.1% below the truth, which is the same fact one level down.

The interval after a stop it chose

A rule that stops when the estimated precision is good enough stops on the samples whose estimate was small. Its interval covers 90% and claims 95%, and a fresh sample of the same random size covers 95.4%.

guarantee · Stopping The one zero neither half of the dependence can touch. A median split's interaction leak at all 30 combinations of copula and marginal, on a log scale. Every one is under 10⁻¹⁶ and the largest is 1.74e-20, which is the quadrature's own noise rather than a leak. The reason is arithmetic and it is short: a centred median split takes the values ±½, so its square is a quarter identically — for every unit, on every draw, whatever the covariate's scale is and whatever joint law the ranks have. The interaction is then orthogonal to both main effects by construction, and there is nothing for either half of the dependence to break. Both of the fields this one joins report this zero holding under their own variation; running both variations at once is what establishes that it is not two coincidences.

The zero that survives both

A median split's interaction leak is under 10⁻¹⁶ at all thirty combinations of copula and marginal. It is the only guarantee in the collection that neither half of the dependence can touch.

compound · Adjustment The cheap repair needs a number nobody has. The obvious alternative to re-randomising is to simulate the design under its null once and use the critical value that comes out — which is what the arm-dropping design does, where the critical value has to be solved for and is 2.313. It does not transfer here. The rule chases outcomes, so how imbalanced the allocation gets depends on how often anything succeeds, and the critical value moves from 1.668 at a success rate of 0.05 to 2.718 at 0.8. Calibrated at 0.3 and used at 0.8 the test's real size is 12.4%; used at 0.05 it is 0.12%. The randomisation test needs none of this, because it conditions on the outcomes that happened rather than on a rate they were supposed to come from.

What the exactness buys

Against a z test calibrated to reject exactly 5% of true nulls on this design, the randomisation test loses nineteen points of power. What it buys is that the calibration needs the success rate — which moves the critical value from 1.668 to 2.718 and is the quantity the trial was run to find out.

exact · Nuisance Where the constraints exhaust the randomisation. At 16 units there are 12,870 equal splits, so the ones meeting a stated tolerance can be counted rather than estimated. With each of the first k standardised imbalances required to be within 0.4 of a coin's own spread, the admissible count runs 3874 → 1006 → 314 → 0 → 0 → 0 — and at 4 functions there is no admissible assignment at all. The count is the number of distinct answers a randomisation test can give: at 3 functions its finest attainable p-value is 1 in 314. Balance improves with every constraint and the reference distribution shrinks with it, and the two run out at different rates.

When the constraints run out

Every function added to a basis is a constraint the assignment has to satisfy with the same units. At sixteen units and a stated tolerance the admissible assignments run 3,874, then 1,006, then 314, then none — and the count is exact, because the assignment space is finite.

basis · Allocation A discrepancy of 4 σ at the top of the range: one bad run, or a line that bends there, in two designs of twelve runs. The same twelve errors in every panel. With one run at the top (upper panels) a bad observation and a bent line produce identical data, point for point. With three runs at the top (lower panels) the bad observation leaves one run far from its two neighbours and the bent line moves all three together. The dashed line is the true line; the grey line is the least-squares fit.

Three runs at the end of the line

With one run at the top of a regression's range, a bad observation there and a line that bends there produce data with exactly the same distribution, so no residual, influence measure or test can say which happened. Move six of twelve runs to the ends, three at each, and a discrepancy of four σ is named correctly as a bad run 84.1% of the time and as a bend 91.6% — while the slope's standard error falls from 0.0836 to 0.0709. The design, not the diagnostic, decides whether the question has an answer.

lineup · Qq Where the bias lands. The drift in the log variance ratio, fitted across 12 blocks over 4000 trials. E[log λ̂_b] is log λ_b plus ψ(k_B/2) − log(k_B/2) − ψ(k_A/2) + log(k_A/2), which depends on nothing but the degrees of freedom — so the tempting sentence is that it goes into the intercept and leaves the slope alone. It does not, because the blocks alternate between allocations and the alternation is correlated with the covariate being fitted: the lopsided blocks carry 0.5383 of bias and the even ones carry none. Uncorrected the slope reads 1.5597 against a truth of 1.5, which is 8.0 standard errors. Subtracting the two digammas block by block leaves 1.4976.

The bias that lands in the slope

The bias in a log variance estimate depends on nothing but its degrees of freedom, so it goes into the intercept — unless the degrees of freedom alternate with the design, which is exactly what a block-randomised trial makes them do.

blocks · Width The pairing recovers most of it and passes nothing. How much of the separating direction each probe carries, over 100 designs of 14 units whose admissible set is enumerated and split in two. The four rows the pairing adds are the dominant direction of what each blocking matrix keeps past its degrees, and the cut that direction's signs induce. Counted, they read 0.5266 and 0.5258 against the counted per-unit share's 0.3854 — most of the gap between that share and the design's own leverage at 0.5395, closed. Modelled, they read 0.4274 and 0.4954 against 0.4272. Nothing built from the active set passes leverage, and the projected fourth power is still ahead of all of them at 0.6583.

A set of pairs, not a vector

The active set is a graph on the units, and every probe built from it so far has been its degree. Read as a graph it recovers 0.1326 of the alignment the summary lost — and draws level with leverage rather than passing it.

blocked · Randomisation What a 8-run fraction of 4 factors confounds. The defining relation is I = ABCD, so the resolution is 4. A is estimated as A + BCD; B is estimated as B + ACD; C is estimated as C + ABD; D is estimated as D + ABC. Each of those is an identity about the design rather than an approximation about the data.

The word a fraction costs

A half fraction estimates each main effect as an exact sum of that effect and everything it is confounded with — no error term, no sample-size argument. With every interaction at 0.8 the design reports a true effect of −1 as −0.20, and the design cannot test the assumption that makes the number mean anything.

design · Factorial What visiting fewer settings costs, 6 parameters. Carathéodory's bound puts the support of an optimal measure between 6 and 21. 6 settings: D-efficiency 88.90%, G-efficiency 57.18%, 0 degrees of freedom for lack of fit; 7 settings: D-efficiency 94.54%, G-efficiency 61.22%, 1 degrees of freedom for lack of fit; 8 settings: D-efficiency 95.99%, G-efficiency 64.60%, 2 degrees of freedom for lack of fit; 9 settings: D-efficiency 97.40%, G-efficiency 82.76%, 3 degrees of freedom for lack of fit. The saturated design has none, and buying the first one costs about five points of efficiency to get back.

How many places a design goes

Carathéodory's bound puts an optimal design's support between six and twenty-one settings, and every design in this field that can fit the model visits exactly nine. The count is not a choice anybody makes, it decides how many degrees of freedom are left to check the model with, and the first spare setting costs six points of efficiency to get back.

optimality · Equivalence What the fit calls the shape, against what it is. One eigenvalue held at −3 and the other swept from −2 to 2, so the truth is a maximum on the left and a saddle on the right and the change happens at exactly zero. At an eigenvalue of −0.25 — a genuine maximum — the fit reports a saddle on 26.4% of studies; at +0.25 — a genuine saddle — it reports a maximum on 25.1%. The standard error of a squared coefficient under this design is 0.3791, and the region of confusion is about that wide either side of zero.

The sign the curvature has

A fitted surface reports a maximum, a minimum or a saddle, and the report is a comparison of two estimated eigenvalues against zero. At a true second eigenvalue of −0.25 the fit calls a genuine maximum a saddle on 26.4% of studies, and at +0.25 it calls a genuine saddle a maximum on 25.1%.

surface · Optimum An exact test rejecting a true hypothesis a fifth of the time. How often each analysis reports an effect when the average treatment effect is exactly zero and the effect varies between units, at 150 units with 25% treated. The permutation test on the difference in means reads 4.20% where the effect is constant — where the two nulls coincide and its exactness applies — and 22.93% where the effect varies with a standard deviation of 3. The same test on the studentised difference reads 6.27% there, and the ordinary large-sample t, which makes no exactness claim at all, reads 6.60%.

The null the exactness is for

A permutation test is exact under the hypothesis that the treatment changed nothing for anybody. Under the hypothesis it changed nothing on average, with a quarter of the units treated and the effect varying between them, it rejects a true null 22.93% of the time.

exact · Nuisance What the guess is worth, when it is worth anything. The variance cost of an even split relative to the variance-minimising one for a risk difference, against the first arm's proportion, with the second at 0.3. The cost is a pure number: it does not depend on the trial's size. It is exactly zero at 0.3 and at 0.70, where the two arms have the same p(1 − p); it is 0.19% at a half and 4.36% at a tenth. Across the whole range from a tenth to nine tenths it never exceeds 4.36%, which is what the variance-minimising rule is worth here — and what it is worth is the reason it is safe to use with a guess.

The arm whose variance is its answer

With a binary outcome the allocation rule is a function of the proportions the trial exists to estimate. It costs at most 4.36% of variance to ignore it anywhere between a tenth and nine tenths, because √(p(1−p)) stays within a factor of two of its peak across 98% of the unit interval.

allocation · Allocation How many of 100 points a fitted slope rests on, by the distribution the covariate was drawn from. The effective number of points behind a slope is n divided by the design's kurtosis. Over designs of 100, its median share is two points, half at each: 1.000; uniform: 0.551; normal: 0.349; exponential: 0.160; Student's t, 3 df: 0.148; lognormal, σ = 1: 0.076 — 34.9 of 100 points for a normal covariate and 7.6 for a lognormal one.

The points a slope rests on

Before any outcome is measured, a regression's design already says how many of its points the slope will rest on: n divided by the covariate's kurtosis, exactly. A hundred points of a normal covariate carry a slope on about 35 of them; a hundred of a lognormal one, on 7.6. The largest leverage follows the same tail — above one half in 15.7% of lognormal designs of a hundred and in none of the normal ones — and on a correct straight line, deleting the single highest-leverage point moves the slope by more than its standard error in 16.3% of lognormal datasets of a hundred and 0.1% of normal ones.

spread · Summary The same number of assignments, held less tightly. For each basis size the box admits at a tolerance of 0.4, the Mahalanobis rule is given exactly the same number of admissible assignments — the ones with the smallest Mahalanobis distance — so the two differ only in the shape of the region. Each bar is the worst named function's remaining imbalance variance as a share of a coin's, with the mean over the named functions beside it. At one function the regions are the same interval: 0.0523 both ways. At 3 functions and 314 assignments, the box leaves 0.0771 in its worst function and the ellipse 0.1545; on average 0.0633 against 0.0927. The ellipse spends its assignments on a direction no function in the basis names.

What the ellipse spends its assignments on

Rerandomising on Mahalanobis distance never runs out of assignments, where a box of per-function tolerances does at four functions. But at the same number of admissible assignments it holds the worst named function half as tightly, because it spends its constraint on a combination of the basis nobody named.

basis · Allocation A basis drawn rather than chosen. The share of the worst of seven shapes a balancing rule removes, over two covariates and a dictionary of twenty-four functions, for the best fixed basis of each size (lower line) and for a basis drawn at random from the least favourable mixture (upper line), at one to four functions. The fixed basis guarantees 0.0000, 0.0000, 0.4244, 0.6366: nothing at one function or two, because two of the shapes are pure interactions and two of them are main effects of different covariates, and no one or two functions touch all of them. The drawn basis guarantees 0.2042, 0.4084, 0.6133, 0.7920 in expectation over the draw — about a fifth of a shape for every function up to three. The marks above the upper line are the best reply to nature's mixture, 0.2051, 0.4098, 0.6142, 0.7931: the game's value lies between the two.

A basis drawn rather than chosen

On two covariates no fixed basis of one or two functions protects every shape an outcome might have — its worst case is exactly zero. Draw the basis at random from the right mixture and the guarantee is a fifth of a shape for every function. The price is per trial — at three functions, two draws in five leave one shape wholly unprotected.

product · Optimum The Box–Behnken design in three factors: 15 runs, none at a corner. Twelve runs at the midpoints of the cube's edges and 3 at its centre. Every run holds one factor at zero, so no run puts all three factors at an extreme — which is what makes it runnable where a corner is not. The three panels are the design's coordinate projections, with repeated positions marked.

The design that refuses the corners

Box–Behnken runs three factors in fifteen runs and puts none of them at a corner, which is what makes it usable where a corner cannot be run. It predicts the corner 1.84 times worse than the seventeen-run design that goes there, and 1.31 times worse at the middle of a face, and all three numbers are matrix computations with no simulation in them.

design · Factorial The certificate when 4 runs are already spent. a 2² factorial has already been run and 2 further runs are to be placed. The stationarity condition is no longer max d = p; it is max d = (p − λ·tr(M⁻¹M_fixed))/(1 − λ) with λ = 0.6667 the share of runs already spent, which is 7.0985 here. The search reaches 7.098508790 against it, and the largest value anywhere on a 41×41 grid is 7.098508790.

Augmenting a design that has already run

The equivalence theorem still certifies when some runs are already spent, and one number in it changes: the bound is no longer p but (p − λ·tr(M⁻¹M_fixed))/(1 − λ). It equals p again exactly when the runs already made can still be absorbed into the design that would have been chosen — so the certificate says whether the experiment is still recoverable.

optimality · Equivalence The ridge, when the fitted optimum is outside the region. One fitted surface. Its stationary point is at a radius of 2.289 and the fit calls the shape a maximum. The ridge is the best setting at each radius, found by the Lagrange condition (B̂ − μI)x = −ĝ/2; the fitted response rises along it from 59.93 at the centre to 62.10 at the edge. The true optimum is at (0.4, 0.3).

When the best setting is outside the region

On a flat surface at twice the noise the fitted optimum lands outside the experimental region on 24.9% of studies and more than three coded units out on 11.8%. The answer is a ridge — the best setting at each radius, with a closed form — and the two obvious rules for using it turn out to be within four per cent of each other.

surface · Optimum Three contrasts on one dataset, three different splits. The variance-minimising allocation for each of three ways of reporting the same two-arm comparison, against the first arm's proportion, with the second at 0.1. A risk difference wants the arm with the larger p(1 − p) to get more units; a log odds ratio wants it to get fewer, and the two curves are exact reflections of each other in the half line. A log risk ratio wants something else again. At a first-arm proportion of 0.6 they ask for 62.0%, 21.4% and 38.0% of the units. A trial reporting more than one of them cannot be optimal for either.

Two contrasts, one split

A risk difference wants 62.0% of the units in the first arm, a log risk ratio wants 21.4% and a log odds ratio wants 38.0% — on one dataset, with one pair of proportions. The difference's rule and the odds ratio's are exact reflections of each other, so no split can be near-optimal for both.

allocation · Allocation The variance of the baseline-adjusted estimate for four arrangements of visits, against the correlation between visits. At a correlation of 0.5 between visits, the adjusted analysis has 0.750 of the unadjusted variance with one visit before treatment and one after, 0.667 with two before, 0.500 with two after and 0.417 with two of each.

A visit before or a visit after

A trial that can afford one more measurement per patient can take it before treatment, to sharpen the baseline adjustment, or after it, to average the outcome. With every pair of visits correlated 0.5, the second visit after treatment removes a quarter of the variance and the second visit before it removes a twelfth, and it is never the other way round: the baseline's contribution is ρ²(1 − ρ)/(1 + ρ), which peaks at ρ⁵ = 0.090 when ρ is the golden ratio's reciprocal. Below a correlation of one half, a single extra follow-up beats any number of baselines. When the correlation fades with the time between visits, averaging an earlier baseline into the adjustment makes the trial less precise, not more.

planned · Power A point that decides a coefficient, on two covariates correlated at 0.9. A hundred points on two normal covariates with correlation 0.9. Dashed lines mark where the most extreme five per cent of each covariate begin. The point carrying the largest share of the first coefficient's information, 14.2%, is inside the lines on both axes, so neither covariate's plot on its own would show it. The coefficient rests on 26.1 effective points, with a variance inflation of 6.27.

A point ordinary on every axis

In a multiple regression each coefficient rests on its own number of points, n divided by the kurtosis of that covariate's residual after the others, exactly. Correlation between normal covariates inflates a coefficient's variance fivefold and leaves it resting on the same third of the points. What correlation does instead is hide the point that decides the coefficient: at a correlation of 0.9 it stands out on no single covariate's plot in 77.3% of designs of a hundred, and the largest leverage a normal design reaches depends on the number of points and covariates and on nothing else.

spread · Summary Separation, how far against whether. The median separation each probe carries between the pieces of a split admissible set, over the 100 designs whose set is split. The share refused per unit carries 1.953; the refusal pairing's direction 3.103, and with the saturated pairs dropped 2.917. The depth pairing's direction carries 2.345, and each unit's mean depth 3.836 — exactly the leverage's. A random direction carries 0.942 and the projected fourth power 5.080.

How far, not whether

The blocking matrix records whether a tolerance box refuses an exchange. Dropping the pairs it refuses for certain changes nothing. Recording instead how far each exchange moves the balance vector turns the pairing into its degree plus a part that lies inside the rule's own span — and the degree is the design's leverage, exactly. The heuristic that beat every reading of what the rule blocks was the unclamped pairing's degree all along.

blocked · Randomisation What a stated power leaves of the main effects. The main-effect efficiency of the best design reaching a power of 0.8 at level 0.05, as N|δ|² rises through 50, 60, 70, 80, 100, 130, 170, 220, 300, 400, 600, 1000. a saddle, two-degree test, the compound path: out of reach, out of reach, 0.806, 0.841, 0.881, 0.914, 0.937, 0.952, 0.966, 0.975, 0.984, 0.990. a bowl, two-degree test, the compound path: 0.782, 0.823, 0.851, 0.871, 0.898, 0.923, 0.941, 0.955, 0.967, 0.976, 0.984, 0.990. a bowl, its own one-degree contrast, the compound path: 0.827, 0.859, 0.880, 0.896, 0.918, 0.937, 0.952, 0.964, 0.973, 0.980, 0.987, 0.992. a bowl, a factorial with centre points: 0.914, 0.930, 0.940, 0.948, 0.959, 0.969, 0.976, 0.982, 0.987, 0.990, 0.993, 0.996.

A design chosen for a test

A screening experiment that wants its main effects and a check on curvature has been told to trade them with a weight nobody can state. It can state a power. Read against a power of 80%, the frontier collapses to one curve in runs times the squared size of the curvature, and a large enough trial pays almost nothing. But the curve belongs to a test. A factorial with a few centre points beats the entire frontier on a bowl, spending 4% of its main-effect information where the frontier spends 10%, and against a saddle of the same size it has exactly its 5% level.

criteria · Criterion What one lost run costs a 16-run factorial fitting 11 coefficients. Every run is worth the same: dropping any one multiplies every coefficient's variance by 1.2000, which is 1 + 1/(N − p) with N = 16 and p = 11, and gives every pair of coefficients a correlation of 0.1667 where the complete design had exactly zero.

The run that did not happen

Lose one run from any orthogonal design and every coefficient's variance is multiplied by exactly 1 + 1/(N − p), and every pair of coefficients acquires a correlation of exactly 1/(N − p + 1) where there was none. The price is set by the design's spare capacity and by nothing else, and a saturated design cannot survive it at all.

design · Factorial Where each criterion's optimum puts the information. the D-optimal design's smallest eigenvalue is 0.09927, attained once; the A-optimal design's smallest eigenvalue is 0.16516, attained once; the I-optimal design's smallest eigenvalue is 0.17541, attained once; the E-optimal design's smallest eigenvalue is 0.19999, attained 2 times. A criterion that reads the smallest eigenvalue has no derivative where that eigenvalue is repeated, and the E-optimal design is exactly there.

The criterion with no derivative

E-optimality maximises the smallest eigenvalue of the information matrix, and at its own optimum that eigenvalue is attained twice — which is exactly where the function has a corner. The multiplicative search this field's other three criteria use assumes a derivative that is not there, and stops at 37.2% of the optimum.

optimality · Equivalence What a confirmation run at the chosen setting would find. The true optimum is worth 62.348. At σ = 2 the fit predicts 62.679 at the setting it recommends and the truth there is 61.821 — a gap of 0.858, which is 0.72 of the prediction's own standard error. The setting itself gives up 0.527 against the best available.

The run that confirms it

The setting a response-surface analysis recommends was chosen because the fitted surface was highest there, so the height the fit predicts at it is a maximum over a random field. At twice the noise the fit predicts 0.858 more than is there — 0.72 of the prediction's own standard error — and the gap is not noise, it is the selection.

surface · Optimum The extra cost of each follow-up schedule over the cheapest, across every correlation curve with adjacent visits at 0.7, when a patient costs 5 visits to recruit. With one baseline, 1 follow-up costs at most 28.0% more than the cheapest schedule, 2 follow-ups cost at most 11.1% more than the cheapest schedule, 3 follow-ups cost at most 24.1% more than the cheapest schedule, 4 follow-ups cost at most 34.8% more than the cheapest schedule, over every stable share from 0 to 0.7. The schedule whose worst case is smallest is 2 follow-ups; its worst case is at a stable share of 0.00. Under compound symmetry the cheapest is 3; under a stable share of 0.4 it is 1.

The correlation a pilot can see

Two models of repeated visits agree that adjacent visits correlate 0.7 and disagree about how many follow-ups a trial should buy: three under equal correlation, one when the correlation fades. Planning on the wrong one costs 28.0% or 9.9%. Two follow-ups cost at most 3.2% more than the best under either, and at most 11.1% under any curve with that adjacent correlation. A pilot of twenty patients at four visits chooses worse on average than that rule when the correlation fades; it takes about forty to draw level with it, and at an effect of half a standard deviation those forty patients' extra visits cost fifty times what the better schedule saves.

planned · Power A point that decides an interaction, on two covariates correlated at 0. A hundred points on two normal covariates with correlation 0. Dashed lines mark where the most extreme five per cent of each covariate begin. The point carrying the largest share of the interaction's information, 10.2%, is inside the lines on both axes, so neither covariate's plot on its own would show it. The coefficient rests on 19.0 effective points.

A term built from the others

A regression's squared terms and interactions are columns built from its covariates, and each coefficient on one rests on the number of points divided by the kurtosis of what the built column adds — fifteen for the square of a normal covariate, nine for the product of two uncorrelated ones, three for a treatment's interaction with a normal covariate, which is the same as the covariate's own coefficient. A design of a hundred rests on more than the population says, 12.2 points behind a square against 6.7, because a hundred draws rarely hold the tail that sets the kurtosis. Centring changes nothing about the square's coefficient, to the last digit. And correlation between two covariates hides the point that decides each one's own coefficient while it reveals the point that decides their interaction.

spread · Summary What a trial's own first patients buy when they choose the follow-up schedule for the rest, powered for an effect of 0.2. Internal pilots measured on two follow-ups, the schedule that needs no pilot, whose correlation curve chooses the schedule for every later patient. Extra cost over the cheapest schedule at stable shares 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7: no pilot, two follow-ups throughout, 11.1%, 9.8%, 8.1%, 5.9%, 2.9%, 0.0%, 0.0%, 3.2%; an internal pilot of 20, 10.5%, 9.8%, 9.3%, 7.9%, 5.9%, 3.7%, 4.2%, 5.2%; an internal pilot of 40, 5.9%, 5.7%, 5.6%, 4.7%, 3.4%, 2.0%, 2.4%, 2.5%; an internal pilot of 80, 3.6%, 3.4%, 3.2%, 2.8%, 1.9%, 1.0%, 1.6%, 1.6%. Worst cases: no pilot 11.1%, 20 patients 10.5%, 40 patients 5.9%, 80 patients 3.6%.

A schedule the trial chooses for itself

A separate pilot cannot pay for itself in choosing a trial's follow-up schedule, because its patients are recruited, measured and discarded. A trial can instead measure its own first patients on the schedule that needs no pilot — two follow-ups — fit the correlation curve to them, and give everyone after them the schedule that curve makes cheapest. Those first patients are counted in the analysis and cost no extra visit. Forty of them cut the trial's worst-case excess cost across every curve with adjacent correlation 0.7 from 11.1% to 5.9% in a trial of two hundred an arm, and eighty cut it to 3.6%; twenty buy almost nothing. And because the curve is read from within-arm deviations, the choice cannot see the effect: the pilot's effect estimate is the same whichever schedule it chose.

planned · Power What the sixteen-run factorial's surviving runs estimate when about 1.3 runs are lost, by why they were lost. A 2⁴ factorial fitting eleven coefficients; true A = 1, B = 0.5, AB = 0.5, everything else 0, noise 1; 20,000 datasets. Runs are lost when the response exceeds 2.5, or at random at the high corner of A and B, or at random anywhere, each losing 1.26 runs on average. Bias of the fit on the surviving runs — lost at random: A 0.001, B 0.003, AB 0.000, C -0.002; lost at the high corner: A 0.003, B 0.001, AB 0.000, C -0.000; lost when the response passes the threshold: A -0.131, B -0.120, AB -0.117, C -0.000.

A run lost for a reason

A sixteen-run factorial that loses a run at random loses precision and nothing else. One that loses a run because the process failed — the response passed a threshold — loses about the same number of runs and biases exactly the effects that pushed the response there: an interaction of 0.5 reads 0.383, a main effect of 1 reads 0.869, and the interval around each still covers about 94%. Treating the lost runs as censored above the threshold overshoots with eleven coefficients and works with six, because a run's own leverage decides how much a censored observation pulls its own estimate.

design · Factorial Seven ways to report the height at the setting a response-surface analysis recommends, at noise 2. Bias and root mean squared error of the reported height about the true height at the chosen setting, over 2,000 studies with a central composite design of thirteen runs, or the same design run twice. 13 runs, the fit's own height: bias 0.884, error 1.537. 13 runs, optimism bootstrapped away: bias 0.183, error 1.396. One confirmation run alone: bias -0.005, error 2.005. Corrected fit and one confirmation run: bias 0.056, error 1.121. 26 runs, the fit's own height: bias 0.475, error 0.998. 26 runs, optimism bootstrapped away: bias 0.050, error 0.924. 26 runs split: choose on 13, estimate on 13: bias -0.014, error 1.213.

The height at the chosen setting

The height a response-surface fit predicts at the setting it recommends reads high, because the setting was chosen where the fit was highest. The obvious repair — choose on some runs, estimate on the rest — is impossible in the usual thirteen-run design, which has nine distinct settings where two separate fits need twelve. A parametric bootstrap of the optimism removes four fifths of it at no cost in runs, 0.884 down to 0.183 at twice the noise. With the design run twice, correcting the full fit beats splitting it: an error of 0.924 against 1.213, at a better setting.

surface · Optimum

Named alongside it

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

Covariate balanceRandomisation testMonte CarloClosed formOptimal designD-optimalityLeverageInformation matrixReference distributionRerandomisationOrthogonalityRandomisation

All concepts