Every essay — page 18
A guarantee that needed a symmetry
What a balancing rule can remove of an interaction is decided by parity, and parity is a statement about a symmetry of the law rather than about the covariate. Keep the dependence and change the marginals — a Gaussian copula under a monotone transformation — and the two things every trial balances come apart. A median split is a function of the sign of the latent normal whatever the marginal is, so its exact zero survives every transformation to the last digit. A mean is odd only when the marginal is symmetric, and its zero is gone at a skewness of one. The separating case is a marginal that is heavy-tailed and symmetric, where every zero holds exactly: it is not normality the guarantees needed. And a threshold at a value on the covariate's own scale never had one at all.
The cut that is not a quantile
A protocol that says split the covariate at a threshold and one that says split it at the median read the same and are different rules. One has an exact guarantee under every marginal and the other has none under any.
Balancing a skewed covariate
The worst case of the rule every trial runs goes from exactly zero to somewhere between a quarter of a per cent and two and a half. Which is small, and is a number that cannot be stated without the covariate's distribution in it.
A share at a hundred units
On a skewed covariate the closed form credits a rule that balances the means with removing 26% of the covariates' product and 52% of a square, shares that rest on symmetry having failed. A trial does not collect them. At a hundred units the rule removes 12% of the product and 37% of the square on the clearly skewed covariate, and at four hundred units barely more, because the shares live in the long tail and the tail is a handful of units. On the most skewed covariate the rule removes a fifth of the median split at a hundred units and nothing of it at four hundred: balancing the mean of such a covariate is balancing its largest few values.
The covariate on its ranks
A skewed covariate spoils every parity argument a balancing rule relies on, and a rule that balances ranks instead of values looks like the way out, since a rank has the same uniform law on any marginal. The measurement says the skew really does belong to the scale for one kind of shape: a rank holds a median split at exactly a quarter of a coin's imbalance whatever the covariate. But it pays with a cap on any shape linear in the value, at nearly half a coin's imbalance on the strongest skew. And a rule holding values and median splits together holds both, beating the rank on the shape a rank gives up and matching it on the one it gains.
Where a taper's case begins
A tapered block was measured against a plain one on the exact bias each implies, and the answer was that the taper's advantage has not arrived at any block length a hundred and twenty rows can afford — with the crossing at ℓ = 20 and a difference there of three tenths of a point, too small for the critical-value table to resolve. Two of those three statements are about the wrong quantity. What a sample reports at ℓ = 20 is four points rather than three tenths, because the autocovariances the window is applied to are themselves attenuated and the window that discards the long lags loses less of them; and the comparison is made at a shared block length where each window has its own best one. Read at each window's own setting, on the error rather than on the bias, the ordering reverses at a hundred and twenty rows.
The gap a sample shows
The exact difference between two block windows at a block length of twenty is three tenths of a point. What a hundred and twenty rows report is four and a third, because the autocovariances the window is applied to are attenuated too.
Bias is not the whole of it
A window that reaches zero at its ends attenuates less and uses less of each block. The block length that minimises its bias is not the one that minimises its error, and comparing two windows at one length compares one of them mis-tuned.
Measuring a variance rather than a quantile
A resample's implied long-run variance can be computed from the sample with no resampling in it at all. A critical value cannot, and the difference is a factor of three in the draws before any of the resampling is counted.
The error no window repairs
Every block window's best estimate of a long-run variance is wrong by about forty per cent at a hundred and twenty rows, and the largest part of that is not a bias at all. Choosing the window moves a twentieth of it.
A model against a window
A block resample's long-run variance is attenuated by its window, and the construction that escapes the window is a model: fit an autoregression and sum its dependence to every lag in closed form. Measured on both halves of the answer, the escape is narrower than it looks. Correctly specified, on 120 rows of a first-order autoregression, the model's error is 45.5% of the truth against the trapezoid block's 42.4% — it removes the bias and buys a spread larger than the bias was — and it pulls ahead only from about 480 rows. Wrongly specified, on a five-period moving average, it is off by 84% to 105% at every length, and even an order chosen by the criterion trails the window throughout.
A covariance with no parameter
A regression's coefficients and its errors' dependence can be fitted together when the dependence is one number. When it is an estimated covariance there is nothing for “jointly” to mean — until a family is named, and then the family's own width is the parameter. Three things the measurement says, two of them the opposite of the guess: the plug-in's shortfall is the taper's rather than the data's and is nothing under two of four laws; nothing in the likelihood chooses the width, because nested families buy about a unit of it a lag and that is what a criterion charges; and a fit-only objective does not run away, because a unit diagonal fixes the trace.
A family before a fit
A regression's coefficients and one correlation can be maximised together. Replace the correlation with an estimated covariance and there is nothing left for "jointly" to mean — until a set of covariances is named, and the set turns out not to contain the truth.
The plug-in and the maximum
A tapered covariance estimate sits five and a half log-likelihood units below the maximum of the likelihood it is substituted into. Four fifths of that is what the optimiser would have found if nothing were missing.
Nothing in the fit picks the width
A wider band is always a better fit, and it is better by about one unit of log-likelihood a lag — which is the order of what a criterion charges for a parameter. Three defensible rules choose widths a factor of three apart.
What fitting them together buys
Maximising over the coefficients and the covariance together beats the two-step under one of four dependences and ties under the other three. It is the one the band family contains, and the likelihood said so before any coefficient was compared.
An interval that knows the band was maximised
A regression whose coefficients and eight-lag error band are maximised together prints an interval that covers 72% to 83% where 95% was promised, because it reports nearly the precision of a known covariance and delivers less. Two repairs were named. A sandwich standard error read off the whitened residuals repairs nothing — 71% to 83% — because the maximisation has already made those residuals look independent. A parametric bootstrap that redraws from the fitted band and maximises the band again on every redraw repairs about half the gap, to 85.5% to 89%, by pricing the maximisation through repeating it. What it cannot price is that the band it redraws from is the maximised one.
How long the list is
A tuning parameter chosen per candidate rather than once for the table costs something, and the window's figure was measured while the order's was not — because their lists are different lengths and matching them changes what each rule is. Matched at every length, the two cost the same. What separated them was not the list at all: a per-candidate whitening is a different error model for every candidate, the Gaussian likelihood has a term that says so, and the sieve's rule in this collection never carried it.
The comparison that was not made
Choosing a whitening's window separately for every candidate costs 0.00401 of regret. The same question about an order was named and left, because the two lists are different lengths. The order's answer is 0.00360, and matching the lists changes almost nothing.
The volume a whitening moves
A sieve's whitening has a determinant and this collection's criterion for it never carried one. Shared across a table the term cancels exactly, which is why nothing ever noticed; used per candidate it is worth more than a parameter and the whole comparison turns on it.
A list is not a rule
How often fifteen candidates disagree about a tuning parameter runs from nothing at two values on the list to two draws in five at thirteen. What the disagreement costs does not move at all.
Two values the criterion cannot tell apart
Fifteen candidates disagree about a whitening's order on two draws in five when the list holds all thirteen orders — and on more than one in five when it holds only one and two. Every adjacent pair of positive orders disagrees about as often as that pair does, and every pair holding an order the criterion never picks disagrees on none. A list's disagreement is set by the pairs of values on it that the criterion cannot tell apart, not by how many values it has, and one candidate's own doubt about its pick predicts all fifteen candidates' disagreement with a rank correlation of 0.96.
A quarrel's cost is its width
When fifteen candidate regressions disagree about a whitening's autoregressive order, the disagreement costs in proportion to how far apart the two orders are — 0.0027 of regret for every order of distance, a line that explains 96% of the cost across sixty-six pairs — and not at all by which orders they are: a quarrel between eleven and twelve costs what one between one and two costs, though one of those holds the truth. Fitted on pairs, the line predicts every longer list's cost. The window's quarrels follow no such line.
Two searches over one sample
A break point that has been looked for costs more than a count of parameters says. So does a window chosen from a list, and nothing here had charged for it. A rule that does both is running two searches over one sample, and the charges do not add: the pair manufactures a fifth less than the two apart. What follows is sharper than a correction to a sum — most of what a break search finds under correlated errors is the correlation, so once a whitening has been chosen from the same sample the break's own charge more than halves, and carrying the published one across switches the test off entirely.
Two searches, one sample
A searched break in a regression manufactures 34.7 of likelihood ratio where a count of coefficients says 11.1. A searched window manufactures 84.0. The two together manufacture 99.4, not 118.7.
The charge that is not a sum
Charging two searches what each costs on its own is conservative, and conservative here means the test never fires. At the largest break measured it declares nothing, on every draw, while a calibrated threshold reaches 29%.
A charge that depends on the rule
The break search's charge is 34.7 on its own and 15.4 once a window has been chosen from the same sample. Most of what a break search finds under correlated errors is the correlation, and a whitening has taken it already.
A break that is there
Put a real break in the regression at row 60 and the profile's two-unit set still misses it on most draws — it contains the true row on 12% of draws at a small break and 69% at a large one, without whitening; with the window searched as well, 23% to 75%. A set that would contain it on 95% of draws needs a threshold of about 19 units without whitening and about 8.5 with it, and either honest set holds most of the sample: fifty-seven of the seventy-three candidate rows at a break of one standard deviation. Searching the window does not move where the break is found.
A set calibrated at the estimate
A confidence set for where a regression breaks needs a threshold on the profile likelihood, and no fixed one works: the honest threshold falls as the break grows. Simulating the profile's deficit at the true break, shift and persistence gives a set that covers 93.7% to 95.0% at every break size and narrows from 67 rows to 23. Simulating it at what one sample can estimate gives a set that covers 64% to 76%, and most of the shortfall is one number: a split regression's residuals report a persistence of 0.62 where the errors have 0.8.