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

Every essay — page 3

Essays 49 to 72 of 436, in the same order.

Regression, and what the summary hides

A slope, a standard error and an R² can all be computed from data the model is grossly wrong about, and none of them says so. Leverage is a number known before the outcome is looked at, influence is a number, and whether a residual plot looks bad is a question with a calibrated answer. Two wrong rows placed together hide from every single-row diagnostic, and a loss that bounds a large residual does nothing about a far one.

One row at x = 9 on a wrong line, fitted three ways. Least squares gives slope −0.389, Huber 0.171 with the extra row at weight 0.115, and least trimmed squares 0.420, fitted to the 12 rows it keeps. The MM-estimator carried on from the trimmed fit gives 0.479, with the extra row at weight 0.000 and the scale fixed at 0.319. The twenty clean rows alone give 0.495. Open circles are the rows the trimmed fit leaves out.

The start an efficient robust line inherits

The MM-estimator carries a trimmed fit on through a redescending loss, and it does what it promises on one far row: slope 0.479 at every distance, the row at weight exactly zero, and 87.2% of least squares' efficiency at twenty rows. What it cannot do is choose. At eight far rows of twenty the exact trimmed fit picks the wrong half on 111 datasets; the efficient step repairs none of them, spoils none of the other 89, and ends nearer the wrong line than the start did.

6 figures · Leverage, part 4

A standard error for a model that is wrong

Every standard error a regression prints is a statement about a model that is true, and the repair for a model that is not is one of the most-quoted lines in applied work. It is priced here rather than recommended: the robust error reads a spread the model-based one does not, and its 95% interval covers 88.73% at twenty rows, is the worse of the two under mild heteroskedasticity until fifty, and is the smaller of the two whenever the error variance sits in the middle of the design rather than at its edges. The four corrections differ in one number — what each does with a point's leverage — and on a design with a single far-out point that number takes one of them to 0.3191 of the truth and another to 5.1127 times it.

One wrong model, four designs, four slopes. The slope a straight line converges to when the truth is a quadratic, under four covariate distributions, by two routes: the population projection in closed form, and the mean of 2500 fitted slopes at 200 rows apiece. The even spread over [0, 2] gives 1.6000 and the same spread moved to [1, 3] gives 2.6000, while widening it to [0, 4] gives 2.6000 — the same number as the shifted one, because a symmetric design's target is the truth's tangent slope at the design's own mean and does not read the spread at all. An exponential spread with the SAME mean as the first gives 2.6000. So two studies of one world, each fitting the same wrong model, honestly report slopes 1.0000 apart, and neither is making an error.

What a wrong model estimates

A straight line fitted to a curved truth converges on the tangent at its own design's mean. Two honest studies of one world, fitting the same wrong model, report 2.600000 and 1.600000, and neither is in error.

6 figures · Misspecification, part 1 What each error is a claim about, and what the claim comes out as. Each variance estimate's average over 20000 draws, divided by the variance the slope actually has across those same draws, at 80 rows with the error variance leaning towards the edges of the design (γ = 0.8). One is a standard error that is right. The model-based estimate reads 0.6081 of the spread, so its standard error is 77.98% of the one it should report; the four robust corrections read 0.9576, 0.9821, 0.9961, 1.0362. Two further routes agree with the count and share none of its arithmetic: n times the counted variance is 4.8905 against a population sandwich of 4.9200, and the counted ratio of the two standard errors is 1.2799 against a closed form of 1.2806.

The bread and the filling

The robust standard error is not a safety margin. At one setting of the error variance it is 1.2806 times the model-based one and at another it is 0.8246 times it, and the sign of a single dial decides which.

6 figures · Misspecification, part 2 Robust, at the sample sizes it is reached for. Counted coverage of five 95% intervals for a slope, at six sample sizes, under an error variance leaning towards the edges of the design (γ = 0.8), over 20000 draws at the small end. The model-based interval sits at about 87.06% everywhere and does not improve with the sample, because it is a claim about a variance it is not estimating. The robust ones do improve: HC0 covers 88.73% at 20 rows, 92.70% at 50 and 94.93% at 1,000. Its promise is asymptotic and its use is not, and the gap between those two facts is this picture. The leave-one-out correction read against a t on n − 2 is the only line that is near its promise at the small end: 94.55% at 20 rows.

Robust is not free

A robust standard error's promise is asymptotic and its use is not. Its 95% interval covers 88.73% at twenty rows, and under mild heteroskedasticity it is the worse of the two intervals until a hundred.

6 figures · Misspecification, part 3 What each correction is worth, exactly. Each variance estimate's expectation under a constant error variance, divided by the variance the slope actually has, at 20 rows on an even design, by two routes: the closed form E[eᵢ²] = σ²(1 − hᵢᵢ) carried through each correction's own weight, and the mean of 20000 counted estimates. The maximum leverage here is 0.1857 and the design's fourth-moment share Σu⁴/(Σu²)² is 0.0897, which is the only thing the closed form reads. HC0 comes out at 0.8603 — short by construction, since its factor is exactly 1 − 1/n − Σu⁴/(Σu²)². HC1 reaches 0.9559, HC2 is exactly 1.0000 at every design and every sample size, and HC3 overshoots to 1.1647. On an even design the four are within a fifth of each other and the choice barely matters.

Three corrections and a leverage

On an even design of twenty rows the four robust corrections read 0.8603, 0.9559, 1.0000 and 1.1647 of the truth and the choice barely matters. Add one point at x = 8 and they read 0.3191, 0.3419, 1.0000 and 5.1127.

6 figures · Misspecification, part 4 The price of insurance is noise, not coverage. What a robust standard error costs under a constant error variance — the case where the model-based one is exactly right — at five sample sizes over 20000 draws at the small end. It is not coverage: the leave-one-out interval read against a t on n − 2 covers 95.52% at 20 rows against the model-based 95.06%. It is a wider interval, by a factor of 1.0689 at 20 rows falling to 1.0049 at 250, and it is a variance estimate 2.57 times as variable at 20 rows and 1.85 times at 250 — against a denominator that is exactly V²·2/(n − 2), so only the numerator is counted. The uncorrected estimate is the cheaper of the two at small samples and the dearer at large: 1.31 against 1.76.

Right for the wrong reason

A robust standard error costs no coverage where the risk is absent — 95.52% against 95.06% at twenty rows. It costs a 6.89% wider interval and a variance estimate 2.572 times as variable, and the pre-test that would avoid paying recovers 15.9% of what the insurance is worth.

6 figures · Misspecification, part 5 The rows are held fixed; only the clusters move. Counted coverage of four 95% intervals for a slope, at five cluster counts with the row count held at 300 throughout and a within-cluster correlation of 0.1, over 6000 draws apiece, with the sizes equal. The interval that counts rows covers 53.42% at 5 clusters — a second closed form says 2Φ(z/√D) − 1 = 54.44% for a design effect of 6.900, and reads nothing about clusters at all. The cluster-robust interval read against a normal covers 74.43% there and 94.20% at 100 clusters; read against a t on G − 1 it covers 85.08% and 94.47%. The number of independent things is the cluster count, and every quantity here is blind to how many rows were typed.

The count that is not the rows

Three hundred rows in five clusters of sixty carry 6.9000 times the variance an independent-rows calculation reports, and the interval that counts rows covers 53.42%. The same five unequal sizes laid out two ways give design effects of 9.3158 and 5.4652.

6 figures · Misspecification, part 6 One estimator, three answers, and only the reference changes. Coverage of the cluster-robust 95% interval for the slope against the number of clusters, at 30 rows in each. The estimator is identical in all three curves; what differs is the number it is compared against. At 5 clusters it covers 75.05% against a normal, 85.30% against a t on 4 degrees of freedom and 87.95% against a t on 3. At 80 clusters the three agree to within a point. The correction costs nothing: the same standard error, a different table.

The reference the sandwich is read against

The cluster-robust interval covers 75.05% at five clusters and 93.58% at eighty. The same estimate read against a t on G − 2 covers 87.95% at five, and the estimator is unchanged — three hundred rows grouped into five clusters cover 74.28% where the same three hundred grouped into seventy-five cover 94.63%.

5 figures · Misspecification, part 7

A variable that moves one thing only

An instrument identifies a causal effect by assuming that a path no data can check is exactly zero, and it charges for that assumption in variance. Both prices are computed rather than described: a violation nobody can see is divided by the first stage, so the strength an instrument needs is 2.78 times the violation it is assumed not to have, and the conventional interval turns out to fail not when one instrument is weak but when many are — a failure each row's own first-stage leverage explains, and leaving that row out repairs at a price in width that depends on how strong the instruments are. What is left identified is the effect among the units the instrument actually moved, which at the standing compliance profile is 1.1000 against a population average of 0.5000.

The first stage an instrument needs is set by the violation nobody can see. The error each estimator converges on when the instrument has a direct effect of 0.05 on the outcome — a path the exclusion restriction asserts is zero and no sample can check. The instrument's error is δ/π exactly, so it is the reciprocal of the very quantity that made the method work: 1.0000 at a first stage of 0.05 and 0.0833 at 0.60. Least squares carries the confounding instead, at 0.3440 at a first stage of 0.30. The two cross at π = 0.1389, and the crossing is exactly δ times 2.7778 — the first stage an instrument needs is proportional to the violation it is assumed not to have, and below that line the method being corrected is the better estimator.

The assumption nothing tests

An instrument buys a causal effect with an assumption no sample can check, and the price is set by the same quantity that made the method work. The first stage it needs is 2.7778 times the violation it is assumed not to have, so a direct effect of 0.05 demands a first stage of 0.1389 and least squares wins below it.

5 figures · Exclusion, part 1 A weak instrument gives back the problem it was hired for. The counted mean bias of two-stage least squares at 4 instruments and 200 rows, over 2000 draws a setting, against the standard approximation and against the least-squares inconsistency the instrument was brought in to remove. At π = 0.02 the counted bias is 0.3220 ± 0.0142 where least squares is out by 0.3594 — 89.6% of the way back. At π = 0.3 it is 0.0118 against 0.2647. The approximation, the inconsistency over the population first-stage F, tracks the count at the weak end and sits above it in the middle: 0.968, 0.971, 0.918, 0.810, 0.740, 0.722, 0.846 as the ratio of counted to approximated bias.

Weak, and back where it started

A consistent instrumental estimate at two hundred rows and a concentration parameter of 0.32 is biased by 0.3220 ± 0.0142 against a least-squares inconsistency of 0.3594 — 89.6% of the way back to the problem it was hired to solve. Just identified, it has no mean at all, and that is measured as a rate rather than assumed.

5 figures · Exclusion, part 2 The interval that over-covers when the instrument fails. Counted coverage of two nominal 95.0% intervals for the same causal effect, read off the same 2000 draws of 200 rows at each first stage. The exact Anderson–Rubin set covers 95.3% at every setting — flat, because the statistic it inverts is built from y − tβ, which contains no π at all, and is therefore the same number on the same draw whatever the instrument is worth. The conventional interval covers 99.1% at π = 0.02 and 95.6% at π = 0.6: it goes wrong at the weak end by covering too MUCH, at a median width of 7.320, because its standard error is computed from residuals taken at an estimate that has itself gone wrong. A weak instrument does not make this interval lie about its coverage; it makes it useless while telling the truth.

What the first stage does not know

A single weak instrument does not make the conventional interval undercover — it makes it cover 99.1% at a width of 7.320. Where the promise actually breaks is many instruments — coverage falls from 97.2% to 51.5% while the median width falls from 1.454 to 0.583.

5 figures · Exclusion, part 3 Four numbers, and only one of them is the question. Four quantities in a population where the effect is not the same for everybody: 40.0% compliers, 25.0% always-takers and 35.0% never-takers, with the compliers carrying an effect 1.0000 larger than everybody else's. The population average effect is 0.5000. The compliers' average effect is 1.1000. A valid instrument converges on 1.1000 — the second of those, not the first, and the two differ by 0.6000. Comparing the treated with the untreated as they stand gives 1.4182, wrong by 0.9182 in the same direction, because always-takers start 1.5000 above never-takers before any treatment happens. The instrument removes the selection and changes the question at the same time, and only one of those is reported.

Whose effect it is

With a perfectly valid instrument and no violation of anything, the estimate converges on 1.1000 where the population average effect is 0.5000. The gap is exactly θ(1 − p_c), the always-takers and never-takers cancel out of both halves of the ratio, and five per cent defiers move the answer to 1.2667.

5 figures · Exclusion, part 4 The same error, caught or invisible, by how it is arranged. The overidentification test's rejection rate against the error the violation actually puts into the estimate, so the two rows are the same estimate being equally wrong. With the whole violation on one instrument the test keeps its size at 5.0% under the null and reaches 86.4% by an error of 0.800. With both instruments violating in the same ratio as their first stages the two Wald ratios are identical, the test has nothing to compare, and it rejects at 5.8% at that same error — its own size. Over 1000 draws of 300 rows at each setting, at a nominal 5.0%. The test is a comparison between instruments and it was never a check on either.

Two instruments that disagree

The overidentification test keeps its size at 5.0% and reaches 86.4% power against a violation carried by one instrument. Against the same error carried by both in proportion to their first stages it rejects on 4.6% of draws — its own size — while the estimate is wrong by 0.3000, which is 94.2% of the confounding the instruments were brought in to remove.

5 figures · Exclusion, part 5 Three intervals as one strength is spread thinner, at a concentration of 8. Coverage of three nominal 95.0% intervals on the same 1000 draws of 200 rows at each count, when a total concentration parameter of 8 is spread over 1 to 32 instruments. Two-stage least squares covers 97.2%, 96.4%, 94.0%, 86.7%, 73.2%, 51.5%. Building each row's fitted treatment from a first stage that never saw that row covers 97.1%, 97.2%, 98.3%, 97.8%, 97.9%, 98.7%. Limited-information maximum likelihood with its conventional standard error covers 97.2%, 96.7%, 95.7%, 90.9%, 85.2%, 79.0%. At one instrument the likelihood estimator is two-stage least squares exactly, which is why the first readings of those two agree to the last draw.

Leaving each row out of its own first stage

Spread a fixed first-stage strength over thirty-two instruments and two-stage least squares covers 51.5%. Build each row's fitted treatment from a first stage that never saw that row and the same draws cover 98.7% — through an interval 5.99 times as wide, around an estimate that misses by more than the whole effect on 34.7% of draws. At eight times the strength the same repair covers 95.3% and costs a width factor of 1.66.

7 figures · Exclusion, part 6 Three intervals as one strength is spread thinner, at a concentration of 8. Coverage of four nominal 95.0% intervals on the same 1000 draws of 200 rows at each count, when a total concentration parameter of 8 is spread over 1 to 32 instruments. Two-stage least squares covers 97.2%, 96.4%, 94.0%, 86.7%, 73.2%, 51.5%. Building each row's fitted treatment from a first stage that never saw that row covers 97.1%, 97.2%, 98.3%, 97.8%, 97.9%, 98.7%. Limited-information maximum likelihood with its conventional standard error covers 97.2%, 96.7%, 95.7%, 90.9%, 85.2%, 79.0%. The same estimate with Bekker's many-instrument standard error covers 97.2%, 97.2%, 97.2%, 95.0%, 94.3%, 93.8%. At one instrument the likelihood estimator is two-stage least squares exactly, which is why the first readings of those two agree to the last draw.

A standard error that knows about the instruments

Limited-information maximum likelihood came out least biased when a concentration parameter of 8 was spread over thirty-two instruments, and its conventional interval covered 79.0%. Bekker's many-instrument standard error covers 93.8% on the same draws, at 63% of the jackknife's width — and it gets there with a median standard error of 0.561 against a true spread of 0.797, because it is large on the draws that need it. At eight times the strength it covers 94.9% at 91% of the jackknife's width, and nothing measured here beats it.

6 figures · Exclusion, part 7

What conditioning on a variable does

Putting a covariate into the regression is one arithmetic operation, and it is the right thing to do in one of the three worlds it could have come from. Here the three worlds are built to share a covariance matrix entry for entry, so the estimate they disagree about by 0.348 is a number no sample of any size can settle, and the rule that controls for everything measured is measured: it leaves a larger bias than controlling for nothing on 65.5% of four thousand structures.

The same covariate, three ways round. Three worlds over a treatment, an outcome and a covariate, joined by the same three edges at the same three strengths — 0.90, 0.50 and 0.70 — differing only in which way the two edges touching the covariate point. In the first the covariate causes both and adjusting for it recovers the effect of 0.50 exactly. In the second the treatment causes the covariate, the effect is 1.13, and adjusting returns 0.50 — the direct edge alone, with the part that travels through the covariate deleted. In the third the treatment and the outcome both cause the covariate, the effect is 0.50, and adjusting returns -0.087. The regression that produces those three numbers is one formula, and nothing in the data says which panel it is being run in.

One arithmetic, three decisions

A covariate beside a treatment and an outcome can be a common cause of both, a step on the path between them, or an effect of both. The regression that includes it is the same arithmetic in all three, and it is right in one — returning 0.5000, deleting 0.6300 of the effect, and turning 0.5000 into −0.0872.

6 figures · Conditioning, part 1 One distribution, two effects. Three causal structures fitted to one covariance matrix over a treatment, a covariate and an outcome. Each reproduces it exactly — the largest entry-wise disagreement across all three is 4.4e-16 — so no sample of any size distinguishes them. The regression of the outcome on the treatment and the covariate returns 0.500 in all three, to within 4.4e-16, because that coefficient is a function of the covariance and of nothing else. The effect the three worlds hold is 0.500, 0.848 and 0.848: adjusting is exactly right in the first and off by −0.348 in the other two. The arithmetic cannot see the difference and the difference is the whole question.

The two worlds that look the same

Three causal structures were fitted to one covariance matrix and agree with it to 4.4·10⁻¹⁶. The regression returns 0.5000 under all three; the effect they hold is 0.5000, 0.8481 and 0.8481. What separates structures is a missing edge, and the signature of one is a correlation of exactly zero.

6 figures · Conditioning, part 2 Two structures in three are made worse. What adjusting for every covariate measured does to the bias in the treatment's estimated effect, against adjusting for none, over 4000 randomly drawn structures of 6 covariates each. Each covariate is independently a common cause with probability 0.25, a cause of the treatment only, a cause of the outcome only, a cause of neither, a step on the causal path, or a common effect. The rule leaves a larger bias on 65.5% of structures, a smaller one on 33.8%, and the same on 0.7%. The share is a property of that population of structures rather than of adjustment, which is why the weights are stated; what does not depend on them is that the rule has no direction — it is not a conservative default that occasionally overcorrects, it is a rule whose error is whatever the structure happens to be.

Adjusting for everything

"Control for every covariate that was measured" leaves a larger bias than controlling for nothing on 65.5% of four thousand randomly drawn structures and a smaller one on 33.8%. Its squared error is 4.110 times that of using no covariate at all, and half of it sits in its worst tenth of structures.

6 figures · Conditioning, part 3 A covariate that is prior to everything and still ruins it. A covariate measured before the treatment, caused by neither the treatment nor the outcome, and not a common cause of them. Two unmeasured variables sit behind it: one reaches the treatment, the other reaches the outcome, and both reach the covariate. Every rule of thumb for including a baseline variable is satisfied, and the regression that leaves the covariate out estimates the treatment's effect of 0.50 without bias, while the regression that includes it is off by −0.2000 — because the covariate is a common effect of the two unmeasured causes, and conditioning on a common effect makes its causes dependent. The path it opens runs from the treatment back through the first unmeasured cause, through the covariate, and out through the second to the outcome.

A collider before the treatment

A covariate measured before the treatment, on no causal path, and not a common cause of anything, still biases the estimate by exactly −0.2000 against an effect of 0.5 — while the regression that leaves it out is exact. The bias saturates at 0.3536, and the two paths that make it a collider do not appear in that bound.

6 figures · Conditioning, part 4 The line is the sample, and the sample is the finding. 900 draws of two independent standard normal causes, with the 453 of them past a threshold of 0.00 marked and the 447 that fall short left pale. In the population the two are independent by construction. Inside the selected sample the correlation is -0.4669 in closed form and -0.5050 counted on these 453 rows, and the least-squares line through them has a slope of -0.545. The mechanism is visible in the picture rather than argued: the threshold removes one corner of the cloud, and a cloud with a corner missing is a cloud whose two coordinates carry information about each other.

The sample is a condition

Two independent standard normals, selected on their sum exceeding its median, read a correlation of exactly −1/(π − 1) = −0.4669 inside the sample. Nothing is measured badly and nothing is missing — and both halves of that split read it, in the same direction, while the population containing both reads zero.

6 figures · Conditioning, part 5 The covariate the treatment caused, and what it hides. A covariate on the causal path: the treatment causes it and it causes the outcome, so the treatment's total effect of 1.130 runs partly through it. Adjusting for it returns the direct edge alone, 0.500, which is what somebody wanting the total effect should not have asked for. The dashed variable is the second problem: an unmeasured cause of both the covariate and the outcome. It does not touch the treatment, so the unadjusted regression still recovers 1.130 exactly. It does touch the covariate, so once the covariate is conditioned on the treatment and the outcome are linked through it, and the adjusted coefficient lands on 0.050 — neither the total effect nor the direct one.

The variable the treatment caused

Adjusting for a covariate the treatment caused stops estimating the total effect and starts estimating the direct one. When that covariate shares an unmeasured cause with the outcome it estimates neither: the total effect is 1.1300, the direct effect is 0.5000, and the regression returns 0.0500.

6 figures · Conditioning, part 6 A proxy removes less than its reliability, always. The share of the confounding bias removed by adjusting for a proxy, against how well the proxy measures the confounder. The diagonal is the answer a reader would guess — a covariate that is 80% signal removes 80% of the problem. The curve is what the arithmetic gives: the reliability, times one minus the squared correlation between the treatment and the confounder, divided by one minus the product of those two. That squared correlation is 0.4475. A reliability of 0.8 removes 68.85% and one of 0.6 removes 45.32%. The two agree only at the ends, and the gap is widest where most applied covariates sit.

Adjusting for a shadow

A covariate that is 80% signal removes 68.85% of the confounding, not 80% — the share is λ(1 − ρ²)/(1 − λρ²) and it is below the reliability everywhere. The residual bias is 0.1084 against an effect of 0.5, and at 25,600 rows it is 17.6 standard errors wide.

4 figures · Conditioning, part 7

Corrections, and what each controls

Bonferroni bounds the chance of any false positive; Benjamini–Hochberg bounds the share of the findings that are false. Both get called correcting for multiple comparisons and they are different promises, so every procedure here is made to report both rates and the power each one costs.

The familywise error rate with no correction, α = 0.05. Two routes: the curve is 1 − (1 − α)^m and the points are counted over 6,000 simulated families of true nulls. With twenty tests the chance of at least one false positive is 64.1%.

What the correction corrects

Twenty tests of true nulls produce at least one false positive 64% of the time, and the closed form and the count agree. Bonferroni holds it at 5% and Holm holds it at 5% while finding more. Nobody should still be using Bonferroni.

6 figures · Multiplicity, part 1 Twenty tests, 10 of them real — what each procedure holds. no correction: familywise 40.8%, false discovery 5.3%, power 85%. Bonferroni: familywise 2.8%, false discovery 0.5%, power 49%. Holm: familywise 3.6%, false discovery 0.6%, power 53%. Benjamini–Hochberg: familywise 20.0%, false discovery 2.6%, power 75%.

Two different promises

Bonferroni bounds the chance of any false positive. Benjamini–Hochberg bounds the share of the findings that are false. Both are called correcting for multiple comparisons, and one of them lets the familywise rate reach 20%.

6 figures · Multiplicity, part 2

FieldsThreadsSeriesConceptsFigure librarySearch