sigma
The number of standard deviations a result lies from the prediction; particle physicists convert it to a confidence level, treating 3 sigma as 'evidence for' and 5 sigma as 'discovery.'
As defined in 144 | Solo: Are We Moving Beyond the Standard Model?
Where it comes up · 3
- 144 | Solo: Are We Moving Beyond the Standard Model?5 claims
- Holiday Message 2020 | The Screwy Universe4 claims
- 370 | Daniel Akerib on a Faint Hint of Dark Matter3 claims
Most important claims · 10 of 12Search all claims for “sigma” →
- A 3.5 sigma result does not mean there is only a 1-in-5000 chance the result is new physics; it means there is only a 1-in-5000 chance of seeing such a deviation as a statistical fluctuation IF the error bars on both theory and experiment are exactly correct—the bigger risk is that the theory or experiment contains a mistake.144 | Solo: Are We Moving Beyond the Standard Model?
- The particle-physics conventions that 3 sigma is 'evidence for' and 5 sigma is 'discovery' are completely made up; in reality 3 sigma results happen all the time and even 5 sigma results sometimes appear convincingly and then go away, because the theory or experiment was not understood exactly.144 | Solo: Are We Moving Beyond the Standard Model?
- For the muon g-2, the hard part is computing the strong-force (QCD) contributions, which can be done either by measuring them phenomenologically (giving the 4 sigma anomaly) or by direct lattice QCD computer simulation; a lattice calculation published the same day as Fermilab's result claims the theory fits the data with less than 2 sigma discrepancy, meaning there may be no anomaly to explain—so the result's reality is theoretically unsettled.144 | Solo: Are We Moving Beyond the Standard Model?
- The event's global statistical significance is 2.8 sigma, below the conventional thresholds for hint, evidence, or discovery, and because roughly one in 100 experiments produce a three-sigma fluctuation, particle physicists routinely see such results from chance.370 | Daniel Akerib on a Faint Hint of Dark Matter
- The Standard Model predicts the muon's g-factor as 2.002331841 while Fermilab measures 2.002331836, meaning the muon's magnetic moment is slightly smaller than predicted—a 4.2 sigma combined result (with Brookhaven data) corresponding to 99.9987% confidence, still short of the 5 sigma discovery threshold.144 | Solo: Are We Moving Beyond the Standard Model?
- The Standard Model predicts that neutral B mesons should decay into electron pairs and muon pairs at an equal rate (ratio of 1), but LHCb observes only about 85% as many muons as electrons, a 3.5 sigma deviation corresponding to 99.98% confidence—reaching the 'evidence for' threshold but not the 5 sigma 'discovery' standard.144 | Solo: Are We Moving Beyond the Standard Model?
- In practice, credence in an experimental result depends partly on whether it fits existing theory, as the joke that 'experimental findings need to be confirmed by theory' captures; the faster-than-light neutrino result at CERN had good statistical significance but was disbelieved because it didn't fit theory and was later found to be a mistake.Holiday Message 2020 | The Screwy Universe
- In particle physics, a 3-sigma deviation is required to claim evidence for something and a 5-sigma deviation to claim a discovery, where sigma is the standard deviation measuring how far a result is from the no-effect hypothesis; a 2.4-sigma result corresponds to about 99.2% confidence but is not enough because one effectively tests many hypotheses over a career.Holiday Message 2020 | The Screwy Universe
- The cosmic birefringence claim occupies an intermediate epistemic position: it is easy to explain theoretically (Carroll's own pseudo-Goldstone quintessence model predicts the effect to within an order of magnitude in both existence and magnitude) but the evidence is very tentative—only one group, 2.4 sigma—so Carroll, while rooting for it, would bet at even money that it will go away.Holiday Message 2020 | The Screwy Universe
- The community conventionally requires three, four, or five sigma levels before calling a result evidence or discovery, because with many experiments running, expected fluctuations would otherwise generate false claims.370 | Daniel Akerib on a Faint Hint of Dark Matter