4.5/5 - (62 votes)

In 2012, physicists at CERN erupted in celebration over the Higgs boson. Three years later, an entirely different team went just as wild over gravitational waves. Both discoveries had to clear the exact same statistical bar before anyone in science was willing to call them real.

The number that decides what counts as real

That bar is called standard deviation, usually written as the Greek letter sigma. It measures how spread out a set of values is around its average.

A low standard deviation means most values sit close to the average. A high one means they are scattered far and wide. Unlike variance, which squares the units and makes the result harder to interpret, standard deviation stays in the same unit as the original data, which is exactly why scientists reach for it constantly.

For a normal, bell-shaped distribution, about 68.3 percent of all values fall within one standard deviation of the average. Move further out, and the odds of landing there by pure chance shrink fast.

Why five sigma became physics’ gold standard

Physicists do not accept a discovery at one or two standard deviations from the expected result. They wait for five, a threshold known simply as five sigma.

At five sigma, the odds that a result is just random noise drop to roughly one in 3.5 million. That is the level of certainty CERN required before announcing the Higgs boson, and it took two fully independent experiments each reaching that threshold on their own before the finding was accepted.

Gravitational waves had to earn the same trust

The LIGO collaboration faced the identical demand a few years later. Confirming that gravitational waves, ripples in space itself first predicted by Einstein, were genuinely detected required clearing that same one-in-3.5-million bar.

See also  Why a casino can lose on one spin but never lose over a year

Two of the most significant physics discoveries of the century, from two completely unrelated fields studying completely different phenomena, were both held to the very same statistical standard before either one was allowed to change a textbook.

Two things almost everyone gets wrong about it

Standard deviation carries a couple of persistent myths. The first is that a sample perfectly estimates the standard deviation of the full population it comes from. Even with Bessel’s correction, the common fix that divides by one less than the sample size, the estimate still carries a slight bias, simply because square roots do not behave in a straight, linear way.

The second myth is that a standard deviation of zero should be impossible in real data. It is not. It just means every single value in the dataset is identical, a rare outcome, but a perfectly valid one.

A tool for more than laboratories

None of this stays locked inside particle accelerators. An investor comparing two stocks needs both the average return and the standard deviation behind it. A stock returning 10 percent with a standard deviation of 20 points can be the smarter pick over one returning 12 percent with a standard deviation of 30, depending on how much risk someone is willing to stomach.

The concept took shape in the late 19th century through the work of Francis Galton and Karl Pearson, long before anyone imagined it would one day help confirm a subatomic particle, a ripple in spacetime, and a portfolio decision, all using the exact same piece of math.