01 THINKHow does he reason?

Formulas are hard to read because you cannot say the symbols out loud

I learned statistics symbols together with how to pronounce them. Once I could say them, the shape of the formula appeared first.

This is the entry retold in plain words. The sentences the author wrote are on the other side of the switch. Nobody wrote the sentences on this page by hand.

A formula is usually hard to read not because you do not know the subject. It is because you cannot say the symbols out loud.

Skim a row of unfamiliar symbols with your eyes and it just goes past you. But once you can say it — "add everything from zero to infinity" — you can see how the formula is built.

So I learned how to say them before what they mean.

The fifteen I learned first

I did not try to learn fifty at once. These fifteen keep showing up while reading the rest.

SymbolSay it asWhat it means
αalphaA value used as a cut-off
βbetaA value, or the chance of missing something
λlambdaA value for how often something happens
μmuThe average of everything
σsigmaHow spread out everything is from the average
σ²sigma squaredThat spread, squared
θthetaA value we do not know yet
Γcapital gammaThe gamma function
f(x)f of xThe shape of how likely each value is
F(x)capital F of xThe chance of getting that value or less
E(X)the expectation of XThe average
Var(X)the variance of XHow spread out it is
ΣsigmaAdd them all up
integralAdd them all up, for values with no gaps
infinityA value that grows without end

Three that kept confusing me

σ and Σ — the same name, completely different meanings.

  • small σ is how spread out things are
  • big Σ means add them all up

You cannot tell them apart by sound, so you tell them apart by size.

Γ and γ — the big one is the gamma function; the small one is a value used in the gamma distribution.

X and x — this one took me longest.

  • big X is something whose value is not decided yet
  • small x is the value that X actually turned out to be

The difference between capital and small carries the difference between "possibility" and "result."

The squiggle means "is distributed"

X ~ N(μ, σ²) is read like this — "X is distributed normal, with average mu and spread sigma squared."

A formula I actually read out

M_X(t) = ∫₀^∞ e^(tx) · λe^(−λx) dx

Said out loud — "M sub X of t is the integral from zero to infinity of e to the t x times lambda e to the minus lambda x, d x."

Simplified, it becomes this.

M_X(t) = λ / (λ − t),    for t < λ

"M sub X of t is lambda over lambda minus t, for t less than lambda."

Once I could read it, the reason for the condition t < λ came first. Adding things up must not run off to infinity — and that follows the moment you say the symbols.