16 Lognormal Distribution
The Lognormal distribution governs a world very different from the Normal.
The Normal is the distribution of additive balance.
The Lognormal is the distribution of multiplicative growth.
It appears when change happens not by repeated addition, but by repeated proportional effect.
That difference matters more than it first seems.
A quantity that grows by percentages, ratios, or compounded factors lives in a different statistical universe from one that grows by simple increments.
16.1 🏵 Definition
A variable \(X\) is Lognormal if its logarithm is Normal:
\[ \log(X) \sim \mathcal{N}(\mu, \sigma^2) \]
This implies that:
- \(X > 0\)
- the distribution is right-skewed
- many values cluster lower down
- a few values may stretch far upward
So the Lognormal is a positive, asymmetric distribution with a long right tail.
16.2 🔰 Why it appears
The Lognormal often emerges when many small multiplicative factors combine.
Examples include:
- income components
- biological growth
- reaction times
- file sizes
- prices under compounded changes
- durations shaped by proportional effects
In such systems, change is often better described by multiplication than addition.
A variable may not gain “three units” repeatedly.
It may gain “three percent,” or shrink by a factor, or scale in relation to its current size.
That is exactly the terrain of the Lognormal.
16.3 ☯ Positivity matters
One immediate clue that the Lognormal may be relevant is strict positivity.
A Lognormal variable cannot be zero or negative.
That makes it natural for quantities like:
- time
- size
- concentration
- duration
- income
- cost
- biological mass
Whenever negative values are impossible in the phenomenon itself, the Lognormal often becomes a plausible candidate.
⚜️ This is already one reason it differs deeply from the Normal.
The Normal stretches across all real numbers.
The Lognormal respects one-sided reality.
16.4 ⚙️ Shape
The Lognormal is:
- continuous
- strictly positive
- right-skewed
- heavy on moderate values
- open to large upper extremes
Its shape depends on how spread out the underlying Normal is on the log-scale.
If the variance on the log scale is small, the Lognormal is only mildly skewed.
If it is large, the right tail stretches dramatically.
This makes it a natural bridge between moderate variability and strong asymmetry.
16.5 💡 Mean, median, and asymmetry
In a skewed distribution, center becomes more complicated.
For the Lognormal:
- the median lies below the mean
- the mean is pulled upward by the right tail
This is an important lesson.
In right-skewed worlds, the average may no longer describe the “typical” case well.
A few large values can move the mean significantly.
This is one reason the Lognormal is so useful in practice: it helps explain why many positive quantities have a typical lower cluster and a smaller set of very large observations.
16.6 🪄 Log scale as clarification
One of the most beautiful aspects of the Lognormal is that it becomes simpler after transformation.
Take logs, and the asymmetry often turns into approximate Normality.
This is not just a computational trick.
It reveals structure.
The logarithm turns multiplicative complexity into additive intelligibility.
That is a profound mathematical act:
- growth by factors becomes growth by sums
- skewed shape becomes approximate symmetry
- multiplicative noise becomes additive noise
16.7 ⚠️ When not to use it
The Lognormal is not appropriate merely because data are positive.
Positivity alone is not enough.
One should also ask:
- is the process plausibly multiplicative?
- is the right-skew moderate in a way consistent with compounding?
- do logs produce a more regular structure?
- are the tails plausible under a Lognormal rather than something more extreme?
A positive variable may also be better modeled by Gamma, Weibull, Pareto, or something else.
So the Lognormal should be chosen as a structural idea, not a cosmetic curve.
16.8 🔰 The Lognormal worldview
The Lognormal reflects a world where change scales with what already exists.
The rich accumulate percentage returns.
Organisms grow proportionally.
Durations stretch under compounding influences.
Sizes multiply through repeated factors.
This is not the world of linear accumulation.
It is the world of proportional becoming.
That is why the Lognormal feels so natural in economics, biology, and many measurement contexts.
16.9 🏵 Final thought
The Lognormal distribution is the geometry of positive multiplicative life.
It reminds us that not all variability is additive, and not all asymmetry is accidental.
Some worlds grow by layers of proportion.
And when they do, the data leans right—not by error, but by the logic of the process itself.