Statistics · Simulation · 6 min

Central limit theorem

Take samples from a skewed population and watch their averages form a different distribution.

08/ 08
Make it your own

Random samples vary between runs.

What happensSimulated results

The idea to keep

It is the distribution of the averages that becomes approximately normal, not the original population.

01 / The explanation

What is going on?

This experiment draws independent observations from an exponential population with mean 1 and standard deviation 1. Individual values are right-skewed. Each sample is averaged, and the histogram records those sample means. Increasing the observations per sample usually produces a narrower, more symmetric shape.

Mean of X̄ = μ · Standard deviation of X̄ = σ / √n
02 / Step by step

Work through it

  1. Choose how many observations go into each average.
  2. Repeat that sampling many times to form a histogram of averages.
  3. Increase the sample size and compare the spread with 1 / √n.
03 / A worked example

Put numbers to the idea

For this population, samples of size 1 have standard deviation 1. Averages of 25 observations have standard deviation 1 / √25 = 0.2. Averages of 100 observations have standard deviation 0.1. Their center remains 1.

04 / Common questions

A little more clarity

Does the original population become normal?

No. The original distribution stays exponential. The distribution of sample averages becomes more nearly normal as the sample size grows.

Is a sample size of 30 always enough?

No. Required size depends on the population’s shape and the accuracy needed. Very skewed or heavy-tailed populations may need much larger samples.

Why does the histogram change between runs?

Each histogram uses a finite random collection of samples. A new run changes the observed bars even though the underlying population and theoretical spread stay the same.