5. The Normal Distributions
5.2 **The Normal Distribution as a Limit of Binomial Distributions
The results of the derivation given here may be used to understand the origin of the Normal Distribution as a limit of Binomial Distributions[1]. A mathematical “trick” using logarithmic differentiation will be used.
First, recall the definition of the Binomial Distribution[2] as
(5.2)
where is the probability of success, is probability of failure and
(5.3)
is the binomial coefficient that counts the number of ways to select items from items without caring about the order of selection. Here is a discrete variable, , with .
The trick is to find a way to deal with the fact that ( is a discrete variable) for the Binomial Distribution and ( is a continuous variable) for the Normal Distribution[3] In other words as we let we need to come up with a way to let shrink[4] so that a probability density limit (the Normal Distribution) is reached from a sequence of probability distributions (modified Binomial Distributions). So let represent the Normal Distribution with mean and variance . We will show how where each Binomial Distribution also has mean and variance .
The heart of the trick is to notice[5] that
(5.4)
This is perfectly true for the density . The trick is to substitute the distribution for the density in the RHS of Equation (5.4) to get :
(5.5)
because . The trick is to now pretend that is a continuous function defined at all ; we just don’t know what its values should be for non-integer . With such a “continuation” of we can write[6]
(..)
Equation (5.8) has no limit; it blows up as . We need to transform in such a way to gain control on (getting it to shrink as ) and to get something that converges. To do that we introduce and a new variable . With this transformation of variables, the chain rule gives
(5.9)
and the RHS of Equation (5.8) becomes, using
(..)
Using Equation (5.9), for the LHS, and Equation (5.14), for the RHS, Equation (5.8) becomes
(..)
where means terms that will go to zero as , and we have used the relation to get Equation (5.16}) and to go from Equation (5.17) to Equation (5.18). Dividing both sides of Equation (5.19) by leaves
(5.20)
Our transformation, with its , has given us the exact control we need to keep the limit from disappearing or blowing up. Integrating Equation (5.20) gives
(5.21)
where is the a constant of integration. Switching back to the variable
(..)
To evaluate the constant of integration, , we impose because we want to be a probability distribution. So
(5.25)
so
(5.26)
and
(5.27)
which is the Normal Distribution that approximates Binomial Distributions with the same mean and variance as gets large.
You may be wondering why that transformation worked because it seems to have been pulled from the air. According to Lindsay & Margenau, it was Laplace who first used this transformation and derivation in 1812. What this transformation does is pull the Binomial Distribution back to have a mean of zero (by subtracting ) which keeps from running off to infinity and, more importantly, allows us to define a function with that has a constant variance of that we can match to when we transform back to at each , see Figure 5.1. Looking at it the other way around, the Normal Distribution[7] with is an approximation for Binomial Distribution that “asymptotically” approaches as .
This is not the only way to form a probability density limit from a sequence of Binomial distributions. It is one that gives a good approximation of the Binomial Distribution when is fairly small if the term in Equation (5.18) becomes small quickly. If is very small, this does not happen and another limit of Binomial Distributions that leads to the Poisson Distribution is more appropriate. When and are close to 0.5 or more generally when and then the Normal approximation is a good one. Either way, the density limit is a mathematical idealization, a convenience really, that is based on a discrete probability distribution that just summarizes the result of counting outcomes. Counting gives the foundation for probability theory.
- The formula for the Binomial Distribution was apparently derived by Newton according to: Lindsay RB, Margenau. Foundations of Physics. Dover, New York, 1957 (originally published 1936). For that claim, Lindsay & Margenau quote: von Mises R. Probability, Statistics, and Truth. Macmillan, New York, 1939 (originally published 1928). The derivation of the Normal Distribution presented here largely follows that given in Lindsay & Margenau's book. ↵
- In class we denoted the Binomial distribution as . Here we use to avoid using too many P's and p's. ↵
- Remember that the Normal Distribution is technically a probability density but we slur the use of the word distribution between probability distribution (discrete ) and probability density (continuous ) like everyone else. ↵
- for the Binomial Distribution. ↵
- Remember that and use the chain rule to notice this. ↵
- You can probably imagine many ways to continue the Binomial Distribution from to . It doesn't matter which one you pick as long as the behaviour of your new function is not too crazy between the integers; that is, should exist at all . ↵
- Our symbols here are not mathematically clean; we should write something like instead of or composed with at , , instead of . But to emphasize the intuition we use . In clean symbols, the function asymptotically approaches where . ↵