I understand the concept of standard deviation as the square root of the square of the mean of each sample value - the mean of the sample values.
Here is the mathematical representation (I've solved out the proof independently) :

1.) $\sigma = \sqrt{\{x^2\} - \{x\}^2}$
where $\{\,\}$ is the average and $x$ is a sample value.

2.) There is an alternate mathematical representation using summation sigma (for discrete random variable also) that more people are probably acquainted with. Or does this one have a slightly different meaning, I'm not sure?

My question is, can someone explicitly show me the derivation for the standard deviation of a binomial distribution.
Here is the information I know:

1.) Final formula: $\sigma = \sqrt{pqN}$

2.) $p =$ probability of event A occurring AKA $p = n(A)/N$

where $A$ is an event OR the first binomially distributed random variable, $n(A)$ is the amount of times event $A$ happens, and $N$ is the total number of events

3.) $q =$ probability of event $B$ occurring AKA $p = n(B)/N$ where $B$ is an event OR the second binomially distributed random variable, $n(B)$ is the amount of times event $B$ happens, and $N$ is the total number of events. Also, $q = 1-p$ because there are only two events, $A$ and $B$.


First prove that if $n=1$ then the variance is $p(1-p)$.

Then prove that if $X_1,\ldots,X_n$ are independent, then $\operatorname{var}(X_1+\cdots+X_n)$ $=\operatorname{var}(X_1)+\cdots+\operatorname{var}(X_n)$.

Then show that if each of $X_1,\ldots,X_n$ has a Bernoulli distribution with parameter $p$ and they are independent, i.e. each has a binomial distribution with parameters, $1,p$ and they are independent, then $X_1+\cdots+X_n$ has a binomial distribution with parameters $n,p$.

(But "first" and "then" need not be construed literally; you can do the steps above in any order.)

Then put it all together.

  • $\begingroup$ Hi Michael! Thanks so much for all the edits and this answer! I think your explanation may have lit a light bulb in my head. Is the binomial distribution a special case of the normal distribution (where you only have TWO events instead of MANY)? $\endgroup$ – Nova Feb 10 '15 at 2:59
  • $\begingroup$ Also what is Xn in this case? Any random event value? For example, when rolling two dice, Xn is any number between and including 2 and 12? $\endgroup$ – Nova Feb 10 '15 at 3:02
  • $\begingroup$ That $X$ has a Bernoulli distribution means $\Pr(X=1)=p$ and $\Pr(X=0)=1-p$. The case of two dice doesn't fit here. The binomial distribution is not a special case of the normal distribution; that would mean that every binomial distribution is a normal distribution. $\endgroup$ – Michael Hardy Feb 10 '15 at 3:27
  • $\begingroup$ I'm a bit confused now. If X can only have two values (1 and 0) what is the point of talking about Xn in your explanation? $\endgroup$ – Nova Feb 10 '15 at 3:30
  • $\begingroup$ You have a sequence of Bernoulli-distributed random variables: $X_1,X_2,X_3,\ldots$. Each of them can assume either of the two values. Suppose, for example, that $n=3$. Then $(X_1,X_2,X_3)$ can be any of eight triples: $(0,0,0)$, $(0,0,1)$, $(0,1,0)$, $(1,0,0)$, $(0,1,1)$, $(1,0,1)$, $(1,1,0)$, $(1,1,1)$. The probabilities assigned to these are $(1-p)^3$ for the first one, $p(1-p)^2$ for each of the next three, $p^2(1-p)$ for each of the three after that, and $p^3$ for the last one. The sum $X_1+X_2+X_3$ has a binomial distribution, so it is equal to $0$ with probability $(1-p)^3$,$\ldots$ $\endgroup$ – Michael Hardy Feb 10 '15 at 3:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.