When using the central limit theorem to calculate for $n$ I come across having to use $2\Phi(x)-1$ to find this. However, I'm unsure why I have to use this and what it means and how it's derived. It just seems to be rather common as an qpproach to follow when trying to derive $n$, what's the explanation behind $2\Phi(x)-1$?

where symbol $\Phi(x)$ denotes the cumulative distribution function of a standard normal variable.

For example: Suppose that a random sample of size $n$ is to be taken from a distribution for which the mean is $\mu$ and the standard deviation is $3$. Use the central limit theorem to determine approximately the smallest value of $n$ for which the following relation will be satisfied: $$Pr(|\bar{X}_n-\mu|<0.3)\ge 0.95$$

By following the $Z$ distribution: $Z = \frac{\sqrt(n)(\bar{X}_n-\mu)}{\sigma}$ I can get: $Pr\left(\frac{\sqrt(n)(\bar{X}_n-\mu)}{3}<0.3\right)=Pr(|\bar{X}_n-\mu|<0.1\sqrt{n})\ge 0.95 $

For the next step I do: $Pr(|\bar{X}_n-\mu|<0.1\sqrt{n})\approx2\Phi(0.1\sqrt{n})-1\ge 0.95$

Then re-arranging to find $n$.

So what's the intuition behind using $2\Phi(x)-1$?

  • $\begingroup$ You seem to think $Pr\left(\frac{\sqrt{n}(\bar{X}_n-\mu)}{3}<0.3\right)=Pr(|\bar{X}_n-\mu|<0.1\sqrt{n}).$ Cam you explain why? $\endgroup$
    – BruceET
    Sep 3, 2021 at 15:08
  • $\begingroup$ @BruceET Perhaps it makes more sense saying $P(|Z| < 0.1\sqrt{n})$ given that we already know $Pr(|\bar{X}_n-\mu|<0.3)$ $\endgroup$
    – Stackcans
    Sep 3, 2021 at 15:46

1 Answer 1


For a standard normal random variable $Z \sim N(0,1)$ and a positive real number $\alpha$, \begin{align} P(|Z| \leq \alpha) &= P(-\alpha \leq Z \leq \alpha)\\ &= \Phi(\alpha) - \Phi(-\alpha)\tag{1}\\ &= \Phi(\alpha) - [1 - \Phi(\alpha)]\tag{2}\\ &= 2\Phi(\alpha) - 1 \tag{3} \end{align} where about the only place where any sort of intuition might have been used is in recognizing that symmetry of the standard normal pdf about $0$ allows us to recognize (or intuit) that $\Phi(-\alpha)$, the area under the pdf to the left of the point $-\alpha$, must equal $1-\Phi(\alpha)$, the area under the pdf to the right of the point $\alpha$, and so $(2)$ follows from $(1)$. In particular, note that tables of $\Phi(\cdot)$ show that $\Phi(1.96) = 0.9750$ and so $$P(|Z| \leq 1.96) = 2\times 0.9750 - 1 = 0.95.$$ Since the right side of $(3)$ is an increasing function of $\alpha$, choosing $\alpha$ to be larger that $1.96$ will certainly give us a probability larger than $0.95$.

Next, given a random sample $X = (x_1, x_2, \ldots,x_n)$ from a distribution with known mean $\mu$ and known standard deviation $\sigma$, it is known that the sample mean $\bar{X}$ is a random variable with mean $\mu$ and standard deviation $\dfrac{\sigma}{\sqrt{n}}$ and so genuflections in the general direction of the CLT allows us to claim that $$\frac{\bar{X} - \mu}{\sigma/\sqrt{n}} = \frac{\sqrt{n}(\bar{X} - \mu)} {\sigma} \sim Z \sim N(0,1)$$ or at least when $n$ is "large".


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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