Suppose $X_1$ is a standard normal variable. Define, $$X_2=\begin{cases}-X_1, & \text{if } |X_1|<1,\\X_1, & \text{otherwise}\end{cases}$$

Show that $X_2$ is also a standard normal random variable.

My approach:

Let,$$F(X_2)=P[X_2\leq x_2]\\=P[X_2\leq x_2\mid|X_1|\leq 1].P[|X_1|< 1]+P[X_2\leq x_2\mid|X_1|\geq 1].P[|X_1|\geq 1]\\=P[-X_1\leq x_2].P[|X_1|<1]+P[X_1<x_2].P[|X_1|>1]$$

But I cannot compute the probabilities.May be,my approach is not right.

  • $\begingroup$ Maybe use MGFs? $\endgroup$ Commented Jun 10, 2016 at 14:49
  • 2
    $\begingroup$ If you were to draw a picture of the bivariate distribution of $(X_1, X_2)$ it would likely suggest a solution. $\endgroup$
    – whuber
    Commented Jun 10, 2016 at 16:03
  • 3
    $\begingroup$ Possible duplicate of Normal Distribution Existence Non-affine Invariant Transformation? $\endgroup$
    – Xi'an
    Commented Jun 10, 2016 at 17:40

2 Answers 2


You have the correct idea so far, but are missing a key trick. Recall that if $X_1$ is a standard normal distribution, then $X_1$ is symmetric around 0. This means that

$$P(X_1 < x_2) = P(X_1 > -x_2) = P(-X_1 < x_2) = P(-X_1 \leq x_2). $$

Using this, the result will follow directly from your last step.


I would divide it into 3 cases:

$X_2 \le 1$

$-1 \lt X_2 \le 1$

$1 < X_2$

The first is trivial, since $X_1$ is standard normal. For the second,

$ \begin{align} F(x_2) &= F(-1) + P(-1 \lt X_2 \le x_2) \\ &= F(-1) + P(-x_2 \le X_1 \lt 1) \\ &= F(-1) + P(-1 \lt X_1 \le x_2) ~by~symmetry \end{align} $

which is standard normal since $X_1$ is.

The third case is similar.


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.