We know if a collection of events are independent, then they are pairwise independent. In general, the converse is not true. However, I'm wondering if there's a condition under which the converse holds. I haven't been able to find anything on this. Any help is appreciated.

  • 3
    $\begingroup$ When there are only two events in the collection. $\endgroup$ – bof Oct 11 '14 at 2:14
  • $\begingroup$ it is not true in the other direction either. $\endgroup$ – Calculon Oct 11 '14 at 2:18
  • 2
    $\begingroup$ @bof . . . and if you want to fit a straight line, then obtain only two data points. $\endgroup$ – Michael Hardy Oct 11 '14 at 2:30
  • 1
    $\begingroup$ @Calculon why isn't it true in the other direction? Say we have three events $A,B,C$ if they are independent then we have $P(A \cap B) = P(A)P(B)$, $P(A \cap C) = P(A)P(C),$ $P(B\cap C) = P(B)P(C)$ and $P(A\cap B \cap C) = P(A)P(B)P(C)$. Surely, they are pairwise independent, no? This generalizes to a collection of $k$ events. $\endgroup$ – epsilon-delta Oct 11 '14 at 2:38

One situation in which this is true is when the random variables involved are jointly normally distributed. If $X_1,\ldots,X_n$ are jointly distributed in such a way that for every sequence $a_1,\ldots,a_n$ of constants (i.e. non-random), the random variable $a_1 X_1+\cdots+a_n X_n$ has a $1$-dimensional normal distribution, then that is joint normality of the distribution of $X_1,\ldots,X_n$. If these are pairwise independent, then they are independent.

At the opposite extreme, you have the case of $Y_1,Y_2, Y_3$, where $Y_1, Y_2$ are independent and identically distributed and $\Pr(Y_1=1)=p\in(0,1)$ and $\Pr(Y_1=0)=1-p$, and $Y_3$ is the mod-$2$ sum of $Y_1$ and $Y_2$. These three are pairwise independent, but the values of any two of them determine the value of the third.


I was also wondering the same, and I think it's instructive to just restate mutual independence in an equivalent way. I will do it below for 3 real random variables but the statement is generalizable.

Let $X, Y, Z$ be pairwise independent random variables and also assume that $(X,Y)$ is independent of $Z$. Then $X,Y,Z$ are mutually independent.

To prove this, just pick Borel sets $A, B, C$, then it holds that:

$$ \begin{aligned} \Pr[ X \in A, Y \in B, Z \in C] &= \Pr[ (X,Y) \in A\times B, Z \in C] \\ &= \Pr[(X,Y) \in A\times B] \Pr[Z \in C] \\ &= \Pr[X \in A, Y \in B] \Pr[Z \in C] \\ &= \Pr[X \in A] \Pr[X \in B] \Pr[Z \in C] \end{aligned} $$


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.