Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

How do you find the range ( min and max) for a probability function such as $$\frac{P (B|A) − P (B)} {1−P (B)}\;?$$

What I tried was to use Venn diagrams, but I couldn't find a solution as the circles must overlap completely to max P (B|A). But that causes P(B) to decrease.

Any thoughts?

Thank you.

share|cite|improve this question
Presumably you have some concrete problem in mind. Without further information, $P(B|A)$ could be $0$, or it could be $1$, or anywhere in between. The ratio you wrote down could be huge negative. It is never greater than $1$. If there are additional details that you know, perhaps a more explicit answer could be supplied. – André Nicolas Oct 4 '11 at 3:03
That is all to the question. the value of P(B|A) can be anything. So the upper bound of the equation is 1. How would one solve to a possible lower bound. Thank you. – Jimmy Oct 4 '11 at 3:06
I will give an answer to the lower bound issue, because it is unpleasant to type it as a series of comments. – André Nicolas Oct 4 '11 at 3:10
When does the probability function get its minimum and maximum values. – Jimmy Oct 4 '11 at 3:17
I am concerned about your use of the term "probability function." For a probability function, you would need first of all a sample space. And as has been observed in my comments and answer, the expression you gave can in some situations be negative. Then certainly it cannot be interpreted as a probability. – André Nicolas Oct 4 '11 at 3:48

No lower bound: We give an example to show that the expression $$\frac{P(B|A)-P(B)}{1-P(B)}$$ can be arbitrarily large negative.

Imagine tossing a very unfair coin which has probability of head equal to $10^{-6}$, and therefore probability of tail equal to $1-10^{-6}$.

Let $B$ be the event "tail" and let $A$ the event "head." It is clear that $P(B|A)=0$. Thus our expression is equal to $$\frac{0 -(1-10^{-6})}{1-(1-10^{-6})}.$$ This simplifies to $-999999$.

By choosing $10^{-66}$ instead of $10^{-6}$, we can make our expression inconceivably huge negative. So there is no universal lower bound for our expression. (If suitable restrictions are put on $B$ and $A$, there may be a lower bound.)

An upper bound: Choose any $A$ such that $P(B|A)=1$, for example choose $A=B$. Then our expression is equal to $1$. It cannot ever be larger than $1$, since for given $P(B)$, the numerator is maximized if $A$ is such that $P(B|A)=1$. Thus $1$ is an upper bound for our expression. This upper bound can be attained, so there is no cheaper universal upper bound.

share|cite|improve this answer
Thank you for that great analysis. I was wondering what is A -> B? – Jimmy Oct 4 '11 at 3:28
Your meaning is not clear to me. One can make sense of $A \implies B$ in a probabilistic context, but you would have to frame a specific question. – André Nicolas Oct 4 '11 at 3:41

As another way of looking at André Nicolas's solution, note that your expression is the same as $$1-\dfrac{\Pr(\text{not }B|A)}{\Pr(\text{not }B)}$$ where the quotient can take any non-negative value and so the whole expression can take any value from 1 downwards including any negative value.

As he says in a comment, this may be a function of probabilities but it is not a probability itself.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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