enter image description here

A class of $400$.
       • $220$ women.
       • $180$ men.
Divided into study groups of 10 students each.
Q. What is the probability that a certain group will have exactly $4$ men?

$$ P(x) = \frac{n!}{x! \cdot (n-x)!} \cdot p^x \cdot (1-p)^{(n-x)}$$

$ n=4 \qquad x=4 \qquad p=0.45 $

$ P(x) = \frac{10!}{4! \cdot (10-4)!} \cdot 0.45^4 \cdot (1-0.45)^{(10-4)} $

$ \qquad \,\, = 23.84\%$

This is from the Binomial Distributions topic of a Udemy course I'm taking. That's the given answer. This seems like how to find the probability of a certain group having four men if each person was chosen from $10$ different groups where probability of choosing a man is $0.45$.

So, is that answer correct? What am I missing?

  • 2
    $\begingroup$ This is random variable $X$ follows the hypergeometric distribution (without replacement) rather than the binomial distribution: $$P(X=4)=\frac{\binom{180}{4}\cdot \binom{220}{10-4}}{\binom{400}{10}}=24.1\%$$ $\endgroup$ – callculus Feb 17 '18 at 16:12
  • $\begingroup$ I appreciate that you recognized this flaw in the video. Respect. I would you give 10 upvotes more if I could. $\endgroup$ – callculus Feb 17 '18 at 19:19

Lets make a small example: You have a group of 2 men and 3 woman. Now you want divide into a study group of 3 students with 2 women ($Y$). The ways to do that is

$\texttt{wwm, wmw, mww}$

Each group has the same probability. Let us focus on the first group. The probability to choose a woman first is $\frac{3}{2+3}=\frac{3}{5}$

The probability to choose woman again is $\frac{3-1}{2+3-1}=\frac{2}{4}=\frac{1}{2}$

Here you can see that you have to use a distribution which regard, that the woman are not replaced.

The probability to choose a man is $\frac{2}{2+3-2}=\frac{2}{3}$

Thus the probability to choose $\texttt{wwm}$ is $\frac{3}{5}\cdot \frac{1}{2}\cdot \frac{2}{3}=\frac1{5}=20\%$.

Since we have three ways to choose $2$ woman, the probability to choose $2$ woman is $P(Y=2)=60\%$. This is equal to $$P(Y=2)=\frac{\binom{3}{2}\cdot \binom{2}{1}}{\binom{5}{3}}$$


It seems that the given example is supposed to show that the binomial distribution can be used as an approximation for the hyper-geometric one if

  • the number of selected items is small in comparision with the size of the population one selects from
  • the probability $p$ of a "success" (here man in group) is relatively stable if only a few items (men) are selected - this is so if $p$ is not too far from $0.5$

As callculus indicated above, the correct solution using the hyper-geometric distribution is $$\frac{\binom{180}{4}\cdot \binom{220}{10-4}}{\binom{400}{10}} \approx 0.241$$

The approximation using the binomial distribution with $p = \frac{180}{400} = 0.45$ is $$\binom{10}{4} \cdot 0.45^4 \cdot 0.55^6 \approx 0.238$$

  • $\begingroup$ But in this part of the video the approximation is not mentioned. So this is a clear mistake. $\endgroup$ – callculus Feb 17 '18 at 18:50
  • 1
    $\begingroup$ Sad to hear. Another case where "educational math" produces erroneous pseudo-math and confuses those learners who really want to understand. $\endgroup$ – trancelocation Feb 17 '18 at 19:03
  • 1
    $\begingroup$ That´s really true. And also: Academic titles sometimes means nothing: He is an Associate Dean of Executive MBA Programs at UCLA Anderson School of Management. Here is the link to the video. $\endgroup$ – callculus Feb 17 '18 at 19:15
  • $\begingroup$ The video is not listed any more on his channel. I good decision from George. But it can be still found at the link $\endgroup$ – callculus Jul 15 at 19:11

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.