1
$\begingroup$

We have two coins. The first one is a fair coin and has $50\%$ chance of landing on head and $50\%$ of landing on tails. for the other one it is $60\%$ for head and $40\%$ for tails. We choose one of them randomly (with equal chance) and toss it three times.The result is Tails-Head-Head. If we toss it again for the 4th time, what is the probability of the coin landing on tails.

At first, it seems like a bayes theorem question, but after some thought I think because the result of every toss is independent of others, the answer is just:

$$\frac{1}{2} . \frac{1}{2} + \frac{1}{2} . \frac{4}{10} = \frac{9}{20} $$

The first $\frac{1}{2}$ is because we choose the coins randomly.

I want to know whether my argument is correct for this probability question or not.

$\endgroup$
  • 1
    $\begingroup$ I think your answer is correct. $\endgroup$ – Kavi Rama Murthy Oct 3 '19 at 23:36
  • $\begingroup$ Intuitively, even though initially the probability is 50-50 for each coin type, by observing THH the odds increase that the coin comes from the biased coin with higher probability of heads. As a limit case, if one of the coins has heads on both sides, by observing tail you would be sure you have chosen the other coin. $\endgroup$ – Momo Oct 3 '19 at 23:42
  • $\begingroup$ No, your thought that every toss is independent leads you to the wrong result. The point is that the a posteriori chance that you have chosen the unfair coin, given you THH experiment thus far, is now a tad more than $\frac12$. So the probability of the next toss being headss is a bit more than $\frac9{20}$. This is indeed a Bayesian question. $\endgroup$ – Mark Fischler Oct 3 '19 at 23:44
3
$\begingroup$

The answer, giving two terms corresponding to the actual coin being fair, then the actual coin being biased, will be $$P(T) = \frac{\left(\frac12\right)^3}{\left(\frac12\right)^3+\frac25\left(\frac35\right)^2} \left(\frac12\right) + \frac{\frac25\left(\frac35\right)^2}{\left(\frac12\right)^3+\frac25\left(\frac35\right)^2}\left(\frac25\right)\\ $$

Doing the arithmetic, $$ P(T) = \frac{1201}{2690} \approx 44.65\% $$ The probability that your coin is biased is about 53.5%, which is why the tails on the next throw is less likely than you might think.

| cite | improve this answer | |
$\endgroup$
1
$\begingroup$

Call the fair coin $A$ and the biased coin $B$.

$$P(THT|A) = (.5)^3$$

$$P(THT|B) = (.4)(.6)(.4)$$

$$P(A| THT) = \frac{P(THT|A) P(A)}{P(THT|A) + P(THT|B)}$$

$$P(B| THT) = \frac{P(THT|B) P(B)}{P(THT|A) + P(THT|B)}$$

(Note that $P(A|THT) + P(B|THT) = 1$, as it must)

Once you have the probability it is $A$, you can easily calculate the probability the 4th coin is $H$ ($0.5$). And likewise if the coin is $B$ ($0.6$).

You can plug in the numbers.

| cite | improve this answer | |
$\endgroup$

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.