I understand that the R-function
pbinom(0, 100, 0.5, lower.tail=FALSE) returns the probability of getting 0 or more heads in 100 trials. R gives the correct outcome, 1 for this problem. But if I use very small values for the probability in a single trial, as in
pbinom(0, 100, 0.0000000001, lower.tail=FALSE) I get an answer close to 0, actually: 1E-08, while I expected a result close to 1. Is this a bug? A similar thing occurs in Excel. But the propbability of ZERO or more successes in a binomial experiment should always be equal to 1. For the example above it is not important, because I know the answer already, but I am using the function pbinom() in a program.
Best Answer
Binomial distribution is
$$ Pr(X = k) = {nchoose k}p^k(1-p)^{n-k} $$
so with $k = 0$ this becomes
$$ {nchoose k}p^k(1-p)^{n-k} = {nchoose 0}p^0(1-p)^{n-0} = 1 times 1 times (1-p)^n = (1-p)^n$$
with lower.tail=FALSE you ask about $Pr(X>k)$, so basically for $Pr(X = 1)$, that is
$$ 1- (1-p)^n = 1 – (1 – 0.0000000001)^{100}$$
what gives you a correct answer. Check it if you want:
> 1-(1 - 0.0000000001)^100 [1] 1e-08 Similar Posts:
- Solved – Exact binomial test using binom.test and qbinom, pbinom
- Solved – An electronics company produces devices that work properly 95% of the time
- Solved – Calculating chi-square and p-value
- Solved – Is an implementation of a density function for a logit-normal distribution available in R
- Solved – why the lower.tail=F is used when mannualy calculating the p value from t score