PDA

View Full Version : not sure what i did wrong binomial probability


mtingt
Nov13-11, 04:00 PM
In a 22-item true–false examination, a student guesses on each question.

If 14 correct answers constitute a passing grade, what is the probability the student will pass?

i did c(22,14)* (1/2)^14 * (1/2)^8

mathman
Nov13-11, 04:14 PM
You have to add in the probabilities for more than 14 correct guesses. Your answer is for exactly 14 correct.

mtingt
Nov13-11, 04:21 PM
so i would have to add every single probability up to 22?
is there any other way i could do this?

SW VandeCarr
Nov13-11, 07:30 PM
so i would have to add every single probability up to 22?
is there any other way i could do this?

You can use an online calculator for p=0.5, n=22, x=14 and solve for P(X\geq x)

http://stattrek.com/tables/binomial.aspx

awkward
Nov13-11, 07:59 PM
so i would have to add every single probability up to 22?
is there any other way i could do this?
You might also use a Normal approximation to the Binomial distribution.

SW VandeCarr
Nov14-11, 03:59 PM
You might also use a Normal approximation to the Binomial distribution.

The normal approximation gives p=0.1004 whereas the presumably exact binomial gives (P(X\geq x)=0.1431 for x=14.

For the normal approximation I'm using mean 11 and SD = \sqrt {11(1-0.5)} = 2.345

awkward
Nov14-11, 04:55 PM
The normal approximation gives p=0.1004 whereas the presumably exact binomial gives (P(X\geq x)=0.1431 for x=14.

For the normal approximation I'm using mean 11 and SD = \sqrt {11(1-0.5)} = 2.345
I get P(X \geq 13.5) = 0.1432 using the Normal distribution adjusted for continuity.

SW VandeCarr
Nov14-11, 05:37 PM
I get P(X \geq 13.5) = 0.1432 using the Normal distribution adjusted for continuity.

I did too, but when Ted Williams was told his 0.3995 batting average would go into the record books as 0.400, he said that wasn't really 0.400 and played through two final season games ending up with a 0.406 batting average. Is 13.5 a passing grade or is 14 a threshold value? I understand the continuity correction and it's fine for some applications but for n=22 and a "threshold" value, why not use an exact calculation? In either case, you will likely use tables or a calculator.

Having said that, it's closer than I would have thought, but I wouldn't have been comfortable without doing the exact approach.

awkward
Nov14-11, 10:19 PM
I agree, in this case the approximation works better than we have any right to expect. Still, it's a useful tool to have around.