c4countdown

Some proper statsy business here, so I guess I'm mainly looking for Raccoon Boy, but if you know then dive in.

Say you have a coin. You do 228 tosses, and 60% come up heads.

What does this tell us about the coin?

e.g., how confident can we be the coin is actually 60/40 biased in favour of heads. What are the chances of it being 50/50? What about 55/45? etc. What would you say about the sample size?

Cheers.

Shall I help?

Matt Morrison wrote:Shall I help?

If you can, sure.

Jon Corby wrote:

Matt Morrison wrote:Shall I help?

If you can, sure.

Raccoon might be able to add to that.

In a word,

Jon O'Neill wrote:In a word,

MWM's made more sense to me.

Jon Corby wrote:Some proper statsy business here, so I guess I'm mainly looking for Raccoon Boy, but if you know then dive in.

Say you have a coin. You do 228 tosses, and 60% come up heads.

What does this tell us about the coin?

e.g., how confident can we be the coin is actually 60/40 biased in favour of heads. What are the chances of it being 50/50? What about 55/45? etc. What would you say about the sample size?

Cheers.

You can't say. There's no answer. You need to have a prior probability. Stattos talk about confidence intervals but that's not the same as saying that there is a certain chance.

You'd just do a two-tailed test and see what the statistical significance is.

We'll call it 137 out of 228. I make the significance 0.0028, or however you phrase it. Basically if the coin is fair, such a result or more freaky would happen about 0.28% of the time. About 1 in 360. A degree.

Gavin Chipper wrote:We'll call it 137 out of 228. I make the significance 0.0028, or however you phrase it. Basically if the coin is fair, such a result or more freaky would happen about 0.28% of the time. About 1 in 360. A degree.

Yeah, so if the coin was 55/45, how likely are we to see that result (of 137/228)?

Jon Corby wrote:

Gavin Chipper wrote:We'll call it 137 out of 228. I make the significance 0.0028, or however you phrase it. Basically if the coin is fair, such a result or more freaky would happen about 0.28% of the time. About 1 in 360. A degree.

Yeah, so if the coin was 55/45, how likely are we to see that result (of 137/228)?

Slim. But the chance of any single outcome with that many repetitions is very slim. If you did a million trials with a perfectly fair coin, you'd still be extremely surprised to see it come out 500k/500k.

But what Gevin said originally is correct, there's no a priori way to answer your question. All you can do is propose a hypothesis and see how consistent it is with the data.

The coin may be heavier on the heads side.

Rhys Benjamin wrote:The coin may be heavier on the heads side.

IQ of 136, don't forget.

Charlie Reams wrote:

Rhys Benjamin wrote:The coin may be heavier on the heads side.

IQ of 136, don't forget.

162 which is higher than Carol V. So there.

Joseph Krol wrote:

Charlie Reams wrote:

Rhys Benjamin wrote:The coin may be heavier on the heads side.

IQ of 136, don't forget.

162 which is higher than Carol V. So there.

Yeah but she's a woman.

Ryan Taylor wrote:

Joseph Krol wrote:162 which is higher than Carol V. So there.

Yeah but she's a woman.

And therefore uses it

Lesley Hines wrote:

Ryan Taylor wrote:

Joseph Krol wrote:162 which is higher than Carol V. So there.

Yeah but she's a woman.

And therefore uses it

What? The coin? In her slot? Filth.

Jon Corby wrote:Some proper statsy business here, so I guess I'm mainly looking for Raccoon Boy, but if you know then dive in.

Say you have a coin. You do 228 tosses, and 60% come up heads.

What does this tell us about the coin?

e.g., how confident can we be the coin is actually 60/40 biased in favour of heads. What are the chances of it being 50/50? What about 55/45? etc. What would you say about the sample size?

Cheers.

No idea what you mean! I'm just a kid!!! I am in Set 1 Maths, but still!

Ryan Taylor wrote: Yeah but she's a woman.

Good job you don't work for Sky.

Kirk Bevins wrote:

Ryan Taylor wrote: Yeah but she's a woman.

Good job you don't work for Sky.

That was the joke?

Kai Laddiman wrote: That was the joke?

It can't have been as Ryan's not funny.

Soph K wrote:

Jon Corby wrote:Some proper statsy business here, so I guess I'm mainly looking for Raccoon Boy, but if you know then dive in.

Say you have a coin. You do 228 tosses, and 60% come up heads.

What does this tell us about the coin?

e.g., how confident can we be the coin is actually 60/40 biased in favour of heads. What are the chances of it being 50/50? What about 55/45? etc. What would you say about the sample size?

Cheers.

No idea what you mean! I'm just a kid!!! I am in Set 1 Maths, but still!

It turned out that he was talking rubbish anyway.

Kirk Bevins wrote:

Kai Laddiman wrote: That was the joke?

It can't have been as Ryan's not funny.

It's OK Coke, I'll pretend that it wasn't the joke to make you a little happier after being displaced as top dog of anagrams.

I see the Royal Mint are producing a £5 coin this year to commemorate the Duke of Edinburgh's birthday with his head on one side and the Queen's on the other.