https://en.wikipedia.org/wiki/Checking_whether_a_coin_is_fair
Math
Sort:
Dec 30, 2017
0
#2
This is cool and I actually googled this before posting, but the answer I want would be expressed as 52% +- (I don't know how to make that symbol on the computer) 0.5%. More ideally, it would be divorced from any outside numbers such as desired 'confidence level'. I only want the equation to look at the number of flips and the number of heads. Is there anything like that?
Dec 31, 2017
0
#4
If there is something like that I think it will inevitably be associated with a confidence interval. That is the nature of statistics. Imagine we toss a coin 1000 and get 800 heads and 200 tails. There is still a chance that the coin is unbiased and the true probability of heads is 0.5; it's just incredibly unlikely. It's much more likely that the true probability of heads is about 0.8. So, in the case I just mentioned, we might say that the probability of heads is 0.8 +- 0.1 (if we want to look at very likely scenarios) or 0.8 +- 0.2 (if we want to include less likely scenarios). Each of these scenarios is a confidence interval and is associated with a particular probability (e.g. 95%).
Dec 31, 2017
0
#5
kaynight wrote:
Pray tell, what is a coin? One deals in £50 notes.
I was talking about bitcoin. What are notes?
corum wrote:
If there is something like that I think it will inevitably be associated with a confidence interval. That is the nature of statistics. Imagine we toss a coin 1000 and get 800 heads and 200 tails. There is still a chance that the coin is unbiased and the true probability of heads is 0.5; it's just incredibly unlikely. It's much more likely that the true probability of heads is about 0.8. So, in the case I just mentioned, we might say that the probability of heads is 0.8 +- 0.1 (if we want to look at very likely scenarios) or 0.8 +- 0.2 (if we want to include less likely scenarios). Each of these scenarios is a confidence interval and is associated with a particular probability (e.g. 95%).
Ok, it makes sense when you say it like that.
So if I want 90% confidence then the strange coin's actual chance of heads is between 43.8% and 60.2%. The range is much larger than I first expected it would be.
Chance: 1/2
Chance:0.5
Chance:50%
Feb 8, 2019
0
#8
Once probability theory (an abstract construct of the mind) is recognized as a myth, a solution is observed.
Let's say I throw a strange coin 100 times. 52 time are heads 48 times a tails. I don't want to throw the coin forever, so I stop and say there will be a 52% chance of heads. Then I realize my sample size is relatively small, but still don't want to throw the coin forever. How can I calculate the 'margin of error'? This margin of error would get smaller and smaller as the sample size got bigger and bigger.
I would eventually want an idea like I get heads 52% of the time plus or minus .5% or something like that.