Coin Flip Experiment

[Camacha]

Is it really that hard to flip a coin?

No. And that is not the point. What I meant was that we know what the result should be, give or take. If it deviates too much, changes are higher that KSP players cannot be trusted in a poll than that we really generated a weird result.

[Drunken Hobo]

My British Pound coin minted in 1995 with a Welsh Dragon design landed on heads. I hope this has been valuable data for your experiment.

[Findthepin1]

We have reached 50-50!

[Javster]

We have reached 50-50!

We're just off it again!

[Ralathon]

My British Pound coin minted in 1995 with a Welsh Dragon design landed on heads. I hope this has been valuable data for your experiment.

Last post before you was nearly 3/4th of a year ago

Anyway, tails here.

[Leszek]

I flipped a coin, it hasn't come down yet. Flipped another, I can't find it. Flipped yet another, tails.

[Kryten]

We have reached 50-50!

Which pretty much proves this isn't actually being done at random. Chance of that is less than 8%.

[ZetaX]

Which pretty much proves this isn't actually being done at random. Chance of that is less than 8%.

Actually, it reaching it at some (arbitrary, non-fixed!) point in time is 100%. And the latter is probably what is happening when someone reports on it.

[Kryten]

It wasn't 50-50 as in 50%, it was 50 heads 50 tails.

[ZetaX]

It wasn't 50-50 as in 50%, it was 50 heads 50 tails.

Ah, ok, didn't realise that.

But while I also think that this is heavily biased (assuming that the 100th person tries to achieve 50-50 makes it more likely by a factor of two; but many might not even look at the result before voting) and even caused by that bias, I wouldn't necessarily dismiss the 50-50 as a result of only that; at least not completely. It is probably enough to happen at pure chance after all, plus we would probably wonder the same about 100-100 and such; if there are "enough" numbers we consider nice/interesting, then the chance for this to happen is still 1.

[Bill Phil]

Actually, some coins are weighted for a particular side. We should specify which coin type is being used...

[ZetaX]

Actually, some coins are weighted for a particular side. We should specify which coin type is being used...

It won't matter if the coins of the world are fair on average; it is not necessary that each single coin is fair.

[Bill Phil]

It won't matter if the coins of the world are fair on average; it is not necessary that each single coin is fair.

It also depends on size, and the actual person flipping, how it's flipped, etc.

If a robot flipped an equally weighted coin 100 times, the results may still not be 50-50.

Btw, the weighting means that a certain side is more likely, changing the results a bit.

[Drunken Hobo]

http://s20.postimg.org/ckt5bsdx9/Thread_Necromancy_3038.jpg

Last post before you was nearly 3/4th of a year ago

Anyway, tails here.

Hah, oops. I had searched for "Flip" to figure out why my rocket was flipping, must have left the window open, came back and assumed this was in the current "Science Labs" forum.

Big surge for tails since I voted. I call shenanigans!

[ZetaX]

It also depends on size, and the actual person flipping, how it's flipped, etc.

If a robot flipped an equally weighted coin 100 times, the results may still not be 50-50.

Btw, the weighting means that a certain side is more likely, changing the results a bit.

I think you are completely missing my point. My point is/was: however unfair a certain coin is, it will probably still average out because there are many types of coins. Restricting the type of coin will probably make it more, not less, biased.

Altogether there is not much reason to assume that coins of all types somehow (dis)favour heads. Any specific coin may, but a random one probably won't.

[LordFerret]

"gamblers-take-note-the-odds-in-a-coin-flip-arent-quite-5050"

[Bill Phil]

I think you are completely missing my point. My point is/was: however unfair a certain coin is, it will probably still average out because there are many types of coins. Restricting the type of coin will probably make it more, not less, biased.

Altogether there is not much reason to assume that coins of all types somehow (dis)favour heads. Any specific coin may, but a random one probably won't.

It depends on what's called "heads". Sometimes the difference is tiny...

It might average out, if there are enough instances.

Restricting the type would allow for a better analysis of the probability. Sure, it introduces bias, but at least it would be known what the bias is for the coin in question.

[K^2]

In either case, our data doesn't show any deviation from fair coin. At 125 tosses, expectation is at 62.5 ± 5.6 heads. We're at 59, which is slightly over half sigma off. p > 0.62 is not what I'd call suspect. Maybe with more data.

[ZetaX]

Restricting the type would allow for a better analysis of the probability. Sure, it introduces bias, but at least it would be known what the bias is for the coin in question.

Even for a heavily biased coin like the one linked above, you would need much more experiments to verify that (assuming people are actually flipping a coin and not making up results). And fixing the coin would probably exclude a majority of people not having access to that specific coin.

[Lukaszenko]

Apparently catching a coin in mid air is more "random" than letting it fall to the ground, due to the fact that coins are usually biased.

Interesting video on the topic:

[Caelib]

The results of this experiment are already known: Law of Large Numbers