Coin Flip Experiment

[gamowin]

Here's what you do, find the nearest coin, and flip it, then vote whether it landed on heads or tails, that's it. Let's see if we actually get a perfect 50/50 split! (Sorry if I posted this in the wrong place, I wasn't really sure where to put it.)

[phoenix_ca]

You'll get a 50/50 split eventually, you just need a large enough sample size.

[Seret]

At first I thought this was a dumb thread, but it's not a bad discipline to reality-check your ideas with a real experiment even if you're confident how it'll turn out.

[shynung]

I'm pretty confident by the time the sample reaches 100, it'd be pretty close to a 50/50 split. However, I'd be quite surprised if it is exactly 50/50 by then.

[bsalis]

Tails here.

Poll result so far = 7:7

[sjwt]

I'm pretty confident by the time the sample reaches 100, it'd be pretty close to a 50/50 split. However, I'd be quite surprised if it is exactly 50/50 by then.

The expected observation is of course as the sample approaches infinity the odds close in on 50/50

[Luis]

This is a bad experiment, since every coin flip happens under different conditions that are impossible to validate. You can't crowd source anecdotes and call the aggregate "data".

I know this was just a lighthearted thing, but sheesh.

[ZetaX]

If you assume that people are honest with their entries, then the different conditions don't matter.

[lajoswinkler]

If you assume that people are honest with their entries, then the different conditions don't matter.

Actually, they do. You don't know what kinds of coins people use. When doing such experiments, number of variables has to be kept at minimum. But let's try it, nonetheless.

[AngelLestat]

I dint flip the coin, I just choose one.

And I guess many people is doing the same.

So this is not a good experiment about probability, becouse it seems that we all are a little more predisposed to choose Tails.

Maybe becouse you place it in second place or just becouse the word sound nice.

[ZetaX]

Actually, they do. You don't know what kinds of coins people use. When doing such experiments, number of variables has to be kept at minimum. But let's try it, nonetheless.

They are still coins, i.e. two-sided, and I presume that everyone maps whatever the two sides show to the two outcomes from the poll. The actual probabilities of each single coin are irrelevant.

[Charzy]

One side will have more metal than the other due to differing designs on the coin, which means that side will land slightly more often.

Also, it's more slightly likely to land on whichever side it's on when you flip it.

[PakledHostage]

They are still coins, i.e. two-sided, and I presume that everyone maps whatever the two sides show to the two outcomes from the poll.

I tossed a loonie... It has a loon on one side (tails) and Queen Elizabeth's head on the other (heads). The coin landed loon side up. Tails it is.

Edited June 7, 2014 by PakledHostage
fixed a typo

[ZetaX]

One side will have more metal than the other due to differing designs on the coin, which means that side will land slightly more often.

Also, it's more slightly likely to land on whichever side it's on when you flip it.

The former only applies if everyone would use the same coin (and then would give a statictically significant result on that in the long run), while in praxis we can probabaly assume that different coins from all over the world will get rid of this effect. Furthermore, I am not sure if your conclusion that it lands more often on that side are correct.

[K^2]

I didn't have a coin nearby. I flipped a key. It's flat, and spins around the long axis just the same. It came up "tails".

By the way, what we're looking at here is the binomial distribution. While the expected average number of "heads" is n/2, the standard distribution is sqrt(n/4), which is quite significant. At the moment, the split is 15:25, which is 40 coin flips. So the expected split is 20:20 with variance of ±3.1. The actual split is less than two standard distributions off, which is a likely enough outcome.

[Bacterius]

This poll might work, except for the "view poll results" button, which will invariably cause a portion of the population to either vote towards making the result closer to or farther from 50/50, thus biasing your experiment. Yep, goes to show, if you want something done right, you really can't trust others to do it for you.

You're welcome.

[Red Iron Crown]

Probability Theory: Inviolable rules, no hard predictions.

[rtxoff]

Someone could repeat the experiment but with a suggestion to try to throw the coin to be heads.

Wondering of the outcome it would produce. Can a human influence the outcome if he tries?

[tutrakan4e]

Tails, I flipped it.

[FanaticalFighter]

I tossed a Rubik's Cube, it came up yellow. I decided that meant it was tails.

[SolarLiner]

That's a cool experiment to do, I one thought of making you guys thinking a number between 1 and 10, and vote the result. Everyone will know it'll come out as a nice Gauss curve, but it'll be a cool experiment to do too.

[ZetaX]

No, studies show it will accumulate at 7.

[softweir]

This is a bad experiment, since every coin flip happens under different conditions that are impossible to validate. You can't crowd source anecdotes and call the aggregate "data".

I know this was just a lighthearted thing, but sheesh.

You have a point. When humans toss a coin (using the accepted thumb-toss technique and allowing the coin to fall onto a surface) the result can have a distinct bias depending on the size and weight of the coin and how far the coin has to fall before settling. If a large number of people engaging in this experiment have the same bias, then the result will be biased.

Still, this is a fun and interesting thread, just so long as people are aware of its limitations.

[N_las]

You have a point. When humans toss a coin (using the accepted thumb-toss technique and allowing the coin to fall onto a surface) the result can have a distinct bias depending on the size and weight of the coin and how far the coin has to fall before settling. If a large number of people engaging in this experiment have the same bias, then the result will be biased.

Still, this is a fun and interesting thread, just so long as people are aware of its limitations.

It is flawed not because of little insignificant things like that. It is flawed because it is not possible to validate if someone really did flip a coin. I picked the answer that was in front - just to increase the difference from 50/50.

[rtxoff]

It is flawed not because of little insignificant things like that. It is flawed because it is not possible to validate if someone really did flip a coin. I picked the answer that was in front - just to increase the difference from 50/50.

Well the true question here is, does it matter if the brain decides or the coin decides. Both can be random. And when the brain decides everyone of us has different reasons for picking one side.

I guess it really does not matter.