Tilted Forum Project Discussion Community  

Go Back   Tilted Forum Project Discussion Community > The Academy > Tilted Knowledge and How-To


 
 
LinkBack Thread Tools
Old 05-02-2004, 10:17 AM   #1 (permalink)
Insane
 
Location: Denver, CO
Probability and Random Chance

My question is a little involved, but here goes:

Let's say you're flipping a coin--you have two possible outcomes, heads and tails. The likelihood of heads coming up once (in one toss) is 1/2, twice (in two tosses) is 1/4, three times (in three tosses) is 1/8.

So we can generalize that the chance of heads coming up n times in n tosses is 1/2^n.

But how can we generalize the chance of heads coming up any x times in n tosses, where x<n? For example, what's the probability of any 7 of 10 tosses coming up heads?

Next, let's say that I have a friend who claims he's psychic, and can predict the outcome of a coin toss. If he gets 7 out of 10 tosses right, that's pretty good, but the generalization above gives us the probability of such an outcome happenning by random chance alone. At what point does this potential for an outcome by chance alone become statistically negligible?

I know that, even with a million tosses, there's still an outside chance that he could get them all correct by randomly guessing. But it's so slim a chance that, if he could get that many right (and especially if he could repeat it), then I'd presume that chance alone did not dictate the outcome.

And then I'd wear a tinfoil hat so that he couldn't read my mind.

Edit: I'm not trying to prove/disprove psychic abilities, I'm just using it as an example.
__________________
"We must have waffles. We must all have waffles, forthwith. Oh, we must think.
We must all have waffles and think, each and every one of us to the very best of his ability."
-- Professor Goldthwait Higginson Dorr, Ph.D.
FleaCircus is offline  
Old 05-02-2004, 01:49 PM   #2 (permalink)
Wehret Den Anfängen!
 
Location: Ontario, Canada
Quote:
Originally posted by FleaCircus
So we can generalize that the chance of heads coming up n times in n tosses is 1/2^n.
That's n times in a row, and correct.

Quote:
But how can we generalize the chance of heads coming up any x times in n tosses, where x<n? For example, what's the probability of any 7 of 10 tosses coming up heads?
It works out to
10C7 * 1/2^10
IIRC.

10C7 is the number of ways you can choose 7 elements out of a list of 10 elements.

It works out to
10!
---------------
(10-7)! * 7!
where n! = 1*2*3*...*n

In this case, it is
10*9*8
----------
3*2*1

which is 120.
1/2^10 * 120 is about 12%.

Quote:
Next, let's say that I have a friend who claims he's psychic, and can predict the outcome of a coin toss. If he gets 7 out of 10 tosses right, that's pretty good, but the generalization above gives us the probability of such an outcome happenning by random chance alone.

In reality, you care about the chance he got 7 8 9 or 10 right (you would be no less impressed by 8 or 9 or 10). That's

(10C7 + 10C8 + 10C9 + 10C10) * 1/2^10
=
(120 + 45 + 10 + 1) * 1/2^10
=
176 * 1/2^10
= approximatally
17%

So, getting 7 coins right is about a 1 in 5 occurrance.

Quote:
At what point does this potential for an outcome by chance alone become statistically negligible?
Statistical negligibility isn't a hard and fast point. How important is it?

However, what you care about is:
What is the chance that his ability to predict coins more accurate than 50%?

My statistics is too rusty to answer that one.

Quote:
I know that, even with a million tosses, there's still an outside chance that he could get them all correct by randomly guessing. But it's so slim a chance that, if he could get that many right (and especially if he could repeat it), then I'd presume that chance alone did not dictate the outcome.
Repeating it is quite important.

If there is a 1/10 chance he can do it once, there is a 1/100 chance he can do it twice, and a 1/1000 chance he can do it 3 times.

Of course, assuming he can cheat somehow is still a good hypothesis. Cheating at coin flips is surprisingly easy, unless you are careful.

A friend of mine used to do a neat card trick.

He'd get someone to shuffle a deck of cards, then select one card, look at it, and put it back.

Without touching the deck, he'd then try to guess what the card is.

By guess, I mean, he'd pick a random card, and say "was it a 7 of clubs?"

1 / 52 he was right, and the person he pulled the trick off on was very impressed... ;-)
__________________
Last edited by JHVH : 10-29-4004 BC at 09:00 PM. Reason: Time for a rest.
Yakk is offline  
Old 05-02-2004, 08:00 PM   #3 (permalink)
Insane
 
Location: Denver, CO
Awesome, thanks for the help.
__________________
"We must have waffles. We must all have waffles, forthwith. Oh, we must think.
We must all have waffles and think, each and every one of us to the very best of his ability."
-- Professor Goldthwait Higginson Dorr, Ph.D.
FleaCircus is offline  
Old 05-03-2004, 06:08 PM   #4 (permalink)
Insane
 
Location: Denver, CO
I did a little more research, and this is what I came up with.

Quote:
If an event occurs N times (for example, a coin is flipped N times), then the binomial distribution can be used to determine the probability of obtaining exactly r successes in the N outcomes. The binomial probability for obtaining r successes in N trials is:

where P(r) is the probability of exactly r successes, N is the number of events, and (pi symbol) is the probability of success on any one trial. This formula assumes that the events:
(a) are dichotomous (fall into only two categories)
(b) are mutually exclusive
(c) are independent and
(d) are randomly selected
Taken from HyperStat Online
__________________
"We must have waffles. We must all have waffles, forthwith. Oh, we must think.
We must all have waffles and think, each and every one of us to the very best of his ability."
-- Professor Goldthwait Higginson Dorr, Ph.D.

Last edited by FleaCircus; 05-03-2004 at 06:11 PM..
FleaCircus is offline  
Old 05-03-2004, 06:33 PM   #5 (permalink)
Addict
 
Fleacircus: If you think the guy is psychic you should take him to vegas and forget the coins. No, seriously in statistics we use alpha and the null hypothesis to determine significance. The null or null hypothesis is the hypothesis of no difference. In the case of your friend the null hypothesis states that his ability to predict the side the coin will land on is no better than anyone else in the population. This is where alpha comes in. Alpha is usually set at .05 which means that he would have to correctly predict the side it landed on more than 95 times out of a hundred to reject the null hypothesis. If he did correctly predict the side the coin would land on 96 times out of 100 you would reject the null hypothesis and he could say that the results of his test throws supported his hypothesis that he was psychic. Just as an add on I have never seen any research that showed a psychic was significantly better at telepathy than the rest of the world.
zfleebin is offline  
Old 05-04-2004, 08:57 AM   #6 (permalink)
On the lam
 
rsl12's Avatar
 
Location: northern va
fleacircus: check out any basic college statistic textbook and look up the two-tailed student t-test. that will give you what you need to know about determining with confidence that your friend has something up his/her sleeve.
__________________
oh baby oh baby, i like gravy.
rsl12 is offline  
 

Tags
chance, probability, random


Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On



All times are GMT -8. The time now is 10:54 AM.

Tilted Forum Project

Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2025, vBulletin Solutions, Inc.
Search Engine Optimization by vBSEO 3.6.0 PL2
© 2002-2012 Tilted Forum Project

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47