Request for Help with Experiment

[complexmelody]

Hi. I am new to this forum. My father has been a member here for a few years and suggested I get input on an experimental design I am working on.

I am in 6th grade and I'm working on a science fair project. My idea for the project is, Can You Control the Roll of Dice with your Mind?. I am going to test two different groups of people; those who do believe they can and those that don't. I will make them roll four dice 25 times (how about ten times?) and I will record the total number of 6's they roll. I will make them roll the dice in a cup to ensure that they don't manipulate the dice with their hands. I will also tell them that they will get money if they roll a certain number of 6's. For instance, if they roll 15 out of 25 6's, they will get $20.

If the "believe" group rolls significantly more sixes that the "non-belief" group then I will have to conclude that there is something to it. I don't know advanced statistics, but I will just look at the numbers and graph the results.

Is there any way I could make this experiment more accurate or fair? How many people for each group do you think I need to test? Thanks for any ideas that you have!

 

[AvalonXQ]

Welcome to the forum!
How do you plan to find people who believe that they can control dice with their minds?

 

[Loss Leader]

CM -

Why are you offering money? That's usually a method researchers use to create an incentive to participate. But the believers already have an incentive: they want to show that they can do it.

If you offer money, and both groups get the same results, you end up having an extra variable to explain. The offer of money causes even people who don't believe in ESP to use their innate powers of ESP.

In fact, I'm a little confused why you've even got a group of nonbelievers. The believers are either going to do significantly better than chance or they're not. Your control group is mathematical chance, not other people.

But I am very impressed at your work trying to set up an unbiased, fair and meaningful experiment. Good luck with your project.

 

[roger]

1. I don't see any point to paying the money.

2. I would not allow them to roll the dice themselves, even with a cup, though that is probably not a big deal giving they are rolling 4 dice at once.

3. This is my bigger objection/observation. You are testing for if somebody can control dice with their mind, but your experimental control is splitting people between people that believe they can vs those that believe they cant. This does not fit in with your stated purpose.

To test if somebody can do something, say, balance a pencil on their nose, you just hand them a pencil and tell them to try. You don't break people up into groups of belief unless you are testing how belief affects performance. Do you see that is a different thing?

Likewise, if you want to test if people can control dice, the test is not how they do compared to some other group, but how they compare to pure chance. For example, suppose people *can* control dice, but not in the way you think. If you believe you can control dice, you really can't, and if you don't believe you can, your psi negatively affect the die. Or something. The point it you'd end up drawing the wrong conclusion - that belief means you can control dice.

So, pure chance says that the dice should roll 6 1 in 6 times. So, I'd run the experiment for some medium mutiple of 6 rolls - say 66 times for each person*. If it's 66, then you would expect 11 sixes to be rolled on average. If you have 10 people, then there should be around 110 sixes rolled in total.

We could tell you how to compute the statistical significance of a result, but that smacks of doing your homework for you, so I won't do it. The most you can hope for is "that's really interesting" or "appears like people don't do much better than chance" results.

You will undoubtedly find some people roll significantly more than 11 sixes in 66 rolls. It would be very interesting to then retest them. It could be they just got that larger number by chance, or it could be they are actually controlling the dice. If they are controlling the dice, then if you run several more trials with that person, they should continue to roll more sixes than average. That's a pretty good way to work around the fact that your math is not yet up to fully analyzing the experiment.

edit: having them roll multiple die at once could be a confounding factor. I don't believe people can control dice with their mind, but if they could, I could well imagine it would be easier to control 1 rolled at a time rather than 4. For example, I can easily catch a baseball, but I'm pretty challenged to catch 2 at the same time. Since we don't have a proposed mechanism for how you might control dice throws with your mind, you shouldn't make assumptions that someone can control 4 at once.


* this is NOT a recommendation to do 66 roles, or 10 people. It's an example. I think it's okay to help you think about your experiment, but not to design it for you.

 

[Professor Yaffle]

Look up the "sheep-goat effect" for similar experiments looking at differences in psi performance between believers and non-believers of psi.

 

[complexmelody]

Welcome to the forum!
How do you plan to find people who believe that they can control dice with their minds?

I don't know how I'm going to find people who believe. I was thinking of not dividing the people into groups. Just test the general public, and not take into account if they believe or not.

 

[complexmelody]

1. I don't see any point to paying the money.

2. I would not allow them to roll the dice themselves, even with a cup, though that is probably not a big deal giving they are rolling 4 dice at once.

3. This is my bigger objection/observation. You are testing for if somebody can control dice with their mind, but your experimental control is splitting people between people that believe they can vs those that believe they cant. This does not fit in with your stated purpose.

To test if somebody can do something, say, balance a pencil on their nose, you just hand them a pencil and tell them to try. You don't break people up into groups of belief unless you are testing how belief affects performance. Do you see that is a different thing?

Likewise, if you want to test if people can control dice, the test is not how they do compared to some other group, but how they compare to pure chance. For example, suppose people *can* control dice, but not in the way you think. If you believe you can control dice, you really can't, and if you don't believe you can, your psi negatively affect the die. Or something. The point it you'd end up drawing the wrong conclusion - that belief means you can control dice.

So, pure chance says that the dice should roll 6 1 in 6 times. So, I'd run the experiment for some medium mutiple of 6 rolls - say 66 times for each person*. If it's 66, then you would expect 11 sixes to be rolled on average. If you have 10 people, then there should be around 110 sixes rolled in total.

We could tell you how to compute the statistical significance of a result, but that smacks of doing your homework for you, so I won't do it. The most you can hope for is "that's really interesting" or "appears like people don't do much better than chance" results.

You will undoubtedly find some people roll significantly more than 11 sixes in 66 rolls. It would be very interesting to then retest them. It could be they just got that larger number by chance, or it could be they are actually controlling the dice. If they are controlling the dice, then if you run several more trials with that person, they should continue to roll more sixes than average. That's a pretty good way to work around the fact that your math is not yet up to fully analyzing the experiment.

edit: having them roll multiple die at once could be a confounding factor. I don't believe people can control dice with their mind, but if they could, I could well imagine it would be easier to control 1 rolled at a time rather than 4. For example, I can easily catch a baseball, but I'm pretty challenged to catch 2 at the same time. Since we don't have a proposed mechanism for how you might control dice throws with your mind, you shouldn't make assumptions that someone can control 4 at once.


* this is NOT a recommendation to do 66 roles, or 10 people. It's an example. I think it's okay to help you think about your experiment, but not to design it for you.

I think offering money to people will make them try their hardest to roll 6s. I was also thinking that maybe I could offer money to some people and not offer to others. I'll see if that makes a difference. I do agree with you that if people could control dice with their mind, it would be easier to control 1 at a time instead of 4. I just thought that it would make the test go a little quicker. But if you think that it would negatively affect the test, I could change it to them rolling it just 1 at a time.

 

[roger]

I think offering money to people will make them try their hardest to roll 6s.

Well, I don't care if you give them money, really. I would think though, that the prospect of being the first human in recorded history to prove they have psi powers would be far more motivating than a Jackson. But then I often overestimate the nobility of humans

 

[Loss Leader]

I think offering money to people will make them try their hardest to roll 6s. I was also thinking that maybe I could offer money to some people and not offer to others. I'll see if that makes a difference.

In that case, you may have a problem with sample size. If you've got some people who believe, some who don't, some believers who are offered money, some non-believers who are offered money, etc., then each distinct group may end up being very, very small - two or three people. At that size, random error and sheer luck will ruin your data.

For example, if you were to take a poll in 2008 as to who should be elected President, you would find that 54% (or whatever) of likely voters were supporting Obama. But if you shrank the sample size to just my house, 100% of likely voters supported Obama. Worse, if you miscoded the data, you would accidentally report that 100% of likely voters supported McCain.

You need to limit your variables to keep your sample size up. Otherwise, randomness is going to make your experiment worthless.

 

[DallasDad]

No matter what you end up doing, I'm pleased that a 6th-grader is contemplating this kind of experiment, and is interested in making sure the design is appropriate.

Good for you!

 

[Cuddles]

I would suggest rather than trying to split people up into believers and non-believers (which has the problem of actually finding enough believers in the first place), instead simply split into a group that will try to affect the dice and a group that won't.

While Loss Leader and Roger are correct that a control group should not be strictly necessary for an experiment like this, I don't think it's a bad idea to include one. Firstly, it's always possible that you make a mistake in the calculations (or the theory is wrong to start with), or that there is some effect that skews the results one way or another. If your "trying" group gets results well above chance that doesn't necessarily mean anything if the control group do as well, it can simply mean there is something not counted for in the theory or experiment, and that can be just as interesting scientifically as getting the intended result would be.

Secondly, it's a school science fair project. The purpose isn't necessarily to demonstrate some amazing original research following post-grad standards, it's to demonstrate that you know how to. Or at least, know how to as much as a school pupil might be expected. Control groups may not be strictly necessary in all experiments, but demonstrating that you know what one is and how to set it up could be just as important for the class as the result of the experiment itself. It's the classic problem of showing your workings - just writing down the answer is nice, but if you demonstrate that you know how to get the answer you can get credit even if a mistake means the final answer is wrong.

 

[fls]

I second what Cuddles said.

I have four kids and I know what it's like having to come up with science fair projects. I'm impressed with your idea. Don't forget you will need to get informed consent working with human subjects (I mention this so you won't go through all the work of putting together an experiment only to have it rejected at the last minute).

Offering money is a good idea. One of the ideas among parapsychology researchers is that we use psi to get rewards. Offering a reward for achieving a goal is better than just giving them a goal. It doesn't have to be money. Rewards that they use include pleasant pictures, playing a fun game, sweets (like cookies), etc.

I would suggest rolling one die at a time, in a cup like you suggested. If you have several dice, then you can run into problems with judgement calls which will skew the results. For example, a die may be tilted against the other dice and you have to decide which face is up.

Your control group could be asked to guess the number on the die (after it has been rolled, but before you look under the cup) instead of trying to get a six. The guesses won't really be part of the experiment. They will just be a way to see how many sixes are rolled when the subject isn't trying to control the die.

Each subject should be asked to do the same number of rolls. The more rolls you can get, the better. Try it yourself and see how many times you can roll the die and write down the results in five minutes. If you can do 100, it should be reasonable to ask your subjects to do it.

The subjects should be randomly assigned to the "sixes" group or the "guesses/control" group, but you probably want equal numbers in each group - if you start with four people, you will want two in each group.

Linda

 

[complexmelody]

I would suggest rather than trying to split people up into believers and non-believers (which has the problem of actually finding enough believers in the first place), instead simply split into a group that will try to affect the dice and a group that won't.

While Loss Leader and Roger are correct that a control group should not be strictly necessary for an experiment like this, I don't think it's a bad idea to include one. Firstly, it's always possible that you make a mistake in the calculations (or the theory is wrong to start with), or that there is some effect that skews the results one way or another. If your "trying" group gets results well above chance that doesn't necessarily mean anything if the control group do as well, it can simply mean there is something not counted for in the theory or experiment, and that can be just as interesting scientifically as getting the intended result would be.

Secondly, it's a school science fair project. The purpose isn't necessarily to demonstrate some amazing original research following post-grad standards, it's to demonstrate that you know how to. Or at least, know how to as much as a school pupil might be expected. Control groups may not be strictly necessary in all experiments, but demonstrating that you know what one is and how to set it up could be just as important for the class as the result of the experiment itself. It's the classic problem of showing your workings - just writing down the answer is nice, but if you demonstrate that you know how to get the answer you can get credit even if a mistake means the final answer is wrong.

I like that idea, splitting people up into trying to roll 6s and not trying to roll anything in particular. I think I'll use that in my experiment. Thanks for the great suggestion!

 

[complexmelody]

No matter what you end up doing, I'm pleased that a 6th-grader is contemplating this kind of experiment, and is interested in making sure the design is appropriate.

Good for you!

Thanks for your support!

 

[Nursefoxfire]

No matter what you end up doing, I'm pleased that a 6th-grader is contemplating this kind of experiment, and is interested in making sure the design is appropriate.

Good for you!

I concur! I know I was nowhere near this sophisticated in my thinking in 6th grade, and I doubt many of my peers were too (probably the result of my strict fundamentalist OT upbringing, quashing my rationality).

Good luck with your experiment, and be sure to keep us posted here!

p.s. is anyone else curious who complexmelody's dad is on JREF? Care to share that with us? I think he deserves some respect for raising an obviously bright and imaginative kid, who can actually write well

 

[complexmelody]

My dad is grunion. I think some of you might know him. Thank you so much for all of the complements.

 

[Emet]

My dad is grunion. I think some of you might know him. Thank you so much for all of the complements.

Does your dad know you identified him?

Best wishes for your upcoming science fair. I hope you have fun, and your project is successful.

 

[complexmelody]

Thanks! Yes, he does know.

 

[Anerystos]

The statistical test you want is chi-squared. Any text book on statistics for psychology should contain it. I'll try to post a reference later - I'm not in my office at the moment.

 

[jakesteele]

There's two things that stand out to me. One, I don't think you should offer money because the people would not be thinking, or participating with the same mindset as if you said, "Give it go."

The second thing is letting them roll their own dice. I worked in the casino world for 10 yrs. and believe me, every roulette dealer learns to control the speed the wheel is turning and releasing the ball at a specific section of the wheel with the same general speed of release to have the ball drop onto a specific group of numbers. I could do it 6-7 times out of 10. Of course, I only did it for hot babes and friends.s with their hands.

So the moral is, can have some influence on the spin of a wheel, throw of the dice/cup. For instance, on a craps table you will have people face up the same numbers on the dice and attempt to throw then straight in the air without any shake or spin to them. The casino won't let you do that.

I would suggest you try to build a small, round cage like you see on the state lottery games.

Good luck, young man and tip of the hat you. You are a very enterprising guy.