Dice rolling statistics.

[Marshal Wilhelm]

Greetings brethren,

 

I have had trouble deciding where to put this OP. I decided here....

Hopefully this has not been spoken of before.

 

+++

 

So I stumbled across this: Dakka dice rolling test. [please excuse the rude gesture in the image]

as a chap on BoLS said ones come up most often and I was quite sure that sixes, being lightest and opposite the heaviest face, one, would be slightly the most probable.

 

Mathhammer says each face has a 1/6 chance of coming up, or 16.67%. With my idea of six being lightest [due to having more material taken away and dice companies not being greatly desperate for mathematical precision to eventuate] this leads me to think that a six would come up perhaps 17-18% of the time, for example.

 

However, the Dakka guy rolled many dice in a controlled setting, many times, and one came up 29% of the time. 29%! That is roughly 50% more than what mathhammer says! That is massive!

This was with GW dice. When rolling square dice [dice without rounded corners] the ones came up some 19% of the time, which isn't really how it should be, ideally, but much better.

 

Has anyone else conducted controlled dice rolling tests and if so, do you have the data to share?

[Deus de Mortalis]

With my idea of six being lightest [due to having more material taken away and dice companies not being greatly desperate for mathematical precision to eventuate] this leads me to think that a six would come up perhaps 17-18% of the time, for example.

 

The article explains, albeit confusingly and poorly, why this conventional wisdom is incorrect.

 

Thank you so much for posting this though. I am going to head to ebay to buy some casino dice ASAP.

[breng77]

Casino dice are a poor choice as in order for them to get a fair roll they must generally bounce off of a bumper as in Craps. With the limited amount of rolling space on most tables it is unlikely you will get a fair roll. Also, it is unlikely that any test really has enough rolls to be statistically relevant (even 1000 rolls for one dice is a fairly small sample). Furthermore, unless you have some sort of dice rolling mechanism controlled dice rolling is not going to occur as slight variations of how you hold the dice or toss them onto the surface will cause differences.

[Marshal Wilhelm]

I guess what I am saying is I just cannot believe it.

 

I am not saying that the test wasn't done or that those weren't the results, but I am saying has anyone else followed his procedure and had similar results?

Trying to be as scientific as possible [meaning nothing is accepted easily as fact nor is anything dismissed out of hand] I wonder if it just happened to turn out the way it did and now we have it as a 'fact' that rounded D6s roll 'ones' 29% of the time.

 

Lets say I roll a purple d6 300 times and record the results. Which I did ^_^

I used a similar technique [not being a robot :lol: ] each time and released the dice from a similar time.

My results:

1 - 51 times

2 - 52 times

3 - 42

4 - 50

5 - 47

6 - 58

 

I know my sample is very small, too small to take results from. However, would you not think that 'one' would have come up more than practically 1/6 times, given the Dakkaites findings?

 

Lets put it another way:

The Dakkaite gets one come up 29% of the time. That leaves 71% for the other five results. 14.2% for the rest, things being equal. That means a 'one' should be coming up twice as often as any other result. That certainly didn't happen with my small sample, nor is it something that I and my two brothers have noticed with decades of rolling dice experience.

 

I am not convinced!

[Deus de Mortalis]

I guess just look for dice that do not have the pips drilled out. That seemed to be what was causing the problem by unbalancing them. But also he and apparently 4 students did all the dice rolling, and they rolled each variety tens of thousands of times. He also asked a Casino about it and they agreed with his conclusions. He also said that when he asked Chessex about it, they said that they save a lot of material making the dice that way. Anyway, idk. I hope we can learn more.

[LardO'Blood]

This frustrates me. From now on I'm going to use block dice when not rolling for morale checks!

I knew my Deathwing weren't as survivable as my PA guys...

[Ming]

Qualitatively I see my dice rolling results being different depending on surface. Glued sand surfaces being the worst towards getting 1's. Smooth felty surfaces do seem to roll fairest. I have a flat tray with raised back edges to carry my army around on - I use it for my dice rolling tray, rather than putting dice on the table. I roll the dice into the back edge so they bounce back, which seems to give me the best comfort towards a random roll. As an alternative I'd recommend bringing a folding felt-lined backgammon case...gets similar to having a gambling table. Another blogger (Thor) noted some studies that dice do wear out (edges get rounder) which also changes how they play with time. In a similar way casinos chuck their dice regularly.

 

Our FLGS tables are 4x8 tables - space for a 4x6 game surface, 2 18inch by 2 foot army trays, and a 1x2 open area for dice rolling or a felt dice tray...something to consider. These are awesome on tournament days. Every table has plenty of space so nothing goes under the table or on the floor.

[pingo]

I wonder if it just happened to turn out the way it did and now we have it as a 'fact' that rounded D6s roll 'ones' 29% of the time.

 

That is a possibility, but the chance would be further reduced with further experimentation. One experiment is not enough to prove something. Science is all about other replicating, or trying to falsify, your findings.

 

Lets say I roll a purple d6 300 times and record the results. Which I did :D

I used a similar technique [not being a robot :) ] each time and released the dice from a similar time.

My results:

1 - 51 times

2 - 52 times

3 - 42

4 - 50

5 - 47

6 - 58

 

I know my sample is very small, too small to take results from. However, would you not think that 'one' would have come up more than practically 1/6 times, given the Dakkaites findings?

 

No. You identified your own problem: sample size. One case is insufficient to establish anything other than the fact that case occurred. You may expect the bias to occur, but the fact it didn't that one time doesn't really tell us anything. You could be right, or your dice could be very well made. Or perhaps you experienced a freak occurrance, or the way you roll, or hold the dice, happens to bias your results.

[Ming]

There is a company in the US that I've seen at histoorical game conventions that sells a dice tower - an actual cardboard device that looks like a tower / keep. You drop dice in the top, they come out the bottom to a collection tray all randomly shaken and stirred...ready to look at and seemingly random. Takes the "practiced" rolling right out of the picture.

[breng77]

The problem I see with dice towers, is that you need space to place them on the table, and they only work correctly with dice of a particular size.

[JamesI]

There is a company in the US that I've seen at histoorical game conventions that sells a dice tower - an actual cardboard device that looks like a tower / keep. You drop dice in the top, they come out the bottom to a collection tray all randomly shaken and stirred...ready to look at and seemingly random. Takes the "practiced" rolling right out of the picture.

I have one of those. It works well with normal GW sized dice, I just tend to forget to bring it with me to my games since it takes some extra space.

[Wildfire]

There is a company in the US that I've seen at histoorical game conventions that sells a dice tower - an actual cardboard device that looks like a tower / keep. You drop dice in the top, they come out the bottom to a collection tray all randomly shaken and stirred...ready to look at and seemingly random. Takes the "practiced" rolling right out of the picture.

 

You can find those things all over the place. I've seen lots of different ones. They're not too hard to make yourself, come to that.

[Grey Mage]

My friends and I did it one weekend, because they were sure at the time *two months in* that my dice were somehow rigged. Got the following results rolling the same 10 dice 100 times each. I know, statisticly its a small sample size but its all Ive got

 

Me:

1 12%

2 17%

3 14%

4 15%

5 24%

6 18%

 

Eric:

1 16%

2 19%

3 14%

4 16%

5 16%

6 19%

 

Rob:

1 18%

2 19%

3 15%

4 16%

5 15%

6 17%

 

John:

1 15%

2 19%

3 15%

4 15%

5 16%

6 20%

 

Overall more 6's than ones, but the person seemed to matter as much as anything. Is that the technique variable? Is it luck? I dunno. Wanted to do more on it, but the group split up before I could convince them to 'waste time' on it again.

 

Edit: For the record, 5 of the dice are from an old yahtzee set, 2 are from a monopoly set, and the other 3 were of unknown manufacture when I get them. Only 2 had significantly rounded sides- the ones from the monopoly set.