steve earth

CS grad student at Drexel Univ

[steve earth @earther66]

This has a χ2 value of 17.7 with a p-value of .54. Or in other words, a fair die would have produced a result as skewed as ours more than half the time. So, the video does not statistically support claims of a loaded die afterall. My apologies for the false alarm.

[steve earth @earther66]

okay, NOW, I'm giving an official retraction of my original post. I redid the math of the 87 rolls under the assumption that when 4+ dice are simultaneously rolled, the highest of those is guaranteed to be one of the 4 displayed rolls.

[steve earth @earther66]

In that case, the expected number of each 1-20 would be 3.3, 3.3, 3.3, 3.3, 3.3, 3.3, 3.3, 3.3, 3.3, 3.4, 3.4, 3.4, 3.5, 3.6, 3.8, 4.0, 4.6, 5.8, 8.2, 13.6. The actual frequencies in the video were 0, 2, 5, 6, 4, 3, 2, 2, 0, 5, 6, 2, 2, 4, 4, 4, 6, 5, 12, 13.

[steve earth @earther66]

Not exactly a retraction, but... it was pointed out that the high roll is always displayed when 4+ dice are rolled. That would create an apparent skew where one might not exist. I will have to go back to the math and see how much of an impact that has, as it could be significant

[steve earth @earther66]

Or to put it another way, if we had played this same game 100,000 times with fair dice, only 7 of those would have ended up with die rolls as wonky as what we got. Conclusion: it is very likely that Arena is loading the dice

[steve earth @earther66]

There were 87 visible rolls when the opponent combo'd off with his pixie. I played back the video at one-quarter speed and recorded them all. It's worth noting that there were thirteen 20s and twelve 19s and no 1s. That smelled fishy to me, so I ran a χ2 goodness of fit test.

[steve earth @earther66]

That's basically a fancy statistical term for checking to see if we believe that the die rolls are fair). It came out with χ2 =51.658536585365866, pvalue = 0.00007441219145143396.

[steve earth @earther66]

@TheAcademy Future is a computer simulation.

[steve earth @earther66]

@SCIENCE_MAN_88 @numberphile @jamesgrime well, 8, if you include the single digit ones.

[steve earth @earther66]

@Veganmathbeagle @Trianglemancsd You need scissors and tape: negative "tics" are cutting out parts of the paper :-)

[steve earth @earther66]

@mathequalslove Possible followups: ignoring the emoji restrictions, how many piece placements are there (96) and why? Are there any other solutions that work (no) and why?

[steve earth @earther66]

@danieltybrown this will happen whenever the hypotenuse of a Pythagorean triple is one more than a leg: b+c=1*(c+b) = (c-b)(c+b)=c^2-b^2=a^2. And this happens for all n: 2n(n+1), 2n+1, 2n(n+1)+1

[steve earth @earther66]

@Veganmathbeagle Did any students reason "no game with an odd (6C2=15) amt of equally likely outcomes can be fair"? Though this doesn't tell who it favors...

[steve earth @earther66]

@AccomplishEdu @ProfSmudge one could also argue for F, the only convex shape.

[steve earth @earther66]

@smartereveryday I have a sci question re your latest under water experiment. it's > 140 chars, so here is a link: http://t.co/JXqLFyvCca

[steve earth @earther66]

@mpershan and the original object couldn't be a hexagon because....?

[steve earth @earther66]

@mpershan all you need now is to pull the Shyamalan: the follow up cutaway long shot showing you holding up a flat irregular hexagon

[steve earth @earther66]

I'm looking to teach (math>English>0) in the Far East (China/Hongkong/Korea) for Fall 2013. Am I too late? Contact suggestions? #TeachAbroad

[steve earth @earther66]

AP score distribution

[steve earth @earther66]

@StatsBosses I would think it's *both*. (hopefully all good statistics are also facts).