BBC article - MP defies 58,000/1 odds in ballot Estimated at 58K? Who by? A seven year old with a crayon? 240 entered the draw this year, oh and what a coincidence 240 squared is 57,600. Rounding takes care of the rest. So they're saying that the odds of an MP being selected twice are approx 58K to 1. The odds are actually 240 to 1. The odds of rolling the same number twice on a six sided dice are 1 in 6, not 1 in 36. Anyone with no grasp of basic statistics needs to go back to school.

The odds of an MP winning twice is 240 to 1. Him winning twice is 58k to 1, but the chance of him winning again is 240 to 1. So his comment about visiting the bookies a few days ago is a little off...

This. The odds of it happening separelty are both 1 in 6, but to happen one after another is the probabilities multiplied together. Your main point about journalists basically making up stats is correct though. In Brazil you have to have to be educated and have a license to be a journalist.

You roll a die and you get 3. That was a 1 in 6 chance. To roll again and get a 3 is also 1 in 6 YES, but what you actually said was about the probability of the event "rolling twice and getting the same number", which actually is 1/36.

The odds of rolling the same number on a dice twice are 1 in 6. I can't think how it couldn't be. There are 36 possible outcomes, and 6 of those are rolling the same number twice.

If it was 1 in 6 then that would be like saying you have a 1 in 4 chance of getting 100% in an exam with 1000 questions each with 4 options, which is of course nonsense. You have a 1 in 4 chance of getting a question right, i.e. you'll average 25%, not 100%.

It is 1 in 6. Pairs of 1,1 + 2,2 + 3,3 + 4,4 + 5,5 + 6,6 is 6 pairs and there's 36 combinations so 6 in 36 or 1 in 6. The odds to roll a predetermined number pair so 1,1 is 1 in 36 though.

What the article states is that he has beaten those odds by being picked/coming top twice. That's the same as saying "number 3 has a 1/36 chance of coming up twice". :edit: Urgh, you're going to twist what I've said because I've left holes in it. Damn my tired mind What Pudney says below.

The article is saying an MP has come up twice and that's the same as 58K to 1 odds. The article would have been written regardless of who the MP was. It's not remarkable that it was John McDonnell twice rather than Douglas Carswell twice, it's remarkable that it was the same MP 2 years in a row and they're saying the odds are 58K to 1.

The odds prior to his first election for him to be elected twice would be 58k to 1, then after he's been elected it would be 240 to 1 to be re-elected. Think that should simplify it enough.

If they said that the chances of him winning again were 58k to 1 then yeah, I see your point, but they didn't. They said the chances of him winning twice were.

I believe the best explanation is the Texas Sharpshooter Fallacy., or the conviction of Lucia de Berk.

I do think I get what you're saying after re-reading everything... the odds for any MP to be elected twice is 240 to 1. So meaning that every 240 times it should happen once. As in out of 58k times it would happen 240 times (due to there being 240 MPs)

If before either of these events took place you calculated the odds of that specific person winning twice, you would get the 58k value. You don't need to explain statistics to me, I understand exactly what you mean but it doesn't apply to what they've stated. What they're implying, maybe... but since when has the news ever implied sensible things ?

Precisely, and it's irrelevant to their article which MP it was. They're reporting on a 240 to 1 chance as if it's something truly remarkable, and reporting it as a 58,000 to 1 chance. The case I linked is of a Dutch nurse who was convicted of multiple murders essentially based on the statistical probability of her having been innocent as 1 in 342 million - it was a fallacy and overturned (after an unsuccessful appeal). Basically it's about looking at an event in retrospect and saying it's incredible because of the chances of it happening were very small.