Statistics

For stuff that is general.
Post Reply
GORDON
Site Admin
Posts: 54400
Joined: Sun Jun 06, 2004 10:43 pm
Location: DTManistan
Contact:

Statistics

Post by GORDON »

The other day I saw a lottery official claim, "You're statistically more likely to be struck by a meteor, than to win this lottery."

Assuming his numbers are correct, doesn't this make "statistics" look like a stupid science? No one is ever struck by a meteor, but someone always wins the lottery, almost every week.
"Be bold, and mighty forces will come to your aid."
thibodeaux
Posts: 8055
Joined: Thu May 20, 2004 7:32 pm

Statistics

Post by thibodeaux »

Well you have to remember that something like 80% of statistics are just made up.

But, we looked at the data...
Thousands of years of historical records show people are likely struck by meteorites surprisingly often.
...
Based purely on statistics, researchers estimate that a space rock should strike a human roughly once every nine years. And with those odds, you’d expect people to get killed by meteorites fairly often.

“I do strongly suspect that stats on ‘death by asteroid’ have been severely undercounted through human history,” NASA Planetary Defense Officer Lindley Johnson told Astronomy via email. “It’s only been in the last half century or so that we’ve even realized that such a thing could happen.”
GORDON
Site Admin
Posts: 54400
Joined: Sun Jun 06, 2004 10:43 pm
Location: DTManistan
Contact:

Statistics

Post by GORDON »

.............. which all sounds like bullshit, right?
"Be bold, and mighty forces will come to your aid."
TheCatt
Site Admin
Posts: 53729
Joined: Thu May 20, 2004 11:15 pm
Location: Cary, NC

Statistics

Post by TheCatt »

GORDON wrote: Tue Nov 08, 2022 4:10 pm No one is ever struck by a meteor, but someone always wins the lottery, almost every week.
You are comparing very different things.

Odds of being struck by a meteor: They're talking lifetime odds. Odds of winning lottery, they're talking a single ticket/instance. That being said, still seems like they are off, but they seem to be about the same order of magnitude.
It's not me, it's someone else.
GORDON
Site Admin
Posts: 54400
Joined: Sun Jun 06, 2004 10:43 pm
Location: DTManistan
Contact:

Statistics

Post by GORDON »

In my lifetime, a thousand-ish people will win millions in the lottery. Probably no one will be hit by a meteor in spite of nasa's "seems like it should happen more often, right? anyway we aren't changing our numbers" numbers.

They say you're more likely to be struck by a meteor than win the lottery, and claim to have the numbers to back that, but it simply isn't true.
"Be bold, and mighty forces will come to your aid."
thibodeaux
Posts: 8055
Joined: Thu May 20, 2004 7:32 pm

Statistics

Post by thibodeaux »

I would be curious to see the calculations here. For the lottery, the probability of picking the correct set of numbers is a straightforward combinatorics problem (I don't know it offhand but it involves factorials so the denominator gets REAL BIG).

For the meteor, I assume it's something like this: you collect the cause of death for N deaths (where N is very large). There are m deaths from meteorites, so the probability of dying of meteor is approximated as m/N. The challenges here are collecting accurate data as well as the fact that as Bayesians we chortle at these silly frequentist games.

But I think, as Catt says, these are two quite different types of events and I'm not sure how useful it is to compare their probabilities. For one thing, you'd only talking about the conditional probability of winning a lottery (conditioned on buying a ticket; your probability of winning without a ticket drops WAY off).

There is of course the practical solution of outsourcing this to an insurance company or a bookie. Find somebody who'll take a bet or write you a policy for both events, see which one costs you more. That's the more likely event.
thibodeaux
Posts: 8055
Joined: Thu May 20, 2004 7:32 pm

Statistics

Post by thibodeaux »

GORDON wrote: Tue Nov 08, 2022 4:43 pm .............. which all sounds like bullshit, right?
Yeah pretty much
GORDON
Site Admin
Posts: 54400
Joined: Sun Jun 06, 2004 10:43 pm
Location: DTManistan
Contact:

Statistics

Post by GORDON »

TheCatt wrote: Tue Nov 08, 2022 5:13 pm
GORDON wrote: Tue Nov 08, 2022 4:10 pm No one is ever struck by a meteor, but someone always wins the lottery, almost every week.
You are comparing very different things.
I didn't make the comparison, I'm just claiming their conclusion is bullshit.
"Be bold, and mighty forces will come to your aid."
Post Reply