statistics: that's not how they work.

17 posts

Flag Post

i see a lot of people saying “x is probably true”
or “there is a n% chance of x happening”
or similar. x refering to some specific thing, and n being a number between 0 and 100.

you must have a random sample from some population for such a statement to make sense.
sometimes it’s implied, but other times it’s not, and is a result of either ignorance or intentional deception.

what “x is probably true” really means:
“p(y) is true more than 50% of the time when y is randomly selected from P”
where p(y) is a predicate (statement with a variable that determines its truth value)
and P is a population that y is selected from.

this means that statements like “god probably exists” or “god probably does not exist” do not make any sense!
there is no predicate, and no population, so a probability does not work.
the attempt at a predicate would be “god exists” or “goed does not exist”, but those are statements with a static truth value, because there is no variable component; they evaluate to either 0 or 1 (even if we don’t know which one, that does not make it a probability).

a way to rephrase it so that it might work:
“in a randomly selected universe y, the probability that (no) god exists in y is > 0.5”
this one makes sense as a probability. (and it is not the same as the original version.)

 
Flag Post

I’m so glad I found what to discuss in this thread.

 
Flag Post

it’s more of a public service announcement.

you can discuss related things if you want, but the main goal is to make people less clueless about logic and statistics.

 
Flag Post

god exists in y is > 0.5”

How can you measure probability in inches?

 
Flag Post
Originally posted by Souperman22:

god exists in y is > 0.5”

How can you measure probability in inches?

Can you read?

“in a randomly selected universe y, the probability that (no) god exists in y is > 0.5”

It is a quote.

 
Flag Post

I said “God probably doesn’t exist” so I feel adressed.

I wasn’t talking about exact mathematics though, probably just means ‘very likely’ or ‘without much doubt’ where I come from. Therefore you could say it’s guessing the probability of something roughly between 50.000…01 and 99.999… percent.
Language is what people make of it. So when people say “We’ll probably be just fine” it’s not because they did some kind of population test, it’s an (often educated) guess.

 
Flag Post

God probably doesn’t exist

You can’t make statements like that, Searth. I agree with this;

this means that statements like “god probably exists” or “god probably does not exist” do not make any sense!

Not a single human knows if he exists. There is no “likely” or “without much doubt”, because we honestly don’t know. There’s no likelihood we can address here. We can only come up to the point where we name reasons why we would think it’s probable or not probable.

what “x is probably true” really means:
“p(y) is true more than 50% of the time when y is randomly selected from P”
where p(y) is a predicate (statement with a variable that determines its truth value)
and P is a population that y is selected from.

I’m not sure how much of statistics you have studied, but this isn’t entirely true. Depending on the statistic you’re aiming on. You set a certain alpha, you set a certain population average, take a sample, then find the probability of the sample average to be found assuming the population average is true. Alpha is usually set to 5% (meaning there’s a 5% chance to find such a sample average assuming the population average is true). If the probability you find is higher, there is not enough evidence to say the assumed average is wrong. If it is lower, there is enough evidence to withdraw the population average (there’s still another matter with a certain beta, but I’ll leave that part out).

a way to rephrase it so that it might work:
“in a randomly selected universe y, the probability that (no) god exists in y is > 0.5”
this one makes sense as a probability. (and it is not the same as the original version.)

There’s a theory multiple universes exist, perhaps even in the same space we live, but I don’t think we will ever be able to find a probability a God exists anywhere.

But I feel this topic should be more general and not just about God. You are saying people shouldn’t just make statements? Well, I obviously agree, I more often than not see people make statements they can’t back up or just say to insult a certain group, without any backup at all. I always address them, and you should too. We can’t force them not to, only show that it doesn’t make sense.

 
Flag Post

@darkrunner:
you’re talking about confidence and tolerence intervals? (if you’re not, then i didn’t get that far in statistics yet)
if you are:
there still has to be a sample, or something to sample.

@searth:
my point is not that your guess is inaccurate, or that you don’t know the probability, it’s that the probability does not exist.

 
Flag Post

Mm. I wasn’t aware that colloquial definitions and uses of words had to be in compliance with their formal mathematical counterparts.

 
Flag Post

darkrunner

Wut.

you’re talking about confidence and tolerence intervals?

That’s one thing. I studied the one I meant some time ago, so I might have some things wrong, I believe it is called the Z-test.

there still has to be a sample, or something to sample.

There should be, I do not disagree on that.

 
Flag Post
Originally posted by Yreval:

Mm. I wasn’t aware that colloquial definitions and uses of words had to be in compliance with their formal mathematical counterparts.

as far as i can tell, the colloquial term is completely meaningless.
what does it do other than attempt to make a guess sound like more than a guess?

in some contexts, “probably” isn’t so bad (like, where it isn’t much of a stretch to make an actual statement about probability).
“high chance”, or made up percentages, on the other hand, should always mean the mathematical term (unless the misuse is funny).

Originally posted by Darkruler2005:

darkrunner

Wut.

you’re talking about confidence and tolerence intervals?

That’s one thing. I studied the one I meant some time ago, so I might have some things wrong, I believe it is called the Z-test.

there still has to be a sample, or something to sample.

There should be, I do not disagree on that.

the keys are right next to eachother, i swear. :V

and yeah. i intended this thread to be general, but thought i should include an example.

 
Flag Post

as far as i can tell, the colloquial term is completely meaningless.
what does it do other than attempt to make a guess sound like more than a guess?

Colloquially, probably is used to indicate inference or to imply the application of common sense to an unknown outcome.

i.e.
“I just drove here and the roads were empty, so my friend probably won’t be caught up in traffic.”
“That chair looks flimsy. It will probably break if you sit in it.”

In neither statement is there any real statistical evidence. The arguments logically follow, but by saying “probably” the speaker emphasizes a sense of uncertainty.

 
Flag Post
Originally posted by Yreval:

as far as i can tell, the colloquial term is completely meaningless.

what does it do other than attempt to make a guess sound like more than a guess?

Colloquially, probably is used to indicate inference or to imply the application of common sense to an unknown outcome.

i.e.

“I just drove here and the roads were empty, so my friend probably won’t be caught up in traffic.”

“That chair looks flimsy. It will probably break if you sit in it.”

In neither statement is there any real statistical evidence. The arguments logically follow, but by saying “probably” the speaker emphasizes a sense of uncertainty.

those uses aren’t completely seperated from the methemetical meaning, though.

there’s still an implied sample. like, of all the previous times you drove on empty roads (that you recall), not many people got caught in traffic.
or chairs that look flimsy break often enough to warrent caution.

these statistics aren’t necessarily accurate, but it’s roughly the same meaning.

the issue i have with misuse of the term is in things like:
previously stated god example
probability of some universal morality
probability that science will discover “truth”
probability of contact with intelligent aliens (until/unless we find at least one)
etc. (basically applying probability to philosophy or pseudo science or something like that)

 
Flag Post

There is an implied sample in those other cases, too. In your God example, the implied population is “universes superficially similar to ours”.

 
Flag Post

67% of statistics are made up on the spot, 58% of people know that.

 
Flag Post

There is a 98% chance this thread does not have the proper statistics to make it’s case.

 
Flag Post

I am normally all for limiting the abuse of technical terminology, I just don’t see the problem here. While you can claim that there is no rationally-determined value for the probability of God’s existence (and I do so claim), that does not make it a useless concept. You can still say useful things about it:

1. The genetic sequencing of a genetic drift in a newly discovered sub-species supports current evolutionary theory. The probability of an interventionist god goes down slightly.

2. While analyzing the human genome, a sequence decodes as the ‘On-chromosome help file.’ Televangelist’s rejoice.

Yes, statistics talks about populations, but these populations or outcomes are rarely tangible things. You don’t need alternate dimensions with their own gods, you just need the set of possible configurations of this universe that you can’t disprove.

Originally posted by notallama:

"in a randomly selected universe y, the probability that (no) god exists in y is > 0.5”

this one makes sense as a probability. (and it is not the same as the original version.)

I would say they mean exactly the same thing. Even statisticians make use of implicit statements when they are speaking natural languages. That is why we have this artificial language called mathematics: so we can be precise when necessary. When you start assuming there is a one-to-one relationship between the two…well, that way lies madness.

this means that statements like “god probably exists” or “god probably does not exist” do not make any sense!

there is no predicate, and no population, so a probability does not work.

the attempt at a predicate would be “god exists” or “goed does not exist”, but those are statements with a static truth value, because there is no variable component; they evaluate to either 0 or 1 (even if we don’t know which one, that does not make it a probability).

Actually…yes…yes it does. At the end of the day, there has to be a final, unchangeable result to which the probability is referring. Statistics are not limited to talking about groups of events, or to things we know about, or even to events that haven’t happened yet. It is the mathematical art of determining what the information we have says about the information we don’t have.

Is it probable that the loaf of bread in my pantry contains more than the legal limit of rat droppings? Is this question meaningless, as the predicate value was set at the time of baking? I would argue no, the answer may well determine whether I will eat toast in the morning. You, I assume, know very little about me or my pantry, but the FDA website would allow you to make an educated guess to the solution. I might give a different probability, because I have more information on the bread’s source.

To take a better-known example, pretend you are on a game show. The host explains that there is a new car behind one of three curtains. You pick Curtain 1. The host, who knows where the car is, opens Curtain 2 to show empty space. What is the probability that the car is behind your Curtain 3? My answer would be 2/3. Would you give the same answer? Or would you claim,“that does not make a probability”?

Even our values for the physical constants have a margin of error, even though the real values are supposedly…you know…constant. Although…some people say there is a small probability that they aren’t.