This site makes extensive use of JavaScript.
Please enable JavaScript in your browser.
Live
PTR
10.2.7
PTR
10.2.6
Beta
Why Surge of Light?
Post Reply
Return to board index
Post by
OscarDivine
:)
More than once I've presented essentially that exact calculation in response to your claims regarding CoH and SoL. But it's ok -- probability theory really is foreign territory for most people. Which is why it's amazingly easy to
lie with statistics
...very few people understand how to think about random or uncertain events.
Ha! you have no idea...
Uh I'm a doctor. My life is telling people rough and relative statistics just to make them feel better.
Post by
Paolo
Lol I used to teach doctors how to think like a Bayesian.
"I'm sorry Mr. Applebaum, the test came back positive."
"You mean I have cancer!"
If you can answer that one correctly, I'll be your patient for life. Hint: it's not
that
hard, but almost no doctor has a clue. ("Yes, I'd say it's pretty likely that you do in fact have cancer.") Another hint: there's some essential missing information that any good doctor would be able to fill in. Given that information, you can tell the patient
exactly
what the probability is that they have cancer.
Heck, if you can even name the essential missing pieces you'll get half credit.
Go.
(And this is less off-topic than you might think. WoW is a game of probabilities, obviously, and knowing how to navigate them correctly is a Good Thing.)
Post by
179389
This post was from a user who has deleted their account.
Post by
OscarDivine
Lol I used to teach doctors how to think like a Bayesian.
"I'm sorry Mr. Applebaum, the test came back positive."
"You mean I have cancer!"
If you can answer that one correctly, I'll be your patient for life. Hint: it's not
that
hard, but almost no doctor has a clue. ("Yes, I'd say it's pretty likely that you do in fact have cancer.") Another hint: there's some essential missing information that any good doctor would be able to fill in. Given that information, you can tell the patient
exactly
what the probability is that they have cancer.
Heck, if you can even name the essential missing pieces you'll get half credit.
Go.
(And this is less off-topic than you might think. WoW is a game of probabilities, obviously, and knowing how to navigate them correctly is a Good Thing.)
uh dude, did I mention that I'm an EYE doctor? I don't do jack when it comes to cancer. If it's cancer, they're having their eyeball plucked or cryo'd. So here's a few things that I'd need to understand about said test: What was it testing for? Does the test have a higher degree of SENSITIVITY or SPECIFICITY? Now if the test was a lab looking at the cells and determining that there's a genetic defect in the apoptotic mechanism of the cells leading to the continuous growth of the cells, then I might be able to take a swing at this one. Otherwise, I don't know the answer. The most important thing I need to know about this test is it's degree of sensitivity over specificity.
Now here's an interesting statistical analysis regarding glaucoma, straight out of the OHTS, which was a several year prospective study regarding the outcome of high eye pressure treatment with eventual degradation to glaucoma.
In this study, the analysis boils down to these relative figures: Given 100 patients with eye pressure over 26, 10 will develop glaucoma if left untreated. Now if you were to take the same 100 patients and treat them all with eye pressure lowering medications, instead of 10 developing glaucoma, you now find that 5 do.
So in order to discuss starting your patient on medication, you can say it to your patient one of two ways:
1) If I give you these eye drops, it will cut your risk of developing glaucoma by 50%! (which is of course, untrue).
OR...
2) If I give you these eye drops, your risk of developing glaucoma is decreased by 5%.
We all claim that we want to hear the truth, but honestly, that's not what you want, nor what many physicians give. In fact, most doctors will cite the first of the two.
edited for grammar.
Post by
179389
This post was from a user who has deleted their account.
Post by
179389
This post was from a user who has deleted their account.
Post by
309579
This post was from a user who has deleted their account.
Post by
Paolo
@Hidden: no need to power-quote everything.
Gonna pull one outta me arse:Is everyone around here freakin' British? Did I get redirected to wowhead.uk for reindoctrination, and just never notice? Sheesh. Ima start talkin' street, just to, yaknow, re-BALance the situation, yo.
@Ignayshus: I wrote this before I read your second post. You're getting very warm. Unfortunately you didn't account for false positives. Read on.
The most important thing I need to know about this test is it's degree of sensitivity over specificity.Almost entirely correct. You also need to know the
prevalence
. (EDIT: again, most of this was written before you guys caught this omission.)
Probability is a pure science, untethered to any particular domain such as medicine or Surge of Light procs. This is why it doesn't matter whether what we're looking at is cancer, glaucoma, or spell crit rating. Medical people have created their own language; I'll translate Oscar's (and my) post:
Prevalence: the overall level of in the relevant population.
Sensitivity: for a diagnostic test, what's the probability that someone already known to have will correctly test positive?
Specificity: what's the probability that someone known NOT to have that condition will correctly test negative?
(Sens & spec are related to, but not exactly, the false positive rate and the false negative rate.)
Instead of going too far down this windy path, I'll just point y'all to some offsite resources for Bayesian calculations. Understanding probability theory is good for your health, I promise. I'd write pages about it if that didn't ensure an off-topic thread-lock. Which I believe we're dangerously close to anyway.
*
Conditional probabilities
(math-heavy, but skip the symbols. read the fallacy section, it's relevant!)
Calculating conditional probabilities
(with calculator) (medical context)
the simple cure to cancer is abolish diseaseUnfortunately, real-world "cures" are rarely 100% effective, and they are rarely side-effect free. Jus' sayin'.
Post by
OscarDivine
Ha at least I don't fail at medical statistics. ;-) completely... I didn't realize that you wanted me to talk about prevalence as well, but regardless, excellent discussion.
Post by
Paolo
If you're going to reassure Mr. Applebaum -- which means let him know the correct probability that he has, well, whatever you just tested him for -- then you need to know all 3 pieces of information, in addition to his test result.
My probability professor in university gave us a test once.
You're arrive at a bus stop, and 20 minutes later the bus arrives. What is the bus schedule?
Or, in today's parlance:
A priest casts Penance and crits on two out of three ticks. What is her crit rate?
That, ladies & germs, is the stuff that doctoral theses are made of.
Go.
No don't, I'm just kidding.
Post by
OscarDivine
No don't, I'm just kidding.
oh thank God I was about to say that if you knew the answer to those questions you are far beyond me.
Post by
91278
This post was from a user who has deleted their account.
Post by
179389
This post was from a user who has deleted their account.
Post by
Paolo
What is
her
crit rate?
The other 46% was
his
Awesome Factor.Higher math: questionable.
Birds & bees: Please report to your mum for remedial education.
Post by
91278
This post was from a user who has deleted their account.
Post by
Paolo
I'm bilingual. I hate the sound of that word, but used it to put you at ease. Too many English friends...how many 'murkins you know that use the word "winge"?
Post by
91278
This post was from a user who has deleted their account.
Post by
105534
This post was from a user who has deleted their account.
Post by
OscarDivine
Finally I understand! I should spec surge of light because Mr Applebaum will feel better. Awesome.
Wait, who is Mr Applebaum?
He's the poor guy who will die of cancer unless you can crit. You have once shot...
GO!
Post by
318206
This post was from a user who has deleted their account.
Post Reply
You are not logged in. Please
log in
to post a reply or
register
if you don't already have an account.