Curse

kooomar · Nov 07, 2007, 08:29 PM // 20:29

Quote:

Originally Posted by Martin Alvito

Stated rate: 63%
Actual rate: 64.2% (derived by: (977-350)/977)

See how easy these things are to test? If the lockpick algorithm "works", 95% of repeated 1000-ish openings should be within 3-4% of the stated rate.

Odds are, this is the majority of what you'll see when people post larger blocks of results.

This isn't by any means meant to put you down, nor get flamed by the people that know STAT better than me, I've got my math minor and have forgot alot of the stuff now. But, to calculate retain rate, is it really as simple as using an "error analysis" type division problem?

Again, I could be entirely wrong here, flames aren't needed.

But, it seems to me, that lets say you have 10 LPs with a 50% chance to retain, SO, people think quickly that you should get 15 items (yes, we're in a perfect statistics world right now, welcome to it.) But in fact, you shouldn't get 15 items should you?

The way I see it, you have your 10 LPs, and use them all, that leaves you with 10 items, sweeeet. Now half your picks broke, you have 5 picks left, so, you use them all, you know have 15 items (the originally thought amount) and 2.5 picks left, so you use those, you know have 17.5 items, and 1.25 picks left, etc etc. Which will approach 20 I believe? giving a mathematical retention rate of 50%

Hmmm, after looking at where I am going with this, I kinda prove that Martin is right, but like I said, this wasn't meant to say he was wrong. I am leaving this post though to show a little more detail on WHY he is right, so there aren't other people like me curious as to whether or not you can use a simple equation like that to do it.

Martin Alvito · Nov 07, 2007, 09:49 PM // 21:49

I happen to know off the top of my head roughly the approximate 95% confidence interval for a 1000-ish member sample. Although I do not do survey research myself, I am a graduate student in political science and I know many faculty and students that work with such surveys. As it turns out, the standard size for a national survey is 1000 or so due to the fact that the rate of return for adding additional members of the sample drops off quite precipitously at about that sample size. (The rate of return for adding another person to the sample decreases exponentially.)

With large populations (such as the GW population of lockpick uses or the population of the United States), it turns out that a sample size of 1000 gives you a 95% confidence interval of just over 3%.

If you don't believe me, type 'sample size confidence interval calculator' into Google and visit some of the first few links.

I'd post direct links, but site policies frown on this.

strcpy · Nov 07, 2007, 10:04 PM // 22:04

That is true - however those 1000 people have to be VERY carefully chosen.

Look at a presidential election poll set - you have 5 or more polls all at your confidence level yet as much as 10 or 15 points difference in them. That is only statistically viable assuming you actually got a sample that is representative of the population.

That sample size is used because it is "close enough" most of the time and the logistics for gathering more is hard. But it isn't unusual to have researchers or polls to be wrong enough that they cause problems with that small a sample. You generally get a VERY different idea on sample sizes when you cross from the mathematicians over to the social scientists.

Dante the Warlord · Nov 07, 2007, 11:16 PM // 23:16

Quote:

Originally Posted by Savio

Or you can look at the next lockpick. If he had broken 1 lockpick already, he has a 1/2 chance that the next one will break. If he had broken 2 lockpicks already, he has a 1/2 chance that the next one will break. If he had broken 10,000 lockpicks already, he has a 1/2 chance that the next one will break. Regardless of the number of lockpicks broken already, he still has a 1/2 chance to break the next one. His odds will not get better except for an increase in title rank.

Thats true, but really its not the way i personally look at it, yea of course you chances don't get better individually, but i view it as a whole, that your chances that you could have gotten it are smaller.. Basically i think that from the first lockpick to you next one should range out to be around 50%. Kinda like a parabola and using my musty stat skills kinda like a SD band(not even sure if im even saying it right) Where you have 95% of it average and everything outside very minute, almost unachievable. Heh maybe i just forgot those skills i learned freshmen year. But correct me since i am curious...

Quote:

Originally Posted by strcpy

You aren't vague, you are grossly incorrect. This is one of the absolute most common classic mistakes out there in statistics, any decent course on it should have drilled this into your head since day one.. What you are talking about is called Gambler's Fallacy

I had written out a fairly large post, but that link has the same "overview" I had and goes into more detail. I also rather suggest that most of the people in posting here click on the "Law of averages" link in that article - so far only a VERY small handful of people have any idea whatsoever what they are talking about. I'm reminded of the saying "A little knowledge is a dangerous thing".

Alright then, well i can proudly say that i am incorrect. But please don't call me a retard(what i interpreted from other messages that i have read). I don't like to brag, but im in a really great high school that specializes in mathematics, nearly the best in my state. Although I am not the best at my school, I don't want anyone to mock me of this error, so please i willing understand my incorrect answer. Although most of the other part of my post were kinda right.

Its just my viewing... thats all this isn't any of my current knowledge rather the knowledge from when i was a 1st grader. I haven't used statistics really since last year lol. So i would be happy if we could call an academic truce :-)... well not really since i lost but you get the point.

kooomar · Nov 08, 2007, 02:14 AM // 02:14

Quote:

Originally Posted by Dante the Warlord

Alright then, well i can proudly say that i am incorrect. But please don't call me a retard(what i interpreted from other messages that i have read). I don't like to brag, but im in a really great high school that specializes in mathematics, nearly the best in my state.

Nobody is calling you a retard, but most of the people including myself on here that know a little more about statistics, have graduated college/are close to graduating/have a math minor or major. There is the crust of statistics that you learn in high school, and then the pie filling that you learn in grad level stat classes, which is 99% theory.

strcpy · Nov 08, 2007, 02:29 AM // 02:29

Quote:

Originally Posted by Dante the Warlord

Alright then, well i can proudly say that i am incorrect. But please don't call me a retard(what i interpreted from other messages that i have read). I don't like to brag, but im in a really great high school that specializes in mathematics, nearly the best in my state. Although I am not the best at my school, I don't want anyone to mock me of this error, so please i willing understand my incorrect answer. Although most of the other part of my post were kinda right.

I didn't call you a retard - I said that is a common mistake that any decent course will drill into your head, if your course didn't do that I don't care if you took it at MIT it wasn't a decent course. Nor are you "kinda correct" - that is like saying 2+2=5 is kinda correct because it is only one off. It is wrong.

Nothing the matter with being wrong - all of us are all the time. We will all even be wrong on important things in our lives, let alone the chance of retaining lockpicks. More the issue is continuing to argue over something you don't really understand. It's best if when you are on shaky ground don't argue with people who obviously aren't and if you do end up being wrong either go your merry way or go "oops". Defending yourself from being wrong by saying "I'm brilliant and sorta correct" doesn't come off very well. That tends to be the cause of the less than polite "you are mistaken", no one minds an "oops, my bad".

Dante the Warlord · Nov 08, 2007, 02:14 PM // 14:14

Quote:

Originally Posted by strcpy

I didn't call you a retard - I said that is a common mistake that any decent course will drill into your head, if your course didn't do that I don't care if you took it at MIT it wasn't a decent course. Nor are you "kinda correct" - that is like saying 2+2=5 is kinda correct because it is only one off. It is wrong.

Nothing the matter with being wrong - all of us are all the time. We will all even be wrong on important things in our lives, let alone the chance of retaining lockpicks. More the issue is continuing to argue over something you don't really understand. It's best if when you are on shaky ground don't argue with people who obviously aren't and if you do end up being wrong either go your merry way or go "oops". Defending yourself from being wrong by saying "I'm brilliant and sorta correct" doesn't come off very well. That tends to be the cause of the less than polite "you are mistaken", no one minds an "oops, my bad".

But I DID admit i was wrong, I just happy to say i was at least outsmarted by a person 5 or more years older then me.... and its not like i didn't know i just didn't think it all the way. I also never referred to myself as brilliant, i merely tried to help someone else out, although defending it may have definatly pushed it. I learned statistics and data analysis back in freshmen year, so i figured i should have been able to hold my ground. Trouble is i forgot it a little, since i am now a junior. So Thank you for the correction and to the mods, I think this question has long been answered, please close this thread.

Ministry Of Peace · Nov 08, 2007, 06:26 PM // 18:26

level 3 treasure hunter here... I went through a REALLY bad streak where it seemed like I got only 20% retain rate or so... This went on for maybe 20-30 picks... then I retained 10 in a row. It's pretty much dumb luck, I'm sure if you retain more than usual, you don't really notice... It just seems like the retain rate is wacked because people have a general tendency to focus on bad things, and call it unfair.

wolfwing · Nov 08, 2007, 07:48 PM // 19:48

Okay yes it's quiet possible to open 1000 chests and have 1000 lockpicks break, but you got a better cahnce of winning the lottery at 51% chance of retaining them having it happen.

Now with my own record, I've had 396 break, 320 retained with maybe 15 at most in HM, though I retained maybe 2-4 of those 20. When I hit 50k lucky 5k unlucky they were both within maybe 10k of each other, now there is a cap closer to 20k.

And while I have had streaks where I might get 3-4 chests in a row where I retain a lockpick, they are always followed by runs of 5-8 where they break every time. And it's not bias or falacy, you can check it by simply seeing how big a gap there is between luck and unlucky. If the unlucky title is going up faster then the luck you know there is something wrong.

One thing that I've noticed and wonder if anyone else has that it seems that if I have any good luck runs it's late at night or early in the day. And if I open chests from about 11am EST to 1-2 PM est I have way below average. Not saying I always have good luck in the off hours but thats when I seem to for the most part get luck runs.

wolfwing · Nov 08, 2007, 07:55 PM // 19:55

heh actually I keep close track, because I buy in lots of 10 lockpicks, and I see how many chests I've opened by the end. On good luck runs I open maybe 17-19 chests with 10 lockpicks, though on average it's closer to 13-15, with like that last run where it was about 14 chests and 12 lockpicks wich is why I started to post, and that started with a run of 8 lockpicks broken in a row.

Faer · Nov 08, 2007, 09:20 PM // 21:20

Quote:

Originally Posted by wolfwing

How about you people that say it's random get a clue.

l2statistics plz

Seriously. These threads pop up all the time, and it's always a case of some people feeling that they are getting shafted. You aren't. That's just the way the cookie crumbles. Stop gambling if you don't like losing at it.

The usual.