Dismiss Shaella. Semi-arsed troll; at least D.E.A makes some effort. Shaella has proven to dismiss arguing with anyone who's able to speak their mind.
Dismiss Shaella. Semi-arsed troll; at least D.E.A makes some effort. Shaella has proven to dismiss arguing with anyone who's able to speak their mind.
I am sorry but 5-6 a day means that I would have to spend what ?Lets say 45min/game x 5 = 225min
And that is a low avrg for me...not accounting for que time and the bathroom/smoke break.This is like the 6th part of a day,..It is insane.
What I take from this is either most of dota community are :kids,slackers or insanely rich...
Do all of you really play 5-6 games per day;|?
@ Shaella: 30+ players is more than enough as a sample size.
Quick course:
We assume a factory is producing 1 000 000 needles (this can be any number since it has no influence on any calculations later on). A statistic in the past has shown that every 100 needles there is 1 broken. (1%)
They are testing on a sample size of 10.
a) None of these needles are broken: No conclusion can be drawn. It is likly that nothing have changed. (90.44% chance on happening by pure luck)
b) 1 of these needles are broken: something is suspicious (9.14% chance on happening by pure luck)
c) 2 or more of these needles are broken: something is definitly not right. (0.42% chance on happening by pure luck)
look for binomial distribution. (numbers used are: sample size (100); actual result (0,1,2); chance(1%))
If c) happens the company will have to review the manifactoring process and put up a new statistic(less than 5% criteria!)
But you are right on the part that this experiment cannot be trusted. See below.
----------------
@ OT: Original post sounds promising but now that I have read it through and did some math myself something doesn't seem right....
Is your sample biased against the mute system? I cannot imagine how 33 players out of 34 players were muted in your sample. Did they play normally? Did the spam the chat wheel? Are the players from the sample players who don't like the mute system?
It's hard to believe because I know many people including myself who got never muted. And in your statistic 90% and more players are already muted. This can't sound right.
According to original post the system works really faulty but if it's true it means there are over 40% muted guarenteed (it is the underlimit). And it is over 90% of playerbase has been muted once since implementation of mute system (which was released roughly 3 months ago)
This doesn't add up right. So the conditions you set up your experiment is definitly wrong.
---------------
binomial distribution calculation for Valve statistics: 1% of active playerbase being currently banned:
0 out of 34: 71.06%
1 out of 34: 24.4%
2 out of 34: 4.07%
3 out of 34: 0.44%
4 out of 34: 0.03%
5 out of 34: 0.0020790511828033253%
6 out of 34: 0.00010150249882373139%
7 out of 34: 4.1011110635851074×10-8
8 out of 34: 1.398106044404014×10-9
...
Experts would question the 1% when they have a sample size of 34 and 2 players are muted. (Less than 5% criteria)
I am aware that I cannot do this calculation since there are many smurfs in the active playerbase and some players play more other players play less. But for argumentation sake I am listing this anyway.
That's why I will not draw any conclusions here. (And it does not support the result from original post! Chance on 33 out of 34 players muted is a really small number. It is a number with over 50 decimal places, virtually 0.)
---------------
From your sample size we have 33 players out of 34 players muted once. In order to show any significance (5% criteria) you need a chance of 99.85% on being muted!
(Binominal of 34 from 34 chance: 99.85 equals 95%)
99.85% is upper limit.
From your sample size we have 33 players out of 34 players muted once. In order to show any significance (5% criteria) you need a chance of 12.6% on not being muted! (I am approaching from other direction.)
Binominal from sample size 34, result 1, chance 12.6% is ~5%.
This means the chance on being not muted - according to your result from sample - is less than 12.6%. 12.6% is the absolute upper limit.
This means the chance on being muted is greater than 87.4% (1-12.6%) and less than 99.85% according to the result you posted. That percentage is FAR TOO HIGH. And it proves that your statistic is biased.
~~~~
What I have done above is: I have questioned the validity of the conditions you have set. I made calculation on your result assuming your result is 100% unbiased it. The result from my calculation is that the chance on being muted is greater than 87.4% and less than 99.85% given your results can be trusted.
(Assuming your result is unbiased/right/resembling the community. I calculated the the chance's upper and lower limit of being banned according to your result.)
This is astronomical high and is a strong indication that the condition you set up your experiment is wrong. Do you believe that the chance of being banned for every player - no matter if he is abusing the communication or not - is between 87.4% and 99.85%?
[If I even take it a step further and assume you picked players from a biased group. Meaning people who don't like the mute system. It even PROVES that the mute system is working. But I won't draw any conclusion here since I don't know what you have actually done. I don't know if you really didn't talk. If you didn't chat wheel spam, If you intentionally messed up your team. If you... the list goes on.]
--------------
If someone is not understanding what I wrote:
I buy spagetti noodles
Because of unknown reason 95% of my spaghetti noodles are broken.
I am making some calculation as in original post and it turns out 33 spaghetti out of 34 spaghetti noodles are broken.
This does not necessarily mean that everytime you buy spaghetti noodles 95% of your spaghetti noodles are broken. I should have picked spaghettis from many different packages and then analyze.
Or simpler: Based on 33 noodles being broken out of 34 noodles from one package I conclude that every spaghetti package I buy has 87.4% - 99.85% of its content broken. Does this sound right?
------------
I am glad you have posted your results with such unreal numbers so it was easy for me to prove that the experiment is strongly biased and doesn't resemble the active player base at all.
With this I cannot trust any experiment you post later on. Your later experiment may show results which is a bit more reasonable. But just because they look reasonable doesn't mean they are unbiased.
It is far easier to prove a experiment biased if the numbers are straight up unreal.
The next method of proving experiments wrong is by analyzing the conditions the experiment was made on. And this is very hard to prove on your side and very hard on my side to verify what you posted is actually what happened.
Chances are your next experiment still has biasness in it even though if you corrected some numbers to make them look more reasonable.
TL;DR: This experiment cannot be trusted at all.
Only thing which can possibly concluded is that chat-wheel only users have a 87.4% - 99.85% chance on being muted. And when we analyze that conclusion we will see this this is also not true: Link to my second wall of text
Last edited by GermanViet; 06-23-2013 at 05:30 PM. Reason: typos, clarifitcation and addition
James "2GD" Harding's career as a Dota 2 host on Valve events
* 25.02.16 - † 26.02.16
Spoiler:
@GermanViet
I'm not going to write a post on why you could be wrong, because I'm just not that bored at this point. However, I will say this. If you take into consideration this particular point: "We have taken numbers of muted or players refusing to speak in our games from a period of 2 weeks. Every single one of us was using an account clean of communication bans or reports, All Pick and Single Draft modes, both Regular and Solo matchmaking on US West or East, varied our play times, and most importantly ONLY USED THE CHAT WHEEL FOR COMMUNICATION." Then you would be aware that this behaviour will potentially increase number of reports against subjects. So even though the numbers seem "unreal", could still potentially be true.
James "2GD" Harding's career as a Dota 2 host on Valve events
* 25.02.16 - † 26.02.16
Spoiler:
Just because they only used chatwheel doesn't mean they weren't spamming for nothing or ironically, just saying.
You can use chatwheel exclusively an still be an asshole.