Jump to content


So what is a large enough sample?


  • Please log in to reply
62 replies to this topic

qpranger #1 Posted 22 September 2017 - 08:25 PM

    Major General

  • Player
  • 31557 battles
  • 5,061
  • [T-D-U] T-D-U
  • Member since:
    12-25-2013

As in: sufficient number of battles to determine a statistical anomaly?

The WG trolls on here like to sing the refrain of "this sample is too small".

So how big is big enough? Or will any sample be too small if it rubs the wrong way? :)



Not_DangerUXB_Thats4sure #2 Posted 22 September 2017 - 08:32 PM

    Major

  • Player
  • 9516 battles
  • 2,554
  • Member since:
    03-07-2013

100 according to Aging Jedi...

 

 

...But if he came up with 500?....it would be 1000 for the forum trolls/WG net forum police/Aspiring barristers etc..


Edited by Danger_UXB_, 22 September 2017 - 08:32 PM.


Jigabachi #3 Posted 22 September 2017 - 08:39 PM

    Field Marshal

  • Player
  • 17858 battles
  • 18,583
  • [T-D-U] T-D-U
  • Member since:
    08-12-2011

Entirely depends on the anomaly you want to find.

 

View Postqpranger, on 22 September 2017 - 08:25 PM, said:

The WG trolls on here like to sing the refrain of "this sample is too small".

The sample size that people show most often is... well... zero. That is definitely too small.

Also: "WG trolls"... that trolling path is pretty pathetic. Even for you.


Edited by Jigabachi, 22 September 2017 - 08:42 PM.


Homer_J #4 Posted 22 September 2017 - 08:40 PM

    Field Marshal

  • Beta Tester
  • 27652 battles
  • 29,000
  • [WJDE] WJDE
  • Member since:
    09-03-2010
All of them.

CmdRatScabies #5 Posted 22 September 2017 - 08:42 PM

    Colonel

  • Player
  • 35936 battles
  • 3,832
  • [-MM] -MM
  • Member since:
    10-12-2015

View Postqpranger, on 22 September 2017 - 08:25 PM, said:

As in: sufficient number of battles to determine a statistical anomaly?

The WG trolls on here like to sing the refrain of "this sample is too small".

So how big is big enough? Or will any sample be too small if it rubs the wrong way? :)

 

It depends on the population you're sampling and what you are trying to estimate.

I suspect it takes less than 100 Tier 8 battles to conclude that MM is statistically fucked.  However, you've need a lot more to prove that your team mates are statistically worse than the other team.



Not_DangerUXB_Thats4sure #6 Posted 22 September 2017 - 08:43 PM

    Major

  • Player
  • 9516 battles
  • 2,554
  • Member since:
    03-07-2013

View PostHomer_J, on 22 September 2017 - 07:40 PM, said:

All of them.

 

I can see you now have 'Dinger' blood in you.....you used to be so informative?:(

 

 

''So what is a large enough sample?''



GreatSuprendo #7 Posted 22 September 2017 - 08:47 PM

    Private

  • Player
  • 21138 battles
  • 27
  • Member since:
    09-13-2014

42 games in my Caernarvon so far.  Of those I have been top tier once (in a 5/10 T8/T7 split).  I have yet to see a Tier 6 tank.

 

Didn't watch the video all the way through to see his numbers, but MM is better in a T8 TD.  I see tier 6 on a semi-regular basis there (still way less than 1/3, but I don't care in a TD as at least I have enough pen to do something useful).

 

Won't be completing the Brit heavy grind as I play this game for fun, and being raped by higher tiers every single game stops being fun after a while.  WG really need to check their dictionary to find out what a game is supposed to be:

 

game
[ɡeɪm]
  1. a form of competitive activity or sport played according to rules.
     
  2. an activity that one engages in for amusement


Homer_J #8 Posted 22 September 2017 - 08:49 PM

    Field Marshal

  • Beta Tester
  • 27652 battles
  • 29,000
  • [WJDE] WJDE
  • Member since:
    09-03-2010

View PostDanger_UXB_, on 22 September 2017 - 08:43 PM, said:

 

I can see you now have 'Dinger' blood in you.....you used to be so informative?:(

 

 

''So what is a large enough sample?''

 

All of them.

 

I was being serious.  Get statistics of every single battle by every player and then nobody can suggest you don't have a large enough sample.

 

Plus it's a Q-Pranger thread anyway, nobody takes them seriously.



Spurtung #9 Posted 22 September 2017 - 08:52 PM

    Major General

  • Player
  • 61598 battles
  • 5,900
  • [GW-UP] GW-UP
  • Member since:
    07-05-2013


qpranger #10 Posted 22 September 2017 - 09:01 PM

    Major General

  • Player
  • 31557 battles
  • 5,061
  • [T-D-U] T-D-U
  • Member since:
    12-25-2013

See a lot of squirming on here already, not many straight answers,

except from Aging Jedi and he has not even posted on this thread :)



Not_DangerUXB_Thats4sure #11 Posted 22 September 2017 - 09:03 PM

    Major

  • Player
  • 9516 battles
  • 2,554
  • Member since:
    03-07-2013

View PostHomer_J, on 22 September 2017 - 07:49 PM, said:

 

All of them.

 

I was being serious.  Get statistics of every single battle by every player and then nobody can suggest you don't have a large enough sample.

 

Plus it's a Q-Pranger thread anyway, nobody takes them seriously.

 

But what the 'Minority' suggest and what the 'Majority' suggest as per ;Statistical 'Anomily boundaries'....wouldnt it sway in the way of the majoity???..

 

..After all sometimes these discussions are run a bit like a courtroom so wouldnt this help?....Correct me if im wrong but i thought a 'Majority' thought was better than a minority??(Unless they are 'Experts' and not

relying of 'The feels'?..

 

 

..All i see is the same old same old defending MM (In the relevant topics they been doing it since 2012 or earlier....go see for proof) but hell...Even the CC,s are admitting it??....

 

 

MM is Baulked and needs fixing ASAP...so why these guys keep defending MM??...i dont understand?



geoff99 #12 Posted 22 September 2017 - 09:07 PM

    Corporal

  • Player
  • 37299 battles
  • 173
  • [SPIKE] SPIKE
  • Member since:
    09-20-2012

I could have a go at working it out, via a Chi-squared hypothesis test, but it will take me a while and make my head hurt, so if anyone cleverer than me could do it first that would be great .

 

Actually, wotlabs forums used to be full of very clever people who spoke stats fluently. Might be worth a post there, QP.



Rosetta_and_Philae #13 Posted 22 September 2017 - 09:07 PM

    Captain

  • Player
  • 44881 battles
  • 2,146
  • [FOXYR] FOXYR
  • Member since:
    06-14-2011
As many as you want.  Conclusions are drawn on testing an object numerous times.  What I can't understand is this belief in 100's if not 1000's of battles are required to draw some form of conclusion. 

Homer_J #14 Posted 22 September 2017 - 09:08 PM

    Field Marshal

  • Beta Tester
  • 27652 battles
  • 29,000
  • [WJDE] WJDE
  • Member since:
    09-03-2010

View PostDanger_UXB_, on 22 September 2017 - 09:03 PM, said:

 

...i dont understand?

 

I suggest you have had too much or not enough to drink.

 

Dr Nutter will be along shortly to prescribe a pound of sprouts.



Cobra6 #15 Posted 22 September 2017 - 09:09 PM

    Field Marshal

  • Beta Tester
  • 16332 battles
  • 15,550
  • [RGT] RGT
  • Member since:
    09-17-2010

In this game, 100 is a large enough sample for most things.

 

Cobra 6



Not_DangerUXB_Thats4sure #16 Posted 22 September 2017 - 09:16 PM

    Major

  • Player
  • 9516 battles
  • 2,554
  • Member since:
    03-07-2013

View PostHomer_J, on 22 September 2017 - 08:08 PM, said:

 

I suggest you have had too much or not enough to drink.

 

Dr Nutter will be along shortly to prescribe a pound of sprouts.

 

I suggest you are trying to avoid a 'Majority of the players' question here...If you are going to frequent the forum as you do at least be informative?? #Datdingerblood??

 

It seems that 100 games is enough regarding MM 'Fluctuations'....do you agree?...or should it be more (HIC!!)



CmdRatScabies #17 Posted 22 September 2017 - 09:21 PM

    Colonel

  • Player
  • 35936 battles
  • 3,832
  • [-MM] -MM
  • Member since:
    10-12-2015

View PostRosetta_and_Philae, on 22 September 2017 - 09:07 PM, said:

What I can't understand is this belief in 100's if not 1000's of battles are required to draw some form of conclusion.

 

Well that depends what conclusion you are trying to draw.

Not_DangerUXB_Thats4sure #18 Posted 22 September 2017 - 09:22 PM

    Major

  • Player
  • 9516 battles
  • 2,554
  • Member since:
    03-07-2013

View Postgeoff99, on 22 September 2017 - 08:07 PM, said:

I could have a go at working it out, via a Chi-squared hypothesis test, but it will take me a while and make my head hurt

 

Its divided by 2 (See vid)...Simples:)

Homer_J #19 Posted 22 September 2017 - 09:24 PM

    Field Marshal

  • Beta Tester
  • 27652 battles
  • 29,000
  • [WJDE] WJDE
  • Member since:
    09-03-2010

View PostDanger_UXB_, on 22 September 2017 - 09:16 PM, said:

 

It seems that 100 games is enough regarding MM 'Fluctuations'

 

I'm not sure that Q-Pranger intended this to be another mm thread because I'm sure he wouldn't want the mods to junk it.

 

As for "fluctuations" as far as I can see 3-5-7 is broken working as intended and you can blame everyone who complained about mm in the past, which wasn't me.



qpranger #20 Posted 22 September 2017 - 09:28 PM

    Major General

  • Player
  • 31557 battles
  • 5,061
  • [T-D-U] T-D-U
  • Member since:
    12-25-2013

View PostHomer_J, on 22 September 2017 - 10:24 PM, said:

 

I'm not sure that Q-Pranger intended this to be another mm thread...

 

True, my personal emphasis in this particular instance is on win rate anomalies.




1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users