52 coaches online • Server time: 16:36
Forum Chat
Log in
Recent Forum Topics goto Post Secret League Old Wo...goto Post Creating a custom to...goto Post ramchop takes on the...
SearchSearch 
Post new topic   Reply to topic
View previous topic Log in to check your private messages View next topic
Matthueycamo



Joined: May 16, 2014

Post   Posted: May 01, 2016 - 20:59 Reply with quote Back to top

dode74 wrote:
Matthueycamo - Don't know if you saw, but I posted a link to another poll with very different results.

As for questioning a poll's results, they are suspect if they have not been corrected for errors, or the errors at least identified. This isn't the statistics of dice we're looking at, it's peoples' opinions and they vary wildly with context. At best what we have here is a straw poll with no controls for all the sources of error (in both directions), which means extrapolating it to assume it is representative of the BB community as a whole is simply incorrect. The best you can say of the poll taken here is "of the 598 FUMBBLers who voted 57% thought CPOMB was a problem in the game". Similarly, the best you can say of the poll I linked is "of the 245 TFFers who voted 65% thought CPOMB was fine". You'll note that I'm NOT using a poll to back up my opinion regarding CPOMB at all, and I am questioning those who do (in either direction).


A 2011 poll when the changes were pretty new? So a poll that is 5 years old is more accurate than one that is about a year and a half? Sure, might as well say the 2010 UK election polls indicate how people are going to vote in 2020......

_________________
Image

DLE College 7s
ArrestedDevelopment



Joined: Sep 14, 2015

Post   Posted: May 01, 2016 - 21:04 Reply with quote Back to top

Matthueycamo wrote:

A 2011 poll when the changes were pretty new? So a poll that is 5 years old is more accurate than one that is about a year and a half? Sure, might as well say the 2010 UK election polls indicate how people are going to vote in 2020......


He's saying they're both equally irrelevant.

_________________
Image
Matthueycamo



Joined: May 16, 2014

Post   Posted: May 01, 2016 - 21:07 Reply with quote Back to top

ArrestedDevelopment wrote:
The most basic problem with citing a poll for anything to do with blood bowl is that you have no idea of the total active numbers of coaches (even on fumbbl alone) to begin with. No population, no confidence interval, no representative sample size.

That's without even thinking about the other controls.


You don't need to know the size of the user base to get an accurate poll. Political polls can be very accurate with a sample size of 1000 when the UK electorate is something like 40 million. Are there 40 million FUMBBL users? Are there even 40 million people who play Blood Bowl world wide? So sample size is not an issue at least for the for the FUMBBL poll. The TFF one though, up to 5 options with 240 odd votes? That is a problem as that could represent as many as 240 people which would be getting towards what I would consider a decent sample size or as little as 48 which is not. That is the problem there, knowing the number how many voted in total and how many voted multiple times.

Controls are a different issue as well as making sure the question is neutral.

_________________
Image

DLE College 7s
Matthueycamo



Joined: May 16, 2014

Post   Posted: May 01, 2016 - 21:18 Reply with quote Back to top

Thinking about it we do know the number of coaches on the database. Something like 70000 from the statistics page.

_________________
Image

DLE College 7s
ArrestedDevelopment



Joined: Sep 14, 2015

Post   Posted: May 01, 2016 - 21:46 Reply with quote Back to top

And yet only 4054 have accessed the site and 2174 played a game in the last 90 days. What numbers do we use?

And yes, you do need to know the numbers. Otherwise you can't extrapolate information (which is what electorate polls do) to produce data with any degree of confidence.

_________________
Image
Matthueycamo



Joined: May 16, 2014

Post   Posted: May 01, 2016 - 21:55 Reply with quote Back to top

Yes you can because we know the numbers. You could even do it for all three figures given if you wanted. But with any of those numbers 600 is more than enough of a sample.

It's how you control it that is the thing that is questionable with such a poll system.

_________________
Image

DLE College 7s
Wreckage



Joined: Aug 15, 2004

Post   Posted: May 01, 2016 - 21:58 Reply with quote Back to top

I think to demand a poll on a gaming site to live up to scientific standards for the purposes of a forum debate is pretty dull. You just take what you have in terms of informations and should accept what conclusions you can draw from it.
I can see perhaps how it would be a valid argument to question a polls conclusions when you have other indications that make its conclusions seem silly or the like. But any other indication we look at indicates the same thing.

If you feel like the community of Fumbbl has gone off track of what is important, nothing debated here will be fruitful, Dode. I think perhaps if you are unhappy with the conclusions you should go out into the BB world and conduct some research that satisfies your standards and present it to us when it is concluded and then perhaps there is a better basis for a dialogue.
dode74



Joined: Aug 14, 2009

Post   Posted: May 01, 2016 - 22:14 Reply with quote Back to top

Matthueycamo wrote:
A 2011 poll when the changes were pretty new? So a poll that is 5 years old is more accurate than one that is about a year and a half? Sure, might as well say the 2010 UK election polls indicate how people are going to vote in 2020......
Neither one is relevant, which is my point. The sources of error haven't even been identified, let alone accounted for.

The 600 number would give a 95CI margin of error of 3.7-4% depending on whether you use 4000 or 70000. Thing is, if you want it to be representative of what Blood Bowl players think it assumes the polling is randomly taken from all Blood Bowl players, and we know for a fact it is not. The two sources of error I mentioned (FUMBBL players and the name of the thread being leading) are just some of the things you'd need to control for if you want to claim some sort of representative sample.

Wreckage
Quote:
I think to demand a poll on a gaming site to live up to scientific standards for the purposes of a forum debate is pretty dull.
I think to claim a poll is representative you have to be able to ensure it is regardless of the reasons for using it. The polls (either one of them) cannot be claimed to be representative of anything more than the people who voted in them, and to claim they are representative of the BB world in general is either ignorant or dishonest. To claim one is while the other is not is crass stupidity.
Quote:
I can see perhaps how it would be a valid argument to question a polls conclusions when you have other indications that make its conclusions seem silly or the like. But any other indication we look at indicates the same thing.
Oh really? What "other indication" are you looking at? Because I presented an "other indication" which indicates the opposite. Not that I think it is valid to draw the same conclusions from that one, either.
Corroborating evidence (if it exists) does not suddenly make a poorly conducted poll accurate. Bad methods can lead to the right conclusions as well as the wrong ones. The old saw about stopped watches being right twice a day springs to mind.
Quote:
I think perhaps if you are unhappy with the conclusions you should go out into the BB world and conduct some research that satisfies your standards and present it to us when it is concluded and then perhaps there is a better basis for a dialogue.
I think perhaps if people want to present evidence to support their conclusions then they should make sure it does so rather than simply asserting it does. I'm not the one claiming the polls support the conclusions, and the burden of proof lies with those who are.
DarthPhysicist



Joined: Jun 14, 2015

Post   Posted: May 01, 2016 - 22:17 Reply with quote Back to top

TBF, its not Dode's responsibility to refute a suggested change to the rules. Its incumbent upon those who think it broken to prove that it needs fixing. Even if a majority of people thought claw should be changed, that doesn't make it worth doing.

_________________
Using derivative humor since 2005.
Image
Wreckage



Joined: Aug 15, 2004

Post   Posted: May 01, 2016 - 22:27 Reply with quote Back to top

DarthPhysicist wrote:
TBF, its not Dode's responsibility to refute a suggested change to the rules. Its incumbent upon those who think it broken to prove that it needs fixing. Even if a majority of people thought claw should be changed, that doesn't make it worth doing.


Well it's much easier to proof that CPOMB is broken than to proof that the majority thinks its broken. Yet both is true. After 5 years of these debates it just feels like a waste of time to pick up all the tables again and calculate it through for somebody who probably doesn't understand what you are doing anyways. Every single time somebody new comes along.

It is not true however that Claw is broken. Razz
Nextflux



Joined: Jan 22, 2008

Post   Posted: May 01, 2016 - 22:45 Reply with quote Back to top

I suppose its better to say people on fummbl lean towards against CPOMB instead of the whole world? The poll is just to get a general Idea, it indicates nothing other that people on fummbl lean towards a side, I thought that was representative enough.

The poll isnt that important eather, the important thing was how could it be changed IF it is changed and IF something was broken, and that discussion is merely a tiny part of the intention of this thread and not up for question right now.
dode74



Joined: Aug 14, 2009

Post   Posted: May 01, 2016 - 23:09 Reply with quote Back to top

Wreckage wrote:
Well it's much easier to proof that CPOMB is broken than to proof that the majority thinks its broken. Yet both is true.
How do you plan to prove it is broken? What metrics do you plan to use and why are you using them?
Quote:
After 5 years of these debates it just feels like a waste of time to pick up all the tables again and calculate it through for somebody who probably doesn't understand what you are doing anyways.
I've been along for the whole ride. It's neither been proved to be broken nor has it been shown that there is a majority which thinks it is. Some people do, sure, but it's entirely subjective.
Netflux wrote:
I suppose its better to say people on fummbl lean towards against CPOMB instead of the whole world? The poll is just to get a general Idea, it indicates nothing other that people on fummbl lean towards a side, I thought that was representative enough.
That controls for one bias, but what about others which are yet to be identified. The only thing the poll tells you is that of the 598 who voted 57% think it is a problem. Drawing further conclusions from a poll would require something far more rigorous.
Quote:
I suppose its better to say people on fummbl lean towards against CPOMB instead of the whole world? The poll is just to get a general Idea, it indicates nothing other that people on fummbl lean towards a side, I thought that was representative enough.
Really? Because the relevant CPOMB question was number 3 which is itself a leading question ("commonly accepted" indeed!). Questioning the premise of leading questions reveals the logical error in them.
When did you stop beating your wife, anyway? Wink
plasmoid



Joined: Nov 03, 2009

Post   Posted: May 01, 2016 - 23:44 Reply with quote Back to top

Hey Guys,
this was actually an interesting thread about something else.
Perhaps you could take your discussion here:
https://fumbbl.com/index.php?name=PNphpBB2&file=viewtopic&t=25152

After all, that thread says:
"Creating a new thread pertaining to this topic will get it locked down. Multiple lockings down for the same coach will get them in trouble with a possible forum ban.
Keep your stuff here please!"

Cheers
Martin
Wreckage



Joined: Aug 15, 2004

Post   Posted: May 01, 2016 - 23:53 Reply with quote Back to top

Sorry plasmoid, I blame dode but also us for feeding him with responses in spite of knowing better.
I just wrote a lengthy post about potential criteria to proof the brokenness of CPOMB like he requested but deleted it now and you are of course correct.
Although there hasn't been a new form submitted in a while. Smile Guess we have to wait for that.
dode74



Joined: Aug 14, 2009

Post   Posted: May 02, 2016 - 00:21 Reply with quote Back to top

Wreckage wrote:
Sorry plasmoid, I blame dode but also us for feeding him with responses in spite of knowing better.
Oh please. This turned to CPOMB well before I got involved. Threads wander and this one did.
Quote:
I just wrote a lengthy post about potential criteria to proof the brokenness of CPOMB like he requested but deleted it now
Of course you did Wink

That said, I'll not continue discussion of it here.
Display posts from previous:     
 Jump to:   
All times are GMT + 1 Hour
Post new topic   Reply to topic
View previous topic Log in to check your private messages View next topic