Geography and statistical probability of ENWorld members

[Agback]

You don't have a big enough sample to draw any inferences.

And how big a sample would he need to test the hypothesis that New York and LA are under-represented? at, say, the 90% confidence level.

Go on, calculate it. And if you don't know how, admit that you don't know whether your statement was true or not.

Regards,


Agback

 

[Agback]

Re: Re: Geography and statistical probability of ENWorld members

EN World has some 12,500 registered nicks (that's at best 0.4% of the gamer population - in acutality far less, considering international users, and duplicate nicks). And only a small number of those registered nicks are active posters.

Ergo, the active posters cannot be expected to comprise a statistically relevant sample of the gamer population.

The sample is not statistically relevant because it is self-selected and biased towards people who have free time and internet connections. But it is certainly large enough to be statistically relevant.


A sample does not have to be a large proportion of the population to support statistically significant inferences. Quality control sampling depends on that.

I think a few more people ought to take 200-level statistics courses before they start laying down the law about statistics.

Regards,


Agback

 

[der_kluge]

Crothian is from Ohio, another midwest state. Well, at least central timezone (I think?!).

Urbannen, hey! So you do actually de-lurk, eh? Still working on that campaign. It's coming together quite nicely. I'll start DM'ing for my group in the fall.

Farganger - my bad, I was thinking that Orange County, CA was where LA was. I'm guessing not. In any case, it is the county that contains LA. I suppose that is LA county? I've never been to LA (just near it), so I don't know.

And D20 publishers -
Mystic Eye Games = Austin, TX
Thunderhead Games (before merging) - Tampa, FL
Troll Lord Games - Little Rock, AR

Those are ones that I know...

 

[Urbannen]

I think the RPGs-orginate-in-the-northern-Midwest theory is a good one.

You've got: 1. Cold weather. & 2. A distinct Northern European cultural presence.

 

[The Sigil]

Re: Re: Geography and statistical probability of ENWorld members

Why is the Orange County, CA population relevant? Isn't most of metropolitan LA (and presumably its population) in Los Angeles county?

Just wondering if I missed some reorganization - many years ago I went through the Orange County Sheriff's Academy, and it was quite distinct from its Los Angeles County counterpart . . .

The Orange County CA population is not relevant if you want to talk strictly about Los Angeles population.

However, given that Los Angeles and Orange Counties are contiguous, and that both are fairly dense urban areas, the question may be extended from "LA proper" to "the LA area" (which presumably also includes adjacent and contiguous Riverside County).

I recall reading one time that geographically, LA is HUGE - IIRC, you could fit (by geographical size) New York, Chicago, Detroit, Philadelphia, Boston, Atlanta, and unless I am mistaken, the entire area of all of the top-15 most populous cities in the US (except LA itself, of course) into the area occupied by LA.

--The Sigil

 

[Farganger]

Farganger - my bad, I was thinking that Orange County, CA was where LA was. I'm guessing not. In any case, it is the county that contains LA. I suppose that is LA county? I've never been to LA (just near it), so I don't know.

Yes, it's in Los Angeles County, though as The Sigil remarked above, you could probably define "in" both Orange County and Riverside County as well, creating a first-rate megalopolis (albeit a sprawling and patchy one).

I will soon be reaquainting myself with LA, as my wife is about to start film school at USC.

 

[Olgar Shiverstone]

And how big a sample would he need to test the hypothesis that New York and LA are under-represented? at, say, the 90% confidence level.

You're right -- I'm off by an order of magnitude for EN World (and significantly further off for the population as a whole).

If the populations percentages are right, and my math is right(.0637 pop fraction for NY/LA), with a 90% confidence interval we need a sample of 63 to see if NY/LA are properly represented (about .5%, rather than 5%, of the EN World population).

Needless to say, my 10K ~ 300 mil number is way off.

Next time I'll engage my calculator before engaging my keyboard, just for you Agback.

Though I do stand my my original intent, which was that scanning the message boards hardly consists of a statistically valid sampling means (not to mention our skewed population).

 

[drothgery]

Actually since enworld is a global community you would have to comparer the NY and LA populations to the world population to arrive at a fair percentage for your query. Here is the math.

World Pop: 6,122,627,560
Ny & LA: 17,645,772

% of the world's population that lives in NY or LA: 0.288%


# of enworld member that makes up 0.288% of the community: 36

While ENWorld is global, it's heavily skewed, because it's a gamer-focused site written in English, focused on primarily on an American gaming system, and run by a Brit. I'd guess that all non-Japanese Asians (defined by where they live, not by ancestry), Africans, and South Americans combined are less than 5% of all gamers (and certainly less than 5% of gamers on ENWorld), though that would be a majority of the world's population.

 

[Umbran]

Re: Re: Re: Geography and statistical probability of ENWorld members

I think a few more people ought to take 200-level statistics courses before they start laying down the law about statistics.

Some folks should ask if others have taken such courses before they start making declarations. I'm a physics grad student, I know about statistics. As physicist, I approach it from a practical standpoint...

The active posters on this board probably represent somethting on the order of thousandths of a percent of the gaming community at large. Each one perhaps represents something like 15,000 gamers. One can get meaningful statistics polling one out of every 15,000 units, but unless one goes through a lot of rigamorale, one doesn't expect reliable statistics. At that point, one expects that the sampling will introduce error unless pains are taken to eliminate it.

Statisticians are very quick to point out that they can get reliable statistics with small sample sizes. They aren't so quick to admit that it's very difficult to do so, and that it's very easy to screw it up.

 

[2d6]

Crothian is from Ohio, another midwest state. Well, at least central timezone (I think?!).

Eastern Time Zone!

there are 6 or 7 people on here from Ohio at least.