Polls being conducted

[rust]

Different voters.

Thank you ! :lol:

 

[Golan2072]

Posted the poll on SFRPG as well.

 

[Gruffty the Hiver]

Posted the poll on SFRPG as well.

Ta! :mrgreen: :lol:

 

[EDG]

Is there any explanation why RPG Site and RPG.net have that much different, almost contradictory, results ?

RPGnet are a lot more mellow when it comes to Traveller than CotI. There have been instances of an idea being proposed on CotI that gets ferociously shot down by the natives, and the same idea being proposed on rpgnet that gets cheerfully encouraged by the natives there. The CotI crowd are also heavily into CT, and the rpgnet crowd aren't.


Not sure about rpgsite though - I think that was set up as a rival to rpgnet after some kind of split.

 

[rust]

Thank you very much, that was the kind of explanation I was hoping for.

 

[Allensh]

What "amuses" me is the fact that the party whose goading essentially brought the poll about has become strangely silent on the subject, choosing instead to spend his time ripping on 760 Patrons in another thread (and making blatantly insulting comments regarding those who did buy it), a book which by his own admission he hasn't even read.

Actually it kinda ticks me off...

Allen

 

[rust]

I found it interesting to compare the current CotI poll with the results
of two previous ones:

http://www.travellerrpg.com/CotI/Discuss/showthread.php?t=8073

http://www.travellerrpg.com/CotI/Discuss/showthread.php?t=13570

In these polls 44 % / 47 % named CT as the Traveller version used by
them, in the current poll (right now) just 36 % name CT as one of the
versions they use.

It seems that CT has lost quite some ground even among the CotI re-
gulars since MGT was published.

 

[far-trader]

...It seems that CT has lost quite some ground even among the CotI re-
gulars since MGT was published.

A couple flaws in your conclusion leap out at me.

First neither of those other polls included hybrid, add that factor to (probably mostly CT) the latest one.

Second Mongoose has added players, probably with very few from CT going over. That seems more the case than a big switch from CT to Mongoose. And some 10% hardly seems huge in any case.

(and hey, they're quick fan polls, not hard research )

 

[Allensh]

We might be somewhat on the way to disproving this statement then:

"I'll go on record, now, before the poll gets too far populated, that what I said earlier will be reflected in the poll. You'll see that CT gets about 40% of the vote and that MGT's votes will be cannibalized from the other editions of Traveller."

It was then stated:

"If I'm wrong, I won't hesitate to apologize for my prediction."

we'll see...

Allen

 

[Allensh]

Second Mongoose has added players, probably with very few from CT going over. That seems more the case than a big switch from CT to Mongoose. And some 10% hardly seems huge in any case.

(and hey, they're quick fan polls, not hard research )

If that's true..I'm ok with that. New players means Traveller survives as a commercial entity.

Anecdotal evidence from some of the other boards seems to indicate that people are switching from CT though. More of them actually seem to be using the two together. As more Mongoose books come out with newer versions of older stuff, some of those (by no means all) may adopt the new and drop the old. Speculation, of course; without actual hard research it would be tough to find that out.

In any event, a commercially successful thriving version of Traveller is a rising tide that lifts all boats, in my opinion.

Allen

 

[rust]

A couple flaws in your conclusion leap out at me.

While I agree that my conclusions are just conclusions, not research re-
sults in any way, I see things somewhat differently.

First, with hybrids in the count, CT should also be mentioned by those
who use it as a "secondary" version (e.g. main version MT, elements
from CT) for their hybrid system, not only by those who consider it
their favourite version (remember, the poll asked for the systems
the hybrids are based upon, too).

Second, I doubt that the membership of CotI has very much increased
since MGT was published, and that many players who started with MGT
flocked to CotI recently.

But, these are only assumptions. See it as an interesting thought experi-
ment, not a scientific research project.

 

[far-trader]

A couple flaws in your conclusion leap out at me.

While I agree that my conclusions are just conclusions, not research re-
sults in any way, I see things somewhat differently.

A fair point And certainly goes for me too. Vive la difference.

First, with hybrids in the count, CT should also be mentioned by those
who use it as a "secondary" version (e.g. main version MT, elements
from CT) for their hybrid system, not only by those who consider it
their favourite version (remember, the poll asked for the systems
the hybrids are based upon, too).

Yeah but some of the respondents (me ) only clicked one choice there (didn't see we could click more) and picked "hybrid" as the most honest answer (I'm playing CT in the game but the ref is doing a CT/Mongoose hybrid on his side)

Second, I doubt that the membership of CotI has very much increased
since MGT was published, and that many players who started with MGT
flocked to CotI recently.

Well the CotI membership roll has jumped of late (no hard numbers, just a rough recall of a couple hundred new members last year or so), and it feels like it's new Mongoose players for the most part.

But that's just one persons interpretation. I'd like to think it shows new players (not just new to Mongoose but new to Traveller) and only as a fraction of those actually playing (managing to find their way to CotI).

 

[Supplement Four]

What "amuses" me is the fact that the party whose goading essentially brought the poll about has become strangely silent on the subject, choosing instead to spend his time ripping on 760 Patrons in another thread (and making blatantly insulting comments regarding those who did buy it), a book which by his own admission he hasn't even read.

Actually it kinda ticks me off...

"Strangely" silent?

You're talking about me. (And, I don't care if I tick you off, btw.)

Before the polls were set, I strongly suggested to allow a vote to only pick one Traveller edition of choice...and to only have the official Traveller editions listed.

That would give us a more clear view of what Traveller players prefer. Instead we've got the "homebrew" option in there that muddies the view.

Even with that, though, the poll results are supporting my original statement...that CT represents about 40% of the Traveller playing public, and MGT players are either new players to the game or are cannibalilized from mostly other, non-CT, editions.

As the MGT numbers climb, we're not seeing the CT numbers decline. We're seeing the other editions of Traveller report in with scant numbers.

The polls are supporting my initial statement.

The polls don't "prove" anything--I don't think any sane person would say a poll like that would. But, they indeed are supporting my initial statement (the reason, you allude, that the polls were started in the first place).

 

[Supplement Four]

We might be somewhat on the way to disproving this statement then:

"I'll go on record, now, before the poll gets too far populated, that what I said earlier will be reflected in the poll. You'll see that CT gets about 40% of the vote and that MGT's votes will be cannibalized from the other editions of Traveller."

It was then stated:

"If I'm wrong, I won't hesitate to apologize for my prediction."

we'll see...

Allen

Actually, you're backwards in your reading of the polls. See above.

 

[captainjack23]

Just to be a stats geek, its hard to make much of the results for a very simple reason, even before poll differences, population bias, and etc.. Low numbers. Most stat tests are based on or similar to ratios -and at low numbers, ratios can swing wildly with a small change. thus a 33% to 66% result looks good with 300 votes, but not with 3 (or 30) . 30 is valid, but not stable. for the 30 vote pool, a change of 3 votes from one column to the other makes a big differnece in the percentages. Not so much for the 300 pool.
So, caveat statisticae.....

That was the summary. The below is a bit crunchier versionif one is interested.


Aramis correctly points out that the response number is valid, but not necc reliable. Unfortunately, the concept of a minimum sample is just that - the absolute minimum to have any validity (mathmatically) at all. It isn't a cut point -because more is better (up to a point, also, looking at diminishing gains) .

What that means is that even if valid, a few votes either way can make a big change in the numbers; with 60 respondants, a shift of three from one column to another makes a 10% difference; so, if you ran it again, and everyone tried to give the same answer, simple error would likely account for the differences... magnitude of difference is the only thing one can look at in a small sample that has any hope of being reliable - so, in the sample of 60, a 68% to 12% (40 votes to 5, or 8:1 )comparison is likely reflects a real trend - but(say) a 38% to 52% (~23 votes to ~31) looks decisive, but in fact can be turned into what looks like a dead tie by 3-4 more votes. Too, keep in mind that its easy to build up big magnitudes when the numbers are small, so thats at best an approximation.

Unfortunately, the validity level also is predicated on the parent population being big enough to have sampled being large enough to be assumed to have a reliable distribution of opinions - so an n = 30 sample of a 600 member population isn't necc a better sample than an n=30 sample of a ppulation of 6000. They both are valid, but the sample based on a smaller parent population is much more unstable than the other, becuase it's an unstable estimate of an unstable population (statistically, not clinically )


I can go into greater detail, but only if you really really want me to. And mind, no winding me up just for the fun of it. ....,
8)

 

[rust]

I can go into greater detail, but only if you really really want me to. And mind, no winding me up just for the fun of it. ....

Let me see if I get this right: The smaller the population, the higher the
percentage of votes you need for a reliable result, because even a small
number of voters makes up a high percentage of a small population, and
therefore even a minor error by a low number of votes could change the
end result considerably ?

And for my setting with that young human colony on a remote planet it
would mean that the smaller the settlement, the more colonists have to
take part in elections and referendums to get a reliable result ?

If this is what you mean, I never did see it that way ... :?

 

[Sturn]

Has anyone gone to the other forums which support non-MGT Traveller and posted the same poll links there? A.i. CotI, SteveJackson, etc.

If not, the voting could be skewed considerably if we are talking low numbers. If a given poll had 20 voters, then a link-thread from this one sent 40 new voters from the MGT forum, of course the result is going to be suddenly skewed.

 

[captainjack23]

I can go into greater detail, but only if you really really want me to. And mind, no winding me up just for the fun of it. ....

Let me see if I get this right: The smaller the population, the higher the
percentage of votes you need for a reliable result, because even a small
number of voters makes up a high percentage of a small population, and
therefore even a minor error by a low number of votes could change the
end result considerably ?

And for my setting with that young human colony on a remote planet it
would mean that the smaller the settlement, the more colonists have to
take part in elections and referendums to get a reliable result ?

If this is what you mean, I never did see it that way ... :?

Well, if we are talking about reliable and valid as statistical terms, validity means, does the math work. Reliable means are you comfortable that the results of the sampling represent the actual distribution in the population.

So, yes, you would have to have a larger sample for greater reliability, all else being equal. But ,keep in mind that reliability and confidence cutpoints are determined subjectively, and are also subject to diminishing returns as modeled by precision vs effort.

Any sample that is less than a totality of the population will have a less than 100% confidence -and, one must recall, even a 100% sample will have an error factor (pushed the wrong button, miscounts, etc); so it really is about where you want to set the values for gain in (precision *effort/unavoidable error factor) (sort of) Generally when the gain in precision is < error factor, it's time to stop: the noise is greater than the remaining signal.

The complication is that if the small population was selected for being homogenus on the topic of sampling, all bets are off. A sample of 1 would probably have huge reliability and validity. This whole area of interpretation is based on the idea that the sampling topics have a range of answers, and they are distributed semirandomly amongst the population.

Ask a samll colony of Dyzan cultists "do you believe in the great god Dyzan's divine nature, yes/no" and a sample of one will likely do fine. Ask them if they feel the Pirates should have made the trades they did, you'll need a bigger group ...(assuming 20C baseball stats isn't a key part of the Dyzanic cult -not always a good assumption... :? ) .


Does that help ?

The relationship to the polls being conducted is this: no matter what you decide, you are always also choosing the liklihood that your conclusion is wrong. With smaller samples and smaller populations you need a much bigger difference to have a reasonable chance -if you are at all worried about being right that is....most polls in the real world fall into one of two categories: opinion sampling, and opinion forming. Guess which is more likely to be used in propaganda and advertising ....?

 

[Sturn]

Let me see if I get this right: The smaller the population, the higher the
percentage of votes you need for a reliable result.

Been a while since my college stats class, but yes.

Looking at extremes might help:

Say you had a population of 100,000. 1%, or 1,000 would be large enough to get a good, varied sample.

Say you had a population of 4. Even 25%, which is only 1, would obviously not be large enough to get a good, varied sample.

 

[captainjack23]

Let me see if I get this right: The smaller the population, the higher the
percentage of votes you need for a reliable result.

Been a while since my college stats class, but yes.

Looking at extremes might help:

Say you had a population of 100,000. 1%, or 1,000 would be large enough to get a good, varied sample.

Say you had a population of 4. Even 25%, which is only 1, would obviously not be large enough to get a good, varied sample.

Yes, exactly ! Thank you for a clearer and even more concise example ! I'll probably steal it sometime. (no sarcasm intended !)