http://en.wikipedia.org/wiki/Sampling_(statistics)#Simple_random_sampling

"However, SRS can be vulnerable to sampling error because the randomness of the selection may result in a sample that doesn't reflect the makeup of the population. For instance, a simple random sample of ten people from a given country will on average produce five men and five women, but any given trial is likely to overrepresent one sex and underrepresent the other."

This is the main issue with XFire polling.

Yay 100 more pages of discussing the exact same subject the previous 100 discussed to death!

Originally posted by Sovrath

Originally posted by Burntvet

Originally posted by quentin405

Originally posted by RefMinor

Originally posted by quentin405

X-fire is just about worthless..  Using it to judge anything is... funny to say the least.

 

I wonder if you can justify that using actual reasoning and statistical theory.

 Yeah heres my statistical theory..  people are using data collected by 4,398 players on xfire to make "facts" about ToR.

 

So, how many millions, minus 4.3k = Insane amount of people that agree with me xfire is worthless :)

And please spare me your reply.. Everyone knows those 4.3k people only use Xfire so they can compete with their friends on who is more hardcore by keeping track of how many 1000s of hours they spend playing 1 game.

 

So my theory is, xfire is just about worthless... Using its TINY fraction of players to assume anything is... hilarious.

 

As has been mentioned, you obviously have no idea how statistical modeling works.

I suggest you look up how the Neilsen TV rating systems work. Do you think that they have wired X million households with TV monitoring boxes? No. It is on the order of a couple thousand. And yet, they use those couple thousand boxes to estimate the "share" and thus viewer ship a TV show gets.

Same thing with all the political polling going on these days: it is a sampling of X registered voters or those likely to vote.

Statistical modeling has proven itself over and over to be a useful tool and accurate within a margin of error.

And Xfire, like it or not, falls within the scale of a statistical modeling tool. Maybe not the best, escpecially in how the subjects are selected, but random enough in the general catagory of computer game / MMO players to be considered valid.

If statistical modeling didn't work, or wasn't useful, people/companies would not spend the huge amounts of money they do using it.

Just because you either don't like or don't understand how it works, does not make it invalid.

 

 

Except the Nielsen group took pains to find an actual "sampling' of different housholds from different demographics. That's the key point there.

If I go into several large retirement communities, wire up a portion of the residents, can I safely say that I can now use that data to determine what the rest of the country is doing?

Now, if I wired up a few people in a rural town, upper/lower class, same with a city, I might get better results.

Yeah, I was on that when I mentioned that the X fire sample might not be as carefully selected as it could be, because it is a totally voluntary 3rd party program used by MMO players.

That said, even if it can not pull out a particular data point, i.e. TOR has X subs, it is very good at hitting the trend, because the same sample (well-selected or not) is tracked week in and week out.

So if Xfire is tracking a strong downward on TOR play hours, that is very likely to be accurate. However, that does not directly correlate to a proportionate loss of subs, because one can theoretically be subbed to TOR and not play.

For the love of god why does this post still exist.  123 pages of people saying the same two things over and over.

Originally posted by sirphobos

For the love of god why does this post still exist.  123 pages of people saying the same two things over and over.

i dont know about other servers but mine seems to be getting busier not quieter. Imp fleet had 2 instances running last night and still had over 200 people on when i logged at 0100. this was ToFN server.

Originally posted by sirphobos

For the love of god why does this post still exist.  123 pages of people saying the same two things over and over.

Smiple, because the people that actually use Xfire honestly believe they are always right, on every game, no matter what facts are presented to them showing differently. You just can't make some people see anything but what they are willing to see. No matter how many pages people put in this thread, no matter how many factual counter arguments people prove here, the Xfire users will NEVER believe any of it. And if anyone doubts this, please feel free to bounce back and read the previous 122 pages in this thread, or any of the 100s of threads about Xfire numbers on these forums over the past 3 years.

Originally posted by Yamota

http://en.wikipedia.org/wiki/Sampling_(statistics)#Simple_random_sampling

"However, SRS can be vulnerable to sampling error because the randomness of the selection may result in a sample that doesn't reflect the makeup of the population. For instance, a simple random sample of ten people from a given country will on average produce five men and five women, but any given trial is likely to overrepresent one sex and underrepresent the other."

This is the main issue with XFire polling.

Technically XFire isn't polling - there is no design behind it whatsoever

Happily no one here (I hope) is saying that XFire is without error. Sadly some are saying XFire is worthless without thinking however?

Some pages back I looked at several things that we actually know about SWTOR and compared them to what XFire showed. Not the "my server is full/empty so SWTOR is thriving/dying" useless comments but real data; real observations. I had no preconceived idea one way or the other. Bottomline: yes XFire is not perfect but it is far from worthless.

An example. Knowing that SWTOR had sold 2M copies at the end of the year if you were to use XFire to predict the numberof subs at the start of Feb your answer would have been 1.5M. So XFire underrepresented the population by c.14%.

So there you have it!  XFire is either totally worthless ....... or not perfect but not bad enough to simply dismiss.

This thread should be about xFire, period. I opened a thread with this idea but the moderator closed it, so we are stuck with this never ending discussion which would end if people understood that we should ONLY discuss xFire and not anything beyond that. 

That's what i do at least.

retract )

Originally posted by gervaise1

Originally posted by Yamota

http://en.wikipedia.org/wiki/Sampling_(statistics)#Simple_random_sampling

"However, SRS can be vulnerable to sampling error because the randomness of the selection may result in a sample that doesn't reflect the makeup of the population. For instance, a simple random sample of ten people from a given country will on average produce five men and five women, but any given trial is likely to overrepresent one sex and underrepresent the other."

This is the main issue with XFire polling.

Technically XFire isn't polling - there is no design behind it whatsoever

Happily no one here (I hope) is saying that XFire is without error. Sadly some are saying XFire is worthless without thinking however?

Some pages back I looked at several things that we actually know about SWTOR and compared them to what XFire showed. Not the "my server is full/empty so SWTOR is thriving/dying" useless comments but real data; real observations. I had no preconceived idea one way or the other. Bottomline: yes XFire is not perfect but it is far from worthless.

An example. Knowing that SWTOR had sold 2M copies at the end of the year if you were to use XFire to predict the numberof subs at the start of Feb your answer would have been 1.5M. So XFire underrepresented the population by c.14%.

So there you have it!  XFire is either totally worthless ....... or not perfect but not bad enough to simply dismiss.

 

There is no design behind random polling either, it is random. XFire is a random selection of SW:TOR gamers which has X-Fire installed so I would say it is like random polling.

Now there could be some bias in this selection of people, but I have not seen any facts supporting this statement. Like XFire is used mainly by FPS gamers, young people, etc. The only known bias I know of is that it is mainly a tool for the western audience so it is very inaccurate at gauging the eastern gaming trends.

Another absolute low with almost a 64% drop, but it's forming a new shape with a smoother decline. Will it safely land at 4000 players?

Is it because of Asian players?

Is it the US holidays someone mentioned? (sorry lost the name).

Does server use reflect the same tendency? I can't reach MMO junkies statistics anymore :/

Originally posted by Metentso

Another absolute low with almost a 64% drop, but it's forming a new shape with a smoother decline. Will it safely land at 4000 players?

Is it because of Asian players?

Is it the US holidays someone mentioned? (sorry lost the name).

 

 

 

Does server use reflect the same tendency? I can't reach MMO junkies statistics anymore :/

 

I've asked several times, but are you comparing this graph with other top mmos during the same period? Are you comparing this graph with those same mmo's graph from the first 60 days of launch.

Looking at one line and not accounting for its statistical context - well, you aren't able to do any actual analysis.

Originally posted by Zorgo

Originally posted by Metentso

Another absolute low with almost a 64% drop, but it's forming a new shape with a smoother decline. Will it safely land at 4000 players?

Is it because of Asian players?

Is it the US holidays someone mentioned? (sorry lost the name).

 

Does server use reflect the same tendency? I can't reach MMO junkies statistics anymore :/

 

I've asked several times, but are you comparing this graph with other top mmos during the same period? Are you comparing this graph with those same mmo's graph from the first 60 days of launch.

Looking at one line and not accounting for its statistical context - well, you aren't able to do any actual analysis.

No sir, I'm not, my apologies.

Originally posted by Metentso

Originally posted by Zorgo

Originally posted by Metentso

Another absolute low with almost a 64% drop, but it's forming a new shape with a smoother decline. Will it safely land at 4000 players?

Is it because of Asian players?

Is it the US holidays someone mentioned? (sorry lost the name).

 

Does server use reflect the same tendency? I can't reach MMO junkies statistics anymore :/

 

I've asked several times, but are you comparing this graph with other top mmos during the same period? Are you comparing this graph with those same mmo's graph from the first 60 days of launch.

Looking at one line and not accounting for its statistical context - well, you aren't able to do any actual analysis.

No sir, I'm not, my apologies.

) Well played.

Originally posted by Zorgo

Originally posted by Metentso

Another absolute low with almost a 64% drop, but it's forming a new shape with a smoother decline. Will it safely land at 4000 players?

Is it because of Asian players?

Is it the US holidays someone mentioned? (sorry lost the name).

 

 

 

Does server use reflect the same tendency? I can't reach MMO junkies statistics anymore :/

 

I've asked several times, but are you comparing this graph with other top mmos during the same period? Are you comparing this graph with those same mmo's graph from the first 60 days of launch.

Looking at one line and not accounting for its statistical context - well, you aren't able to do any actual analysis.

Oh no you can't do that here.  "WoW is old, Rift is six months older" blah blah blah.  They always use other game's age as an excuse for declines.  Meanwhile COD4 was released in 2007 and is the 2nd most played game on xfire.

Originally posted by Yamota

Originally posted by gervaise1

Originally posted by Yamota

http://en.wikipedia.org/wiki/Sampling_(statistics)#Simple_random_sampling

"However, SRS can be vulnerable to sampling error because the randomness of the selection may result in a sample that doesn't reflect the makeup of the population. For instance, a simple random sample of ten people from a given country will on average produce five men and five women, but any given trial is likely to overrepresent one sex and underrepresent the other."

This is the main issue with XFire polling.

Technically XFire isn't polling - there is no design behind it whatsoever

Happily no one here (I hope) is saying that XFire is without error. Sadly some are saying XFire is worthless without thinking however?

Some pages back I looked at several things that we actually know about SWTOR and compared them to what XFire showed. Not the "my server is full/empty so SWTOR is thriving/dying" useless comments but real data; real observations. I had no preconceived idea one way or the other. Bottomline: yes XFire is not perfect but it is far from worthless.

An example. Knowing that SWTOR had sold 2M copies at the end of the year if you were to use XFire to predict the numberof subs at the start of Feb your answer would have been 1.5M. So XFire underrepresented the population by c.14%.

So there you have it!  XFire is either totally worthless ....... or not perfect but not bad enough to simply dismiss.

 

There is no design behind random polling either, it is random. XFire is a random selection of SW:TOR gamers which has X-Fire installed so I would say it is like random polling.

Now there could be some bias in this selection of people, but I have not seen any facts supporting this statement. Like XFire is used mainly by FPS gamers, young people, etc. The only known bias I know of is that it is mainly a tool for the western audience so it is very inaccurate at gauging the eastern gaming trends.

Xfire is not a random selection of SW:TOR gamers.

It is not random. Its a subset of SW:ToR player which where included into this group by the criteria of choosing xfire for any other reason than polling, but for the benefits and interest into xfire as a communication software. No the argument s.o. else brought up before that "xfire is not choosing them so its random" is foolish, (if anything it supports that it is not random not the opposite), because its obvious the user are the ones choosing xfire and their choice or even demand for a software like xfire is obviously and evidently influenced by personal preferences and tendencies correlating woth same for games. Otherwise xfire would not have significantly different results from real surveys. (And no, I am not talking about annecdotes to "private" surveys which never happened or torstatus which is not a random survey either, before they are brought up yet again for no good reason)

By your logic we could as well draw preelection polls from greek orthodox church members for U.S. presidency elections. "Hey the greek orthodox church is a random selection of Mc Cain voters who happen to be part of the greet orthodox church. I would say its like random polling! I also wont accept any reason or evidence about possible or proven bias. Look they didnt like McCain! Look McCain wasnt elected. SEE ! It was ACCURATE! Now on to the next election with our PROVEN methods!!!"

The only reason why you "have not seen any facts" pointing at the obvious bias even when multiple have been already posted even in this very same topic, even two pages before your response highlighted in flashy colors, is because you most apparently proactively look away or just deny them because they make almost 50 pages of plain ignorance insisting on xfires "accuracy, proven accuarcy etc due to being a random sample" look just plain dumb and contradict what you attempt to convince others to believe.

Randomized surveys like those of NIELSEN have times and times again not just slighty but grossly contradicted or deviated from xfire. Its almost hillarious when people here insist to compare xfire to "random polling methods" and pretend their acclaimed yet never shown "knowledge about statistics", somehow would affirm it  just by reapting the same claims over and over again, when the very same reputable polling surveys and public accessible suvey results, do not even support the claim about xfires "accuracy", not just that but infact back again the obvious "bias of xfire user statistics vs real stats/trends".

But I am shure despite the fact we should already be over with this matter about 20-40 pages before 10 posts next someone will repeat the same claim about "lack of evidence for bias" again and insist xfire to be "like" a random polls again and compare it to preelection surveys again or pretend xfire was "proven accurate in the past" again, just repeating the same ignorant and debunked claims again and again...

I thought I smelled dead horse when I logged on this site. Not much left of it but you guys are still wailing away on it. I'm sure its corpse can't wait until GW2 or TERA  launches its equines. Then it will get some rest.

I wonder how stupidly long and drawn out this thread would be if XFire was showing a growth of hours?

Originally posted by Sukiyaki

Originally posted by Yamota

Originally posted by gervaise1

Originally posted by Yamota

http://en.wikipedia.org/wiki/Sampling_(statistics)#Simple_random_sampling

"However, SRS can be vulnerable to sampling error because the randomness of the selection may result in a sample that doesn't reflect the makeup of the population. For instance, a simple random sample of ten people from a given country will on average produce five men and five women, but any given trial is likely to overrepresent one sex and underrepresent the other."

This is the main issue with XFire polling.

Technically XFire isn't polling - there is no design behind it whatsoever

Happily no one here (I hope) is saying that XFire is without error. Sadly some are saying XFire is worthless without thinking however?

Some pages back I looked at several things that we actually know about SWTOR and compared them to what XFire showed. Not the "my server is full/empty so SWTOR is thriving/dying" useless comments but real data; real observations. I had no preconceived idea one way or the other. Bottomline: yes XFire is not perfect but it is far from worthless.

An example. Knowing that SWTOR had sold 2M copies at the end of the year if you were to use XFire to predict the numberof subs at the start of Feb your answer would have been 1.5M. So XFire underrepresented the population by c.14%.

So there you have it!  XFire is either totally worthless ....... or not perfect but not bad enough to simply dismiss.

 

There is no design behind random polling either, it is random. XFire is a random selection of SW:TOR gamers which has X-Fire installed so I would say it is like random polling.

Now there could be some bias in this selection of people, but I have not seen any facts supporting this statement. Like XFire is used mainly by FPS gamers, young people, etc. The only known bias I know of is that it is mainly a tool for the western audience so it is very inaccurate at gauging the eastern gaming trends.

By your logic we could as well draw preelection polls from greek orthodox church members for U.S. presidency elections. "Hey the greek orthodox church is a random selection of Mc Cain voters who happen to be part of the greet orthodox church. I would say its like random polling! I also wont accept any reason or evidence about possible or proven bias. Look they didnt like McCain! Look McCain wasnt elected. SEE ! It was ACCURATE! Now on to the next election with our PROVEN methods!!!"

That really makes no sense. Political votes are heavily influenced by religious and cultural mores, so ofc drawing a sample out of a religious or ethic subgroup will falsify the poll. However in this case xfire is measuring ONLY how many people that are already playing swtor while using xfire, decide to keep playing it.

It doesn't matter if they are some subgroup aslong as the subgroup doesn't differ wildly from the general population(which your greek orthodox church example clearly does), they obviously cared enough about the game to buy it(so they where not generally opposed to the genre), install it and play it for some period of time. They expected to enjoy it, and they ended up not enjoying it and playing it less and less. Thats what xfire is showing.

I mean seriously, what kind of "subgroup" is it supposed to be thats showing a "false" decline of swtor but a correct "stable" pop for WoW and other MMOs anyway? Is xfire a melting pot of SW haters? Or maybe people are drawn to xfire that have a rare medical disease that makes them get tinnitus from lightsaber soundeffects? I just can't think of a "kind" of people that are ok(not much delta) with WoW, EvE, Aion, EQ etc yet supposed to show bias against swtor. I mean its easy to draw up a religious subgroup that will be rather unlikely to support a gay rights activist democrat proposing same sex marriage as a bad sample example ... not so easy for a (rather big) bunch of computer guys who merely picked the same messanger.

I mean i never heard talk about "those AOL messanger guys", or "you know those folks that use Yahoo chat ...", are people seriously thinking that there are common traits shared between people that influence their choice of messanger AND will impact their enjoyment of certain activities? Are maybe Google mail people more likely to like soccer while yahoo mail people prefer baseball? Thats what your trying to say?

If a MMO on xfire is showing a delta that other MMOs do not ... i for one don't blame the sample.

Originally posted by bartoni33

I thought I smelled dead horse when I logged on this site. Not much left of it but you guys are still wailing away on it. I'm sure its corpse can't wait until GW2 or TERA  launches its equines. Then it will get some rest.

I wonder how stupidly long and drawn out this thread would be if XFire was showing a growth of hours?

The saddest part of this topic is that the guy who bumps it on regular basis never even went past the beta as he admitted that his PC is only good enough to run WOW. I am not making this up.... /facepalm