Hello Path of Exile players

"

Bars wrote:

The lab straw poll has 2400 voters. This is a more than excellent sample size for PoE's population and has at least a 97% confidence level. If you are talking about something, like mathematics, statistics and sociology, it's recommended to at least know some basics or you'll probably make a fool of yourself.

Article about sample sizes in sociological surveys (table 1 on page 3)

The article you linked says that its "approaches to determining sample size have assumed that a simple random sample is the sampling design". The straw poll you linked, however, uses a self-selected sample. That 97% level of confidence is only valid for a simple random sample.
Self-selection sampling is good if you want to get the views of people who are motivated to respond (eg: people care a lot about the topic). However, it is not good if you want to be confident that you have a representative sample.

The pros and cons of different sampling methods are all part of the basic statistics knowledge that is taught at secondary school, btw. As a wise person once said, "If you are talking about something, like mathematics, statistics and sociology, it's recommended to at least know some basics or you'll probably make a fool of yourself." ;)

I, for one, am hyped as fuck about atlas of worlds.

The lab I don't know about.. I hated it when I did it the first time, because dodging traps just didn't feel like poe. But then I did it again and I realised as long as you don't freak out and do anything stupid you can just run through all the traps with nothing to worry about, aside from a potential DC RIP. The lab brought with it crazy powercreep though, with no real addditional challenge to balance it out - Ascendancies give you more power than even the most OP uniques, with no cost to them.

But as much as people complain about the "clear speed meta".. isn't this what everyone secretly wanted anyway? Everyone wanted to be able to make a build that could just run through packs and have everything melt around them. Now everyone can. *shrug*

This is why I usually stay out of these discussions.

"

dudiobugtron wrote:

"

Bars wrote:

The lab straw poll has 2400 voters. This is a more than excellent sample size for PoE's population and has at least a 97% confidence level. If you are talking about something, like mathematics, statistics and sociology, it's recommended to at least know some basics or you'll probably make a fool of yourself.

Article about sample sizes in sociological surveys (table 1 on page 3)

The article you linked says that its "approaches to determining sample size have assumed that a simple random sample is the sampling design". The straw poll you linked, however, uses a self-selected sample. That 97% level of confidence is only valid for a simple random sample.
Self-selection sampling is good if you want to get the views of people who are motivated to respond (eg: people care a lot about the topic). However, it is not good if you want to be confident that you have a representative sample.

The pros and cons of different sampling methods are all part of the basic statistics knowledge that is taught at secondary school, btw. As a wise person once said, "If you are talking about something, like mathematics, statistics and sociology, it's recommended to at least know some basics or you'll probably make a fool of yourself." ;)

Not to mention selection bias, its mostly posting in places where the topic is already being discussed in a negative manor, thus the results are already guaranteed to be skewed. The poll itself is located in a negative thread of the subject which is bias in of itself, which automatically invalidates it.

Edit - But hey, i dont know what im talking about obviously.

/s

seriously thats why i just wanted to drop it because soo many things were wrong, especially the way he used the link he obviously just googled without understanding. I literally IRL faced palmed and massaged my temples after that.

"

dudiobugtron wrote:

"

Bars wrote:

The lab straw poll has 2400 voters. This is a more than excellent sample size for PoE's population and has at least a 97% confidence level. If you are talking about something, like mathematics, statistics and sociology, it's recommended to at least know some basics or you'll probably make a fool of yourself.

Article about sample sizes in sociological surveys (table 1 on page 3)

The article you linked says that its "approaches to determining sample size have assumed that a simple random sample is the sampling design". The straw poll you linked, however, uses a self-selected sample. That 97% level of confidence is only valid for a simple random sample.
Self-selection sampling is good if you want to get the views of people who are motivated to respond (eg: people care a lot about the topic). However, it is not good if you want to be confident that you have a representative sample.

Producing two mouse clicks isn't exactly a stopping factor, as evidenced by the large percentage of voters who said they didn't care. It can't be compared with, say, having to answer 50 questions and set apart 30 minutes of your time, which happens in a lot of real-life surveys. Also, aren't the players who like the lab also motivated? Tinfoil and some other players' consistent efforts seem to indicate so.

Anyway, you're at least civil in the discussion, so I'll leave it at polite disagreement there. He's a prime candidate for the 'don't bother talking to' category. I knew it was a mistake to dip my feet in the cesspool but I suppose his bait is high-quality. It's just something that has bugged me about the forum's attitude towards steam charts and straw polls with a lot of voters: 'if I don't like the data, the data is unreliable'. The reality is, some of the data is perfectly valuable.

"

Bars wrote:

"

vio wrote:

in before ggg closes this forum and moves it to reddit...
you guys depress me!

lol
gotta admit, there seems to be a great many full-time shitposters lately

trying my best to lower my posting frequency, hehe.
no, the thing is, i went through the (often crap) suggestions i made over the years and that probably put me in bad mood for several reasons.

---

on topic, roughly one third of the people writing here ... i even get a weird feeling in my stomach when they make enthusiastic or otherwise positive posts because i know it's going down even deeper in a week.

for what? i don't know.

all what those hate posts about a game mechanic will achieve is that players who suck at positioning skills and reaction times will look them up and get the impression that it's not their bad skills which are to blame and they should work to improve them ... IT'S THE NEW MECHANIC! it's not their fault!

it's all just another round of the good old game: "if you can't beat the game, just change it's rules to your advantage", kids do that naturally all the time if you play with them.
i'm out.

"

vio wrote:

all what those hate posts about a game mechanic will achieve is that players who suck at positioning skills and reaction times will look them up and get the impression that it's not their bad skills which are to blame and they should work to improve them ... IT'S THE NEW MECHANIC! it's not their fault!

it's all just another round of the good old game: "if you can't beat the game, just change it's rules to your advantage", kids do that naturally all the time if you play with them.
i'm out.

It's never been about not being able to beat the lab. Beating the lab is not a problem at all if you know what you're doing. It's about not liking it for a variety of other reasons. But let's not turn this into the Nth thread which degenerates from something entirely different to lab discussions. I won't post about this in this thread anymore.

https://en.wikipedia.org/wiki/External_validity

The sample size argument is really trying to address this without using the right words.

"

Bars wrote:

The lab straw poll has 2400 voters. This is a more than excellent sample size for PoE's population

We discussed this in one of those 'I hate lab' threads, the size is certainly good but the sample structure is bad.

Until devs make an ingame poll that

"

- randomly selects voters from the pool of people that finished the lab at least once
- gives them something for their trouble so nobody skips voting (a random box or something)
- provides the usual five graded answers (yes, somewhat yes, neutral, somewhat no and no)

we won't have accurate data.

Replying only out of respect for your opinion despite my promise not to post anymore - it's not perfect but it's the best we have to work with and entirely dismissing it like a lot of 'unbiased' posters from the pro lab party would be much more of a mistake than taking it as 100% truth, especially when there are other corroborating circumstances. If, say, everyone you know loved the lab, the lab race ladders were lively and most forum posters seemed pleased with it and then suddenly this poll apeared, it would be strange and suspicious. But this is certainly not the case. We can always construct scenarious to discredit it but it would go against Occam's Razor which I'm very fond of.

"

Bars wrote:

Replying only out of respect for your opinion despite my promise not to post anymore - it's not perfect but it's the best we have to work with and entirely dismissing it like a lot of 'unbiased' posters from the pro lab party would be much more of a mistake than taking it as 100% truth, especially when there are other corroborating circumstances. If, say, everyone you know loved the lab, the lab race ladders were lively and most forum posters seemed pleased with it and then suddenly this poll apeared, it would be strange and suspicious. But this is certainly not the case. We can always construct scenarious to discredit it but it would go against Occam's Razor which I'm very fond of.

Thanks, it's mutual.

I've been thinking something, devs wouldn't want to waste time on doing something like that, but the community could probably pull it off with a bit more elbow grease. On the plus side, there are more of us so we could split the work.

So, the plan would be to gather up twenty trustworthy (and bored) community members and get each to interview 50 random players, ten players every day for five days isn't all that much work. We would randomly select them by using a number generator on the part of the temp league ladder that is above, say, level 40. It would be performed a month into the league when the initial rush is over and the things settled down, we have even the most casual people ascending at least once and the server population is still high.

I'm interested in finding out what the community really thinks here so I wouldn't mind being one of the 20 roninses. Maybe a few of the streamers could get involved too, the process would be fully transparent that way and it might make some good viewing.

Ideally, devs would provide us with access to a script that spits out a random account out of all the players online at the moment, but temp leagues make the majority of the playerbase at that point in time so it should be fine. Or we could split it 30 from ESC, 10 from EHC, and 10 from standard, I think we should get really solid data that way.