There is a correlation between “being his friends” and “leaving”.
CCP should kick him from Eve to increase the retention rate.
2 Likes
Quit with the throw away lines. Give us a valid explanation of why the dataset was poor.
Maybe you meant to write something different, because that statement is so wrong =.
Correlation isn’t always causation, however causation always has one or several correlations that can be observed.
For example, it’s likely that this year there have been multiple murders, however being 2019 isn’t the cause of murders. There’s unlikely to be a causal link between the observation that it’s 2019, and that there are murders occurring.
On the other hand, smoking correlates with an increased risk of developing cancer among the population. There is absolutely a causal link in the correlation of observed rates of cancer among smokers.
Causation always has correlations; and even if you don’t know the exact causes associated with observed correlations (eg. as in the CCP data analysis), the correlations are still important:
The best line from that post:
Can we ever understand systems that we have limited or no control over? This would be a very bleak state of affairs, and fortunately there has been progress in answering these questions in the negative!
That’s what makes the type of data science CCP conduct, valuable. Because it’s not required to know all of the causes in order to take action on the things that are observed.
For example, in the CCP study, people that do have their ship blown up but who never join a corp, still have a 65% chance of not playing by month 4.
That’s the origin of CCP’s statements in recent years that they know that joining a Corp helps player retention (even though the system is too complex to know all of the reasons why). And there’s more useful information out of CCP’s studies than that.
2 Likes
Only half correct, it was changed to improve retention and recruitment.
As I said earlier in this thread, my comment wasn’t meant to be taken literal but more of a poke at the hundreds of people that told me and others that the player count would drastically increase if the mechanics we’re changed.
2 Likes
It is ok. I felt like writing a story. We both got our thing. 
1 Like
This is priceless given how much you were just whining about strawmans…you get so upset when you have to back track.
If there was a causal link between the two then correlations shouldn’t be ignored. And there is a common and causal link between nerfing hi-sec pvp and player activity/retention.
It’s funny you jump on anyone who quotes data suggesting pvp actually helps the game but never on anyone who uses the even more sketchy reference to data that suggests wardecs have a bad effect on the game.
Did you post in the thread on the csm minutes? You must have been dying inside with all that correlation being thrown around.
So when there’s a deal for plex on, player activity goes up? Let’s test that.
The leavers are leaving.
What did you expect?
What did i say carebear 101 was?
4 Likes
Did you watch the YouTube videos? It’s my understanding that they’re packed with statistical expertise.
I would guess they cover “why correlations mean nothing at all, especially as they approach 100%”.
It sounds like a daring point to me, especially to deiver to people trained in statistics.
Maybe I will watch them, actually. Should be fascinating.
no that’s false. There can be causation that can’t be observed, typically when you don’t look at the correct events.
That’s why I said correlations are in the eye of the beholder. Very often people find very correlated events but don’t realize they both are caused by a third one. Even worse, sometimes a causation can be not detected because of the noise, eg. another causation that has more effect.
In your example, smoking could be also linked to people wealth and thus the reduction of lung cancer due bad working environment, or prediction of cancer apparition. In such an environment, even if smoking really increases the cancer risk, it may be statistically correlated to a better health and thus to less cancer.
No it does not. Just because people not joining a corp correlates to people not playing does not mean that making people join a corp means they will play. As it is written in your article
To revise the quote, we might say “there is no correlation without causation somewhere ”
It does not say you can use the correlation as a causation. It says you can search for factor and test if those factors are a causation - if they are not, find other factors, there is at least one factor that is the common cause of those events.
I just don’t give a fock, this sentence means nothing.
nope. There is not.
So my narrative being ? Yeah, you have no idea because I just pointed when you say BS and did not give my opinion. And now you are making a strawman again.
Strange that what you do all the time does not apply to you. Vested interest much ?
Just watch them. Faster than to reply when you know nothing.
1 Like
This is exactly why we use statistics.
For things where we understand causation, we solve our problems using analytics methods.
When we can’t describe what’s happening or what’s causing it, we use statistics.
That’s the point. I would consider this problem to be a brilliant application for statistical methods.
In fact, I’d love to run some of those computations if I could get access to build the queries for the data set.
Would be a fascinating project.
And yeah, if I found a 98% correlation in the behaviors of people… I’d spike the ball and call it a touchdown.
1 Like
No one is saying that. That is crazy to draw that conclusion and that would be totally wrong. Where you get that from is fantasy.
Of course not. That’s also a pretty stupid thing to write. Go learn some basic stats before thinking that one thing being true means another must be.
If you aren’t looking at the correct events, doesn’t mean there is no correlation. Re-read what was written.
You are jumping to stupid levels of interpretation.
If there is a causal link between two or more things, there will always be correlations that can be observed.
Not looking at those correlations doesn’t mean they don’t exist. They still exist.
2 Likes
6 Likes
When the wind blows, the trees move.
That’s a correlation. I could claim it’s causation, but I could get it backwards. I could think the trees are pushing the wind.
I could get the causation totally wrong… but they’re still going to correlate. And… I can still manage to that correlation because I don’t really care which one is the cause… I’m not on a mission to explain it… but rather to fix it.
98% of people who don’t experience PvP in their first 4 months quitting is plenty compelling for any reasonable person to assert that PvP needs to be delivered to every new player (timeframe could be found statistically). You could also statistically determine what type of PvP is most effective.
But that 98% correlation is powerful. Don’t deluded yourself into thinking it’s not.
3 Likes
“powerful”. It’s just a correlation, so it can be totally not linked at all. and forcing people to pvp may actually make them leave more than remain.
No, that’s a causation. People empirically showed than WHEN you blow the tree, it moves. And when you place a wall to prevent the wind from blowing in the tree, it stops moving. So the wind makes the tree move.
But I get you can’t understand that.
1 Like
You are more transparent than you think you are.
Or maybe you’re kidding yourself.
Like when you nerf hisec pvp the game gets less active…
This is just getting better.
1 Like
Look at the post above, that’s actually exactly what he says.
Well that’s what I’m saying from the beginning of my posts.
Re-read what I wrote. I wrote correlation is never a causation. This means you can NEVER take a correlation alone as a causation, when you observe one in an analysis.
So now please don’t change the context , we are talking about observed events. There can be a causation between two observed events, which does not translate into a correlation in the observation.
Or maybe you have no argument whatever and rather ad hominem.
Or like when you quit eve your friends stop leaving.
But, no, definitely not like what you are trying so hard to imagine - and only exists in your imagination.
Give me an example.
It appears that you are saying that one thing can cause another but there is no observable correlation between them. If that is what you are saying, an example would be good, because that runs counter to decades of research.
1 Like
It keeps happening. So I’ll keep saying it.
2 Likes
Please don’t encourage him, we have had leaking buckets with holes, trees, leaves, smoking and lung cancer as examples already.
5 Likes
Changes don’t happen in isolation so you can’t point at them as the only possible reason.
Also the recent wardec change was not a nerf, because it buffed wardecs by limiting who can ally into a war.
So anyone hiding in a corp without structures can’t ally in to defend any friends structures, making it easier to target those structures. And it increases the social elements of the game.
So… you are trying to cherry pick only a single aspect of a single change here, it’s utterly dishonest.
Sure.
You know how i can tell you’re not reading?
1 Like
Complains about people not reading, spams same words again and again, cherry picks the quote they place without actually reading & considering the rest of the sentence, and offers no actual rebuttal, just 'No You aren’t reading"…
Yep… Clearly a well structured and reasoned argument worth discussing further with you.