Maybe it’s because Ima is right?
Lol
this is 100% impossible. He is misusing terms, to the point even if his opinion is correct(which is totally possible), the way he gets to it is 100% wrong.
9 Likes
That is simply false lol
It’s just your opinion, not an actual fact 
he is factually using a correlation as a causation, which is a fallacy leading to his argument being wrong.
This is factual. He is completely wrong. (again, only when stating that CCP data shows anything, not in his opinion)
10 Likes
I guess we are just using the same words for different things. Lets make this simpler so we can find out what you are on about here.
Lets assume we have a dice and someone states that it is rigged and more likely to get a 4.
To test the claim we roll it 60’000 times and get:
- 10’000 times
- 8’000 times
- 12’000 times
- 2’000 times
- 8’000 times
- 20’000 times
Would you say the data does not falsify the claim that the 4 is more likely? Would you also say that this data does not indicate that the dice probably rigged and more likely to get a 6?
your example can not be translated to CCP data. You are not trying to exhibit a correlation between variables.
You are trying to prove that an oracle is correct, by making it predict a value.
9 Likes
I’m trying to find out if there is a fundamental difference between how we use some words. Can you please answer the question? I’m trying to prove nothing
And no it has nothing to do with predicting values. It has something to do with stating an assumption about the dice and checking if that assumption is true by actually rolling it.
It’s not about trying to prove anything, it’s just unrelated.
I can’t find a reason to change the subject.
But to answer : The data you have can only be interpreted to find the probability the dice is rigged.
Here it should be a close to 100% probability it is rigged. In most cases we will define that the dice is rigged and won’t use it anymore.
Then you must define what effectively means “rigged” before launching the dice.
However this is the correct way to get knowledge about something : make experiments that will produce data.
9 Likes
OK, so we agree on that. Progress.
How would you then say would an experiment/measurement has to look like if we want to find out whether the claim “the dice is more likely to get a 4 is true”?
Sorry but I still have no idea why you are asking me this. Plus I don’t know what an experiment “looks like”.
I guess you have very good lessons in statistics in your local university, even college.
Of course, if a term I used seems incorrect or raises questions, feel free to point it out. 
9 Likes
… seriously I genuinely try to find out where the point is where we diverge with the approach to find out how things are. That is why I use the easy example with the dice.
WIth “what an experiment looks like” I mean what experiment do you have to run to find out if the claim “the dice is more likely to get a 4” is true. What kind of data and how would you collect it to find that out. Or are you suggesting there is no way for you at all to find that out?
Obviously this ties in to CCPs study and is not completely off-topic but we can talk about those differences once we have figured out how the simple example works ok?
No, because it s completely unrelated to the example given by CCP. If you cannot accept that examples that simple are not relevant IRL, sorry but I can’t do anything for you.
That said, I assume that you want to know what experimentation could allow us to affirm what kind of relation there is between the “player has been ganked” variable and the “player remains for at least X time”.
Your answer is here Randomized controlled trial - Wikipedia . Anything less strict is biased by definition.
So :
- EVERY new player is RANDOMLY assigned a group : N for NEVERGANK or G for GANK
- make concord reacts instantly for players in group N for their first week.
- be sure to gank every player of the group G at least once during its first week.
- exclude players that venture to low sec during their first week.
Then YOU are the one defining which players are ganked or not, and thus you can extract results such as “being ganked increases the retention rate”.
And yet people will tell you that this experiment has a few biases, and will propose another experiment with less biases , etc.
11 Likes
I’m not sure why you avoid the simple example? Again, I try to find out if we disagree on some basic terminology or the way how we find out if something is true or not… But ok, I guess there is a reason why you avoid the question…
That assumes we have the power to influence the players or the game like that. What if we don’t have that power and are just able to observe? Would there still be a way to find out if the claim "is true or not?
No there is not. If you can’t control the parameters, you can’t prove any relation between the parameters, as I already answer to black pedro before. There is always the possibility of a third parameter not observed influencing on the two other parameters and thus creating correlations when there is no causation at all.
In CCP case you can imagine that the fact people want to remain on the game make them play more, which in turn make them get more ganked. This is just one case of third parameter, there are so many other, that I can’t even imagine.
That’s why making science is so difficult, expensive, and people can’t understand it.
And still scientists publish incorrect data, that need years to be re evaluated, and people keep propagating those false results without checking them…
10 Likes
Condescending much, aren’t you?
the last part is also correct for you.
10 Likes
How?
(five letter limit sillyness)
it means you should learn to do correct sentences.
10 Likes
I see you really can’t help yourself.
No but I can help you by pointing your mistakes. I’m not really selfish 
And I’m very happy you made a real sentence this time. I t looks like you grew a second neuron \o/ victory is mine !
9 Likes