Yes you do. Not necessarily millions, the amount depends on the amount for possible variation.
You messing up things completely, a pool of opinion and consume is a complete different matter. How you calculate it is completely different. It’s not a matter of random results, it doesn’t follow mathematical equations.
In studies like that the company will study the population and that subject and then select places and a number of people that will represent the general response in the whole country. That is then extrapolated. But it’s not an ACTUAL data, it’s predictions, it can be completely wrong (hello, election of 2017?) if the people interviewed doesn’t represent the country’s variety and different opinions.
This has nothing to do with that, you are completely wrong. Here we talking about one equation, actually computer code, that will randomly generate a result. We do not know the parameters of that code and at what rates each item is set to have. In order to figure this out, you need to select data from each tier in huge amounts.
Again, the dice example, 1-100 face dice, how many tries do need to figure out how much faces my dice has if I didn’t know it? Saying just a few will do is not true. 50 tries won’t be able to tell you that, even if you get the 100, you can’t be sure that’s the highest. For that you need a lot of repetition and thousands of tries to have a lower error and be able to confirm it probably really has only 100 faces.
We have no ideia how much is the odds of the rare and epic stuff. It it’s something like 0,5%, that’s 1-200; 0,1% it’s 1-1000! Few tries might actually not show a single of that rolls or actually show a lot of them.
There’s mathematical equations to calculate error, the more results you have, less error, and so a better understand of the ACTUAL odds. A lot less results have such a huge error that the data is completely useless.
Also wrong for the hundred time… please, read, you cannot make a single data from different loot tier with different odds. I’m not even talking about scores, I’m talking about TIERS. If you wanna now the odds for every tier, it’s one study for each tier. If you just calculate all together, your data will mean nothing other than a average for tiers and won’t answer the questions if one drops more than the other.
Other than that, I’m done trying to explain this subject. This is NOT a thread about odds to begin with. So if you guys wanna do some average or whatever, you can go ahead and start a thread about that and do whatever you want.
I’m not doing that, I have no intention to check odds or prove and disprove “wild claims”, most of each were already answered by the developers themselves. If the person don’t believe what should be a fact, any incomplete data will not be prove for them either. Someone will just point out the obvious, your study has a huge error margin, means nothing, and continuous to claim wild stuff. So nope, I’m out.