Qubort said: broll said: scottee said: Why would they lie? If they wanted to just make an expensive package to "make more money", they wouldn't AB test. They'd make only the 3600 package. Forumites are such conspiracy theorists. To give the illusion of caring about what we think is best. It will probably work for some. Clearly it worked for Scott
broll said: scottee said: Why would they lie? If they wanted to just make an expensive package to "make more money", they wouldn't AB test. They'd make only the 3600 package. Forumites are such conspiracy theorists. To give the illusion of caring about what we think is best. It will probably work for some.
scottee said: Why would they lie? If they wanted to just make an expensive package to "make more money", they wouldn't AB test. They'd make only the 3600 package. Forumites are such conspiracy theorists.
astrp3 said: Warbringa said: You are assuming that the resources for group A and B are the same. You are right. I AM assuming that the resources are the same because if the groups were randomly and properly sampled and of sufficient size, they generally would be (the odds of them being skewed enough to substantially affect results, are increasingly small as the sample size increases, to the point of being negligible).Even if they weren't, however, it would not affect, my main point, which was the counter those who seem to be saying that D3's A/B test (assuming they did one) is invalid simply because they didn't offer everyone a choice, even if it was done properly and proper controls were in place. My point is that, with a properly conducted test, you don't need to offer everyone both choices to determine that one choice was preferred.
Warbringa said: You are assuming that the resources for group A and B are the same.
You are right. I AM assuming that the resources are the same because if the groups were randomly and properly sampled and of sufficient size, they generally would be (the odds of them being skewed enough to substantially affect results, are increasingly small as the sample size increases, to the point of being negligible).
Even if they weren't, however, it would not affect, my main point, which was the counter those who seem to be saying that D3's A/B test (assuming they did one) is invalid simply because they didn't offer everyone a choice, even if it was done properly and proper controls were in place. My point is that, with a properly conducted test, you don't need to offer everyone both choices to determine that one choice was preferred.
Warbringa said: Would not a scenario where many more people from group 1 being placed in the 3600 pool skew the results? In this situation would not the preference be incorrectly measured?
keitterman said: I wasn't gonna buy at 2500, or 3600, but this process does make me very concerned with how D3 makes decisions.
mpqr7 said: I'm skeptical that people would choose to pay more.Is there an option so that I can remove this store from my list, so that I don't need to see it, since I'll never use it?
I am mostly indifferent to this, but I do have a little pro/con list...
Pro:
Con:
That is all. Thanks for reading.