That particular discrepancy is easily explained by the fact that the amount of effort required to fill out an entire survey and the amount of effort required to click a check box on a poll are completely different (hence, also, the large discrepancy in overall number of responses). In that case we can simply defer to the simpler poll as a more accurate representation of the player base. The more detailed poll still gives us a very interesting picture of how the player base is responding to the expansion, however, particularly in terms of what they liked and what they didn't like.
The fact that MCH is at the very bottom of every single poll certainly says something. BRD being so close to the bottom every time instead of at the very top like they used to be also says something. Whether or not you take these polls as 100% fact, they still serve as very useful barometers of where the player base stands.
Mind you, when I say it is "indisputable" that there is a problem, I mean it. You can quibble over the accuracy of these polls all you want to. The fact remains that the difference between before Heavensward and after Heavensward is HUGE regardless of what sources you use.
I also have to mention that I never claimed 100% accuracy, just very close to it, which is what the whole point of probability and sample size confidence levels are for. A very high confidence level is still enough to render dispute largely pointless. You're not going to be able to take a survey with a sample so biased that it turns the results around. That's just not going to happen.
