I'm the post author. I suspect aside from conducting extremely unrepresentative surveys, they are also aggressively filtering out respondents. I didn't include that since it is purely speculative on my part, but I'd guess that is part of what is happening.

I'd note that other seemingly innocuous things could have an impact too. For instance, if they surveyed only people who've owned their phone for 90 days or whatnot, you would eliminate the vast majority of people who didn't like it and returned it from the sample.

99% is an extraordinary claim, which is why we think Apple should be more transparent about their methodology.