Just one thing I'd note about the data they're using. They had really horrifically bad survey design in multiple ways. You can see the outcome in Ranger, which the October 2013 showed people were generally really happy with the Ranger.
Really, surveys need to establish baselines. Who is taking the survey. What are their likes and dislikes. Then ask them questions. Surveys are basically useless without that. And in D&D, that means finding out favorite classes & editions. Do people answer the same way each time, etc...