>Especially when 90% of those asked didn't answer the survey at all - upset users are probably more likely to respond to the survey.
Alternatively: many of the 90% didn't answer because they have already left Twitter.
Alternatively: many of the 90% didn't answer because they have already left Twitter.
You may be right that some users saw a survey about twitter in their inbox and thought "I've already left, therefore I'm not the target audience for that survey".