Just think for a second about the term "anecdotal evidence." What really is the problem with "anecdotal evidence?" Well the term is an example of a logical fallacy only when one seeks to deduce broad all encompassing conclusions from such evidence. Thus, for example, if you see one drunken irish person it does not follow that irish are drunks. It is anecdotal evidence. But does the same logic follow for plane crashes? I suppose if you have one recorded event of a device interfering with flight instrumentation, you could call that "anecdotal evidence." But what if the plane crashes and it kills 300 people? Well that would still be a rare occurrance, and you could still call this an anecdote. But it would also be a terrible tragedy.
You see, rare occurrences are not sufficient to create an overall rule about the majority of cases, but they are sufficient to create tragedies. I am willing to bet that even most drunk drivers end up getting home safely. Then why ban drunk driving? Because the danger of death even if relatively rare is still unacceptable. Because we value life in this country. So my point is that the usual rule against anecdotal evidence does not apply when dealing with airline safety. In airline safety even things that happen very rarely can be a problem. Because, again, we value our lives.
Their other point is also rather dumb. They say that they did a survey that showed that many people do not turn off their devices completely and few people (2%) even use them blatantly. The thinking here goes a little bit like this: "since some people broke the rules and no disaster happened then it must be ok for more people to break the rule". This is train of logic is similar to the one used in the Peter Principle. This kind of thinking is basically guaranteed to result in a disaster. Because, every time they would say "well nothing bad has happened thus far, so we might as well increase the sources interference", and this will go on until something bad happens.