Hm. The rest of your article explains that tone can be bad for effectiveness, and when it is, it's reasonable to act on that information.

You do argue that a confrontational tone will put people on the defensive. Which seems true on average, and ceteris paribus will result in persuading fewer people than a nonconfrontational tone. But there are "non-average" effects, for example if people who don't immediately become defensive are also more (or less) likely to be persuaded by confrontation, that gives a positive (or negative) term. And there are "ceteris isn't paribus" effects, where a confrontational tone might get noticed a lot more (or less) than non-confrontation, gaining a positive (or negative) effect that way.

Anyhow, one could find anecdotes both pro (yay google) and con (I guess we'll count you for that one. We'd prefer anecdotes of opinion because of rather than opinion on, but those will be scarce unless Dawkins actually makes people join religions). To get non-anecdotal evidence, maybe we could estimate some long-term impact on atheism? That gets awful theory-laden awful fast, though. I guess you'd need some persuasion studies on the marginal impact of tone, and a model that could account for effects like those I mention above.