Quite right, I have failed to show anything substantive. And yes, I am too lazy to do the measurements on my own, since I have nothing vested in this topic and plenty of other ways to spend my time and money. It was simply my goal to point out an obvious set of factors which appear to have gone unaddressed in his methodology, and the ways in which they may (as I believe) have impacted his results and invalidated his conclusions.
To go off of what you said, not only do I believe—without having the numbers to back me up—that people who have used slide-out keyboards will correlate heavily with those who preferentially chose them at some point, I also believe that that idea is self-evident to most people. And I also believe—again, without the numbers—that the class of user who has considered and rejected slide-out keyboards without ever having used them is significantly larger than the class of user who at some point has used them (i.e. the ones allowed in the survey). In failing to consider those factors, I also believe that Bennett inadvertently weighted the pool of surveyed users towards those most likely to favor slide-out keyboards.
All of which is to say, you're right that I have no factual basis for my assertions (and I'm glad that folks around here still take people to task for valid reasons such as yours), since I haven't done the surveys or pulled together my own results. Even so, that doesn't change that Bennet failed to consider two factors that have the potential to heavily skew his results, and that, as a result, he lacks a solid basis for concluding "that the near-extinction of slideout-keyboard phones in retail stores is probably not in proportion to what people actually want" (unless by "people" he means "people who used slide-out keyboards"). The only thing he can reasonably conclude, based on his methodology and results, is that among users who have used slide-out keyboards, more than half prefer them. That's it, nothing more, and that remains true whether or not my assertions are correct.