Thanks to Ian Hart for signposting the "no significant difference" reference. I came across it a while back and I don't find it so much depressing as still presenting a vigorous and relevant challenge. To take one step back, I see Richard Clark's position (which this literature survey broadly supports) as "eliminativist" by analogy with a current strand of thinking in the philosophy of mind which declares mental states to be, like phlogiston, a popular fallacy. (Now, I'm waiting for the first person to tell me that I'm phlogging a dead horse!). The "elimination" of mediating technologies from the spectrum of whatever can usefully influence learning or offer mathemagenic potential remains a puzzle, as several people have attested. An interesting point which lies in the background to Clark's issue of his challenge is that earlier (Salomon and Clark, 1977) he thought it evident that mediating technologies produce psychological effects. The "eliminativist" thesis he later published showed up, in effect, a puzzling "gap" in the idea of phenomena capable of producing psychological effects (effects on thinking and feeling) but not learning effects. The "gap" seems even more enigmatic today and suggests something odd or limiting about a conception of learning which somehow cocoons it from effects on thinking and feeling. This notion of "something odd" chimes with Georgina Fyfe's message:
[quoting Fife, 30 Jul 96] ...our results made us realize that the evaluation of learning which we employed in that particular unit was at best incomplete and at worst a poor indicator of any learning which may have taken place. Perhaps, as in our case, many of the website examples tried to evaluate good learning innovations and techniques with assessment instruments which could not record or reflect any significant differences??
I'm sure there must be flaws in my way of thinking about the issue that hopefully other people will point out.