> In this experiment, the task was to complete sentence fragments, either in an abstract or a concrete way. For example, some might complete this fragment: "An example of a bird is ______." Others completed this kind of fragment: "A bird is an example of ______." The first requires a concrete example—an indigo bunting, for example, or scarlet tanager—while the second asks for an abstract category—warm-blooded vertebrates, say
There's another difference between these two tasks: one is easier than the other. The experiment did not control for this. It did ask participants how difficult they thought the task were, and found no significant difference between the two groups. From that they infer that the results are not explainable by difference in difficulty - which seems flawed to me. Since when did perceived / stated difficulty become a barometer of actual difficulty? This is repeated across all three experiments. I suspect they are simply measuring differences in real difficulty, and its divergence from stated difficulty.
Link to paper: http://www.uni-konstanz.de/FuF/SozWiss/fg-psy/gollwitzer/PUB... (the lack of raw data in this paper is worrying, and makes validation very tough).