AbstractThe validity of the assertion that rated and self‐reported behavior is “almost entirely under the influence of preexisting conceptual schemes” is questioned. It is argued that the approach previously used to test this thesis is inadequate, and an alternative approach is described. A comparison of findings obtained using the two approaches suggests that the thesis is in need of revision: The presumed correspondence is neither as high nor as consistent as it may be thought to be. Some implications of the present findings for future research in social cognition and in personality are disc