AbstractAlthough multimethod research has a large number of advocates, problems arise in the interpretation of any discrepancies between results. This paper identifies three approaches to their interpretation: hierarchical validity, cross‐validity and triangulation. The first assumes that methods can be ranked in order of validity so that lower ranking methods can be validated against higher ranking methods. The second assumes there is no hierarchy and that multiple methods are desirable to overcome the weaknesses in any one method and, further, that only results obtained from more than one method should be given credence. The third position, which differs from the others in that it arises from a constructionist rather than a positivist framework, holds that multimethods are valuable because of the different insights they produce and focuses attention on the need to examine methods in terms of their contribution in the context of a particular theory. These approaches are discussed in reference to results for children's negotiations from mothers' reports versus those from a quasi‐experimental puppets t