On establishing the validity of 'objective' data: can we rely on cross-interview agreement?

It has been proposed that a high level of agreement between informants' reports about events and happenings ('objective' data) is a sound basis for establishing (a) the validity of the measure which is based on the reports, and (b) the nature of objective reality. In this paper it is argued that such agreement may not be an adequate basis for validation since the accounts may not be independent. Likewise, low across-interview agreement may not signify the inaccuracy of either report because of the phenomenon of 'object-variation'. It is concluded that, although this method of validation can be significantly improved, it should be supplemented, wherever possible, by other approaches.