Statistical considerations in the design and analysis of parallel line assays.

Underlying the design of any assay and further, interpretation of the results, are multitudes of assumptions and implications, ranging from 'biological' assumptions about the nature of the assay system and its response to the materials assayed to 'statistical' assumptions about the form of the dose-response relationship and the distribution of the response data. As far as possible, assays should be designed and analysis carried out to assess these assumptions. Implications for the individual assay are discussed, since this is where all studies of assays necessarily begin. Consideration is then extended to implications for the combination of data and results of several assays.