SIMULATIONS AND OUTPUTS.

Bayesian analysis is now routinely applied for the construction of site-specific stratigraphic chronological models. Other approaches have analyzed the chronology of phases of archaeological activity across regions. The available radiocarbon results--the nature of the samples and their associations--provide the basis for what chronological questions it is possible to address for any site or region. In dealing with regional analyses, due consideration must be made of data selection. While data selection might be a relatively self-evident consideration in the analysis of a site chronology, working with data from a larger region poses a number of specific data selection issues. Robust association between dated samples and a particular type of diagnostic material culture or site may provide one means of producing regional chronologies. However, if the activity under investigation includes a number of different cultural traits, which are related but with each having a slightly different chronological currency, defining the temporal end of data selection becomes more problematic. This article presents one approach, using a case study from the British Mesolithic-Neolithic transition, with 880 simulation OxCal models used to investigate the effect of variously defining the end of a regional archaeological phase. The results emphasize that for a regional case study, sensitivity analysis may provide a useful tool to ensure representative models; the study also highlights the importance of comparing multiple model posteriors. [ABSTRACT FROM AUTHOR]