Abstract:
Intercept-based methods of generating a point estimate of a calibrated radiocarbon date are very popular, but exhibit undesirable behaviour. They are highly sensitive to the mean of the radiocarbon date and to adjustments of the calibration curve. Other methods give more stable results. The weighted average of the probability distribution function is recommended as the best central-point estimate, but more consideration should be given to using the full probability distribution rather than a point estimate in developing age-depth models.