Both the approaches of Rhamstorf and Coumou (2011) and Hansen et al. (2012) are limited in their ability to make firm
probabilistic statements about the changes that are observed because neither uses a validated statistical model in their analysis.
The statistically robust approach used in this paper, incorporates time series modelling, validation and bootstrap simulation
and provides a probabilistic assessment of global warming, strongly complementing the scientific evidence for the
anthropogenic origin of recent climate change. Methods that account for temporal dependencies in climate data have been
considered before in the statistical downs caling literature; see e.g. Charles et al. (2004), but their emphasis was on model
skill and projections rather than attribution, which is essential for the current application.