Drink driving: monitoring data

We currently publish annual provisional estimates of drink drive casualties around 6 months after the end of each year. These statistics are then revised and finalized a year later, based on more complete data.

We have become aware of limitations in the accuracy of the provisional estimates and are proposing changes to improve future processes and are seeking your views as to the best option. The options are to:

• retain provisional statistics with their current form and timing
• stop publishing provisional statistics entirely
• delay provisional statistics to later in the year to allow a larger, more representative sample to be gathered
• adjust the estimates to account for the bias
• produce the estimate as a range, rather than a single “best” estimate