j .
On a related note , we have also tried to improve our methodology for estimating statistical uncertainty by using a bootstrap approach , in part because of the resampling involved in the existing matching analysis . Following on from this , we also decided to avoid reporting statistical significance at the 95 per cent confidence level only , since this is an arbitrary threshold . We now additionally report the full range of estimates from our matching analysis ( and the associated probability density ) through the use of violin plots . 49
This report is an official statistic which falls under the Code of Practice for Statistics . We welcome any feedback on our approach . Please email any comments to Stanley Rudkin at official . statistics @ officeforstudents . org . uk .
49
Available in the datafile associated with this release at www . officeforstudents . org . uk / publications / uniconnect-national-evaluation-updated-analysis /.
68