I'm grateful to correspondents on Straight Statistics who pointed out that the the way "confidence in police" was scored in the national Police Improvement Agency's (NPIA) randomized trial meant that the higher the score, the lower the confidence. I had overlooked the footnote which clearly stated so.

Thus, the NPIA trial - rather than finding that victims' confidence in the police was very much greater than that of non-victims - showed the reverse,  much lower confidence. Correspondents were alerted because  'my headline' contradicted previous research, and a reader who remarked that the the piece on the NPIA trial was below standard for Straight Statistics was absolutely right.

I am very grateful to readers for their prompt correction and appropriate censure, and also for their generosity in allowing that the reporting standard that I had suggested could have bemused the Minister had clearly bemused me. However, I have excuse for having been bemused - the misreading was plain careless. Thank you, readers, for pointing it out.