UK Statistics Authority validates TUC concerns over sloppy figures at HM Treasury
This week, Sir Andrew Dilnot, chair of the UK Statistics Authority, issued a written response to TUC’s concerns about poor data practices at HM Treasury. This was a reply to the formal complaint lodged by TUC general secretary Frances O’Grady last October.
We are pleased with the seriousness with which our complaint has been taken and the alignment of the response with our initial assessments. The Treasury made an attempt to address some of our concerns with a revised press release in early November, but their efforts were a bit too little too late. As the UK Statistics Authority response has put it:
“Whilst it is helpful that HM Treasury published its analytical note, we are concerned that it did not do so until a fortnight after the version that Ms. O’Grady attached to her letter. The Authority’s view is that Government announcements drawing upon official statistics, or making significant reference to them, should be accompanied by analytical articles setting out the details of sources, definitions and any assumptions made.”
Consistent with our original analysis, the UK Statistics Authority reply aptly makes the point that even after more fully citing their data sources, the Treasury’s presentation of female employment growth by industry remains misleading. In case you missed it, this is what the Treasury published:
The Treasury’s graph obscures the fact that female employment growth in the services sector was much higher than in the other three sectors. We showed this by graphing the percentage increase of female employment by sector:
To a similar effect, the UK Statistics Authority has now shown this by graphing the absolute increases to female employment by sector:
Sir Andrew Dilnot’s reply went one step further, noting that the vastly larger number of women who were already working in the services sector should also be acknowledged. His response illustrated this as below.
We hope that the Treasury will take greater care with its statistical releases in the future.