Comments on: One in Five Home Office Statistics Unreliable http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/ Out of the crooked timber of humanity, no straight thing was ever made Sun, 27 May 2012 00:43:58 +0000 hourly 1 http://wordpress.org/?v=3.3.2 By: Decnavda http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/comment-page-1/#comment-184297 Decnavda Wed, 17 Jan 2007 22:27:49 +0000 http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/#comment-184297 Can someone here with a smarter statistical mind than I tell me if there is a way for those of us outside the Home Office to assess the reliability of Home Office statistics? According to the headline statistic, the reliability is 80%, but since the 80% reliability statistic also aplies to itself, I do not know what that means. I do not think I can just say that there is a 80% * 80%, or 64% chance any given statistic is reliable, can I? I think I can say there is a 64% chance any given Home Office statistic is reliable under Home Office measures, and a 16% chance it is unreliable under Home Office measures, but what do I do with the other 20%? Can someone here with a smarter statistical mind than I tell me if there is a way for those of us outside the Home Office to assess the reliability of Home Office statistics? According to the headline statistic, the reliability is 80%, but since the 80% reliability statistic also aplies to itself, I do not know what that means. I do not think I can just say that there is a 80% * 80%, or 64% chance any given statistic is reliable, can I? I think I can say there is a 64% chance any given Home Office statistic is reliable under Home Office measures, and a 16% chance it is unreliable under Home Office measures, but what do I do with the other 20%?

]]> By: Michael Mouse http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/comment-page-1/#comment-184258 Michael Mouse Wed, 17 Jan 2007 16:11:41 +0000 http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/#comment-184258 This drives me up the wall. The Office of National Statistics is the sole beacon of half-decent data in a mess of politically-massaged nonsense. The public - rightly IMO - believe that Government statistics are politically-massaged nonsense. But when you get specific, it's the Departmental statistics that they specifically mistrust (and not just the Home Office). So to restore trust in Government statistics, they're going to make the ONS more independent ... which is a Good Thing, but there isn't a profound problem there and it'll do nothing to deal with the more serious question of the Departmental stats. Gah! [And note that the headline almost certainly understates the case. The one-in-five figure is for datasets that "received a zero rating for reliability". We can safely infer from the fact that the that the figure for datasets receiving a "reliable" rating is omitted that it was considerably worse than 80%.] This drives me up the wall. The Office of National Statistics is the sole beacon of half-decent data in a mess of politically-massaged nonsense.

The public – rightly IMO - believe that Government statistics are politically-massaged nonsense. But when you get specific, it’s the Departmental statistics that they specifically mistrust (and not just the Home Office). So to restore trust in Government statistics, they’re going to make the ONS more independent … which is a Good Thing, but there isn’t a profound problem there and it’ll do nothing to deal with the more serious question of the Departmental stats. Gah!

[And note that the headline almost certainly understates the case. The one-in-five figure is for datasets that “received a zero rating for reliability”. We can safely infer from the fact that the that the figure for datasets receiving a “reliable” rating is omitted that it was considerably worse than 80%.]

]]> By: bert http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/comment-page-1/#comment-184224 bert Wed, 17 Jan 2007 10:32:52 +0000 http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/#comment-184224 Statistics about Cretans available only at disproportionate cost. Statistics about Cretans available only at disproportionate cost.

]]> By: bert http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/comment-page-1/#comment-184223 bert Wed, 17 Jan 2007 10:30:03 +0000 http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/#comment-184223 All Athenians are liars. All Athenians are liars.

]]> By: Sam http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/comment-page-1/#comment-184138 Sam Tue, 16 Jan 2007 19:21:59 +0000 http://crookedtimber.org/2007/01/16/one-in-five-home-office-statistics-unreliable/#comment-184138 Might we call this the bureaucratic variant of Mannheim's paradox? Might we call this the bureaucratic variant of Mannheim’s paradox?

]]>