My favourite anecdote about this is from the Vietnam War:
US General: `We have less reports of VietCong infiltrating villages, hence we are winning the war' Journalist: `In your experience General, how often do the VietCong contact the US Defense Dept to inform them that they have infiltrated a village?'
What a surprise that the US lost the war.
Hence, it comes as no surprise to me that the `facts' dressed up as `hard data' regards schools & the health service (or anything else for that matter) presented by the government ad nauseum bear no correlation with my personal experience. They're just using rhetorical smoke & mirrors to push home what is their own largely hidden agenda.
Both "personal experience" and statistics can be highly subjective. But it can be easier to sell the latter to politicans. Especially if they don't understand what it actually is which is being measured or if whatever it is is meaningful.
But most people fall for the `smoke & mirrors' as every magician knows.
Even when people know trickary is involved. Though with something like politics too few people are skeptical, IMHO. Indeed often much is put into essentially meaningless figures, such as reported crime figures. (Together with such stupidity as assuming that information on reported crimes can tell you everything about unreported ones.)
Unfortunately, an industry has grown up around statistics because people think it has a scientific grounding and therefore they think that they must be telling the truth. But stats are prone to misuse and misinterpretation by people who don't understand mathematics and it's
Hence "lies, damned lies and statistics" :)
limitations or by people who are trying to `prove' some point.
Also as you well know there is no one `Statistical Method' and what
Without knowing the method understanding the results can easily be impopssible.
they all have in common is that you can't prove anything with them in a formal mathematical proof sense ie: by induction, counter-example etc.
Mathematics and engineering on the other hand you can use statistics satisfactorily because you can model simple systems and you can use your statistical results to see whether they conform with your model by empirical means.
I'll address the `positive discrimination' part of your earlier post at a latter data. Needless to say I disagree with you :)
"positive discrimination" is term invented for political purposes. Since any kind of discrimination can be described using either positive or negative languages and phrasing. -- Mark Evans St. Peter's CofE High School Phone: +44 1392 204764 X109 Fax: +44 1392 204763