Friday, 29 August 2014

Scandal: Britain's rulers, judiciary and media are a clique of art and law graduates

There's been lots of press this week, rightfully noting that Britain's rulers are an elite clique of a few schools and two universities.

What was not picked up on was

  1. All of them have arts and law degrees. Usually the arts degree is "PPE: politics, philosophy and economics", which is an oxbridge-only degree with no real-world relevance outside local and national government. 
  2. A whole six percent of the current MPs have science degrees.
  3. It applies to the media too, where again schooling, networking and the willingness to work £0 interns get you a job.
This utter ignorance of the scientific method "let the data guide you, not your ill-informed opinions" shows up throughout much of government policy.
  • Badger culling. The scientists said it wouldn't work. They were right
  • The East Coast Mainline returning more money to the treasury than any private rail company, yet the government opening up the franchise again. What is it about spreadsheets that they don't believe?
  • The price guaranteed per MWh for nuclear power generated at Hinkley Point C, despite the way the price curves for solar and other renewables saying "massively over the odds"
  • Global warming. It may be bad news, but that doesn't stop you pretending it doesn't exist.
  • etc. etc. We have politicians on all sides of the house who have agendas and beliefs and who are not prepared to consider how many of those beliefs are defensible
And as noted, the media are as bad. Which means when the politicians say things that are blatantly wrong, this doesn't get picked up on. The classic example is when politicians say that cycling in the UK is safer than the NL as annual fatalities/head of population are less. They can quote the numbers because they are dividing the death rate by the wrong value ... it is as defensible as saying "the uk is safer as annual cycling fatalities per sheep in the country" is less. It's completely bogus, either said knowingly to defend a position —confident that it won't be challenged by a media without the understanding— or uttered because they are clueless and don't recognise its flaws themselves.

This whole "25% of London Guide dog owners" story is the ignorant abuse of statistics:
"A survey by the charity found one in four blind and partially-sighted people were involved in a crash"
As has eventually reached the press, this came from 14 people who may or may not have had guide dogs, and therefore is pretty much meaningless

In contrast, we are data driven organisation.

Admittedly all our content is self-selected, hence utterly prone to selection bias and observer bias —but we are aware of this fact and happy to admit it.
  1. we back up our assertions with photos and videos, not anecdotes about cyclists hitting wing mirrors and cycling off. 
  2. by surveying the same streets repeatedly, we build up a defensible dataset. For example, we have no evidence that the fabled "stokes croft bike lane" actually exists other than as a short-stay parking area for the post-office and the mercedes owner who lives next door to it. 
  3. When our assumptions are proven incorrect, we are happy to admit it. For example, when we had evidence that the infamous YA55VDY delivery van could park legally when there was a space for it.
we also know what a survey is. 

For the reference of others
  1. Census: you measure the entire population (i.e. set of things you care about).
  2. Survey: you measure a proper subset of the entire population in the hope that this data can be used to reach conclusions about the entire population. 

The critical thing is that for a survey to substitute for a census, it needs to be a representative sample of the entire population.  If it is not, then it can only be expanded to the non-representative portion of the population.

In the case of the guide dogs, it could be expanded to say "25% of the kind of people who follow our twitter feed and can be bothered to fill in surveys claim to both have guide dogs and to have been hit by a cyclist"

That's a very different statement. But it is all that can be made.

Which brings us round to the point. It's bollocks, but it is a bollocks that will be repeated blindly by much of the press then believed by its readers —and it will also be believed by politicians who don't have a clue about how numbers work.

Maybe we should just make up random facts too.

No comments: