Saturday, December 5, 2009

How Science Works

Scientists don't just think of an idea and do a simple experiment that beautifully verifies it - except in extremely rare instances. Normally data is mucky, has noise, that is say that not all the evidence will support the idea, even when it is mostly correct. Some of the data may not be too relevant to the domain in which the idea applies - try proving Newtons 3 laws of motion in everyday life - you can't, this does not mean that the 3 laws are wrong, just that you need to find the appropriate conditions.

When dealing with something as complicated as weather, one needs a good understanding of Statistics, in addition to knowledge about Physics, Chemistry, and Geography. It also helps enormously if you understand the jargon, and the difference between informal discussions and peer reviewed conclusions. Plus it helps to realize scientists are human, and therefore not expected to be diplomatic about climate denialists in private correspondence.

So sometimes scientists have to either omit data, and/or modify it to be useful.

For example if two boys where trying to decide which of their girl friends preferred chocolate the most, by analysing what the girls bought - they would omit any data relating to buying unrelated items like clothes. Similarly if they wanted to to decide the best level of loudness the girl friends liked music, they would adjust the data depending size of room (and several other factors) to 'normalize' the measurements - so that comparing loudness associated with a party of many people in a large room to the volume for 2 people in a small room. Or planet hunters, who deliberately omit the light from a star to pick the light of a planet - they are modifying the data.

So if an email talks about selecting data sets, and the reader does not bear the above in mind, and is naturally disposed to be untrusting - then quite the wrong conclusion could be drawn.

For a good coverage of this, see the articles about the significance of the leaked climate emails and his follow-up by Phil of Bad Astronomy (he started the site to debunk the Apollo Moon Hoax and others of a similar ilk).

No comments:

Post a Comment