How does science work?

A computer hacker recently stole hundreds of internal emails and documents from the Climate Research Unit (CRU) at East Anglia University in England, and made them publicly available online (NY Times story: http://www.nytimes.com/2009/11/21/science/earth/21climate.html). The emails reveal portions of discussions on analysis and presentation of data related to climate change, and some believe they are evidence of conspiracy surrounding climate change. I am not going to address that – skepticism is a prerequisite for good science, and the theory and data behind climate change should and do stand on their own – but this event does raise interesting questions about science as whole.

There can be a public perception of scientists as “analysis machines”: give a scientist raw data and get back objective conclusions, regardless of the context. This does not reflect reality. There are many subjective elements to science. Simply the process of selecting topics to study is subjective. Framing a hypothesis is a highly creative act, requiring equal parts imagination and critical thought. Likewise, there is no concrete guide for selecting methodology, for resolving uncontrollable events during data collection (people gathering data forget to write down something, batteries fail in a data collection device, etc), for selecting analytical approaches, or for drawing inferences.

How then can science “work?” How can it provide any objective, important observations? The first course in PiE is 5100, “Ecology as a research discipline.” Last year it was co-taught by PiE faculty from botany and philosophy. We extensively discussed this question and related questions about the philosophies behind scientific processes and values. I really enjoyed the course – to me, it seems far too rare that scientists can take time to thoroughly examine their implicit assumptions about science as a whole.

Many very smart people have considered the question of how science can work, and a lifetime could be spent trying to answer the question. Here are just a couple thoughts from my personal perspective. I think asking a good question is important; the question needs to address a very specific gap in knowledge, where surrounding material is well understood. Preferably, the question should have multiple potential answers, each with a clear relationship to the data that will be gathered. And, very importantly, the different answers should clearly relate to each other (for example, are they mutually exclusive?). Also, speaking more broadly, transparency is critical. Scientific reports should include enough information so the subjective judgments described above (framing a hypothesis, selecting methodology, selecting analytical approaches, etc) can easily be understood by the reader.

READING

John Platt’s 1964 article “Strong inference” (Science, vol 146, pages 347-353) lays out an interesting possible framework for doing good science.

PS – On an unrelated note, we had the Zoology and Physiology holiday party last night. The local band “The Patti Fiasco” performed and they were great; we even got faculty on the dance floor. Great job by those who organized it.