Two summers ago I had lunch with Professor Jonathan Osborne of Stanford University’s science education department. I was interested in finding a new research avenue for my classroom. Osborne’s work is in scientific literacy. I was looking for a research partner to carry out a study comparing future outcomes of students who had taken IB sciences versus those that had taken AP science courses. I hypothesized that the in-depth lab-work demanded by the IB sciences might have an effect on future success in collegiate science. I never did the study.
Instead, Osborne invoked the 2007 theme from Wired Magazine, Science is Dead. The rest of our conversation consisted of Osbourne arguing that understanding how science is done, and deeply understanding scientific concepts are paramount; practicing the scientific method – not so much. Before we parted, he referred me to this article by Christopher T Hill, that outlines a future for American industry that includes less and less scientific research as it is farmed out overseas.
Really? Maybe. That 2007 issue of Wired magazine identified how more of the ‘R’ in R & D in industry is based on examination of huge data sets instead of the collection and analysis of small ones.
Last week Kaiser Permanente published a research paper that will be all over the news this week. Their findings – women who take antidepressants during pregnancy are twice as likely to have children with autism as those who do not. And get this, they did the study without any experiment. Kaiser has a giant database of patient records. They have published numerous papers in recent years, done by researchers that simply examine huge data sets and look for strong associations. When they find something, they dig in and attempt to retroactively control for everything they can. The result is, well, solid results; associations with strong correlation coefficients that are well controlled.
What is the high school science teacher to do? Do I stop offering experiments in lab time? Certainly reason is not dead. There is definitely much to be learned from the manipulation of materials, the recording of and analysis of data, and the examination of uncertainty as a metaphor for understanding truth in any field. But if the future of science in the US is really meta-analysis, perhaps lab time might be better spent exploring phenomena in ways that will help students identify their own misconceptions instead of repeating a measurement of the acceleration of gravity for the billionth time in human history.
Question Tank 
Don’t believe it. Yes, we’ve collected a ton of data, much of which has gone unanalyzed — and so the kinds of number crunching exercises that Jack describes are valuable. But key to smart science is asking the right question — not just asking the question that your data can address. There will always be great reason for fresh, hands-on investigations, even as the data in storage far exceeds the number of stars in the heavens.