Ben Goldacre always has interesting things to say about the science behind health care and the pharmaceutical industry. In one of his recent posts he writes about a research project that examined 72 trials with negative results, i.e. an investigated drug or treatment did not cause a desired effect. Out of all of these trials, he quotes that only 9 gave any figures in the trials' abstracts and that 28 gave no numerical results at all.
What was in the reports was "spin," or the authors' attempts to project the results in a positive light. In order to prevent this, he says, trials are supposed to be registered before they are performed so that their intended purpose can not be changed. Additionally, there are guidelines that dictate what must be included in a report. These rules, however, are more akin to suggestions since there is no enforcement of them.
Can such a system be implemented in the physical sciences? I don't think so. Often, we're actually learning about the topic as we proceed through the research. No amount of preparation can allow us to establish a hypothesis sufficient for inclusion in a detailed report before we undertake the experiment. Hypotheses, I feel, are best constructed concurrent with an experiment. And as for report guidelines? Well, anyone who has had to deal with reviewers' critiques of their papers will tell you that there is rarely any consensus about what makes a good report.
I suppose that one could argue that a grant proposal tries to satisfy this purpose, but I can't say that I'm experienced enough to comment one way or another on it.
A "lab notebook" about my life in science, programming, and philosophy.
