Develop a Model First Posted on October 19, 2007 Work out your model before you design the study, recommends Martica Hall (bio).

So it starts with an observation. Well, then what do you do with it? Well, you could go chat with your friends about it or sit on the front porch and swing with your grandmom and talk about it and probably come at something that may be close to the truth, but that’s not science. Science has a very specific process, and in order to come up with an experimental design, where you manipulate the thing you’re interested in and look at its outcome; if you have a model, it makes it much easier to design the experiment, even if the experiments you’re talking about are, in my case, observational studies in the community that are as uncontrolled as any experimental endeavor could be.

If you don’t have a model, it’s very difficult to design an experiment that really will at least come close to answering that question. And I read papers all the time or review papers or review grants, where I’m not sure what the model is, and if I’m not sure what the model is by the time I get to the method section, you’re in trouble because it should be evident.

« Back to Article