Department of Statistics, UCLA
Data Modeling and Theory Construction
- Author(s): Jan de Leeuw
- et al.
One of the purposes of this chapter is to compare, and to a certain extent contrast, two different approaches to the use of models in science. The two approaches are illustrated with Figure 1, a picture of a scientist arriving at the Scientific Forum. In the first case the scientist starts by selecting a model from the model box, he then grabs data from the data box, and holds them against the model. If they don't fit, he rejects the model and throws it away. The Forum applauds. On the other hand if the data do fit he holds on to the model. There is some applause, but not as much. If he has time he grabs new data, and he goes on until either they do not fit the model, or until his time is up. In the second interpretation of the picture the scientist begins by selecting the data. He then grabs a model from the model box, and holds it against the data. If it does not fit, he throws it in the trash. Not much applause. He then grabs another model, and so on, until he has found one that fits, or until his time is up. He holds on to the one that fits, if he finds one. There is also some applause in that case.