Far better an approximate answer to the right question, which is often vague, than an exact answer to the wrong question, which can always be made precise.
John Tukey, 1962
At the most basic level, the quality of a goal under investigation depends on whether the stated goal is of interest and relevant either scientifically or practically. At the next level, the quality of a goal is derived from translating a scientific or practical goal into an empirical goal. This challenging step requires knowledge of both the problem domain and data analysis and necessitates close collaboration between the data analyst and the domain expert. A well‐defined empirical goal is one that properly reflects the scientific or practical goal. Although a dataset can be useful for one scientific goal g1, it can be completely useless for a second scientific goal g2. For example, monthly average temperature data for a city can be utilized to quantify and understand past trends and seasonal patterns, goal g1, but cannot be used effectively for generating future daily weather forecasts, goal g2. The challenge is therefore to define the right empirical question under study in order to avoid what Kimball (1957) calls “error of the third kind” or “giving the right answer to the wrong question.”
The task of goal definition is often more difficult than any of the other stages in a study. Hand (1994) says:
It is clear that establishing ...