In the previous chapters we pointed out that there are basically two ways to win the battle against noise. The first and by far the most effective solution is to find situations where the effect that one wishes to investigate is sufficiently strong to give a high signal to noise ratio. For instance, the effect on suicide of a male–female imbalance will be stronger in a population whose sex ratio is 20 than in one whose sex ratio is 1.2. This is what we called the extreme value technique and it played a considerable role in previous chapters. That approach is particularly useful in the first stage of a study when one wishes to identify an effect and to get a rough idea of its magnitude. In a second stage the objective is to perform the most accurate measurements of the phenomenon under consideration. In this case one resorts to the second method of reducing the influence of the background noise, namely increasing the number N of events; this means that one will collect and analyze the largest possible number of (good quality) data.

To reach these goals it is essential to have access to a great variety of large databases. Needless to say, the Internet marks a revolution in this respect. Let us illustrate this statement by two examples drawn from previous chapters.