Background

Epidemiological inferences about the aetiology of a disease can often be made from its seasonal patterns.

However, due to its multifactorial nature, the seasonality component can be obscured by other factors.

It is therefore important to develop statistical techniques which are sensitive to minute temporal changes.

Methods

The Lorenz curve and the associated Gini index are applied for characterizing and testing seasonal variations.

Computer simulations were conducted to compare the powers of the Gini test and other seasonality tests.

We also show that the Gini index can itself be interpreted as a probability related to temporal clustering.

Results

The powers of the proposed tests are shown to be higher than or at least comparable to other tests under various conditions.

Conclusions

Though computer-demanding, the proposed method is well-suited for analysing seasonal data.