Slightly more births at full moon

Abstract

A popular belief holds that the number of births highly increases when the moon is full. To test this belief, we use a 50-year data set of 38.7 million births in France. The signal includes quasi-periodic and discrete components that need to be subtracted. This is done using a non-linear Gaussian least-squares method. It results in residuals with very good statistical properties. A likelihood ratio test is used to reject that the residual means for the 30 days of the lunar month all equal 0 (p-value = $5 \times 10^{-5}$): the residuals show very small but highly significant variations in the lunar month due to an increase of births at full moon and the day after. The reason for the very small increase of birth at full moon is not investigated but can be suspected to result from a self-fulfilling prophecy.