In this paper we present a statistical study of edge fluctuations taken with the Gas Puffing Imaging (GPI) diagnostics. We carry out a comparison of GPI signal from an extensive database including four devices (two Tokamaks and two Reversed Field Pin
ches). The data are analyzed in terms of their statistical moments Skewness and Kurtosis, as done in [B. Labit, et al, Phys. Rev. Lett. textbf{98}, 255002 (2007)]. The data align along parabolic curves, although different from machine to machine, with some spread around the best-fitting curve. A discussion about the meaning of the parabolic trend as well as the departure of real data from it is provided. A phenomenological model is finally provided, attempting to accomodate experimental evidence.
