The analysis of effective connectivity in neuronal networks has become an essential tool for understanding the underlying mechanisms of information transfer in neural circuits. However, its interpretation is not always straightforward, and it can be masked by the presence of artifacts.We scrutinized calcium imaged data from in vitro neuronal cultures grown of PDMS-engineered patterns using a transfer entropy algorithm [1]. Initial analyses revealed the presence of spurious correlations which hampered the actual underlying network. The origin of these artifacts lays on the delicate interplay between the data bin size and the biological timescales of the system, as well as on local noisy regions particular for calcium data. This systematic revision has provided a rule of thumb for the choice of the transfer entropy algorithm parameters, which can also be extrapolated to multi-electrode arrays data, as well as a comparison between Schmitt and OASIS-derived raster plot. Additionally, we explored the use of surrogate data to assess the robustness of transfer entropy estimates. Finally, we also investigated the interpretation of community structures adopting the Stochastic Block Model method [2].Our findings underscore the critical importance of artifact management in deciphering the nuanced dynamics of information transfer within neuronal networks. This work contributes to the advancement of analytical methodologies in neuroscience, facilitating a more accurate understanding of information flow within neuronal networks.[1] O. Stetter, et al., PLoS Comput. Biol. (2012)[2] T. Peixoto, https://arxiv.org/pdf/1705.10225 (2019)