Are GRB980425 and GRB031203 real outliers or twins of GRB060218?

GRB980425 and GRB031203 are apparently two outliers with respect to the correlation between the isotropic equivalent energy E_iso emitted in the prompt radiation phase and the peak frequency E_peak of the spectrum in a νF_ν representation (the so-called Amati relation). We discuss if these two bursts are really different from the others or if their location in the E_iso -E_peak plane is the result of other effects, such as viewing them off-axis, or through a scattering screen, or a misinterpretation of their spectral properties. The latter case seems particularly interesting after GRB060218 that, unlike GRB031203 and GRB980425, had a prompt emission detected both in hard and soft X-rays which lasted ~2800s. This allowed us to determine its E_peak and total emitted energy. Although it shares with GRB031203 the total energetics, it is not an outlier with respect to the Amati correlation. We then investigate if a hard-to-soft spectral evolution in GRB031203 and GRB980425, consistent with all the observed properties, can give rise to a time-integrated spectrum with an E_peak consistent with the Amati relation.