Context. It is still being debated whether the well-known metallicity-giant planet correlation for dwarf stars is also valid for giant stars. For this reason, having precise metallicities is very important. Precise stellar parameters are also crucial to planetary research for several other reasons. Different methods can provide different results that lead to discrepancies in the analysis of planet hosts. Aims. To study the impact of different analyses on the metallicity scale for evolved stars, we compare different iron line lists to use in the atmospheric parameter derivation of evolved stars. Therefore, we use a sample of 71 evolved stars with planets. With these new homogeneous parameters, we revisit the metallicity-giant planet connection for evolved stars. Methods. A spectroscopic analysis based on Kurucz models in local thermodynamic equilibrium (LTE) was performed through the MOOG code to derive the atmospheric parameters. Two different iron line list sets were used, one built for cool FGK stars in general, and the other for giant FGK stars. Masses were calculated through isochrone fitting, using the Padova models. Kolmogorov-Smirnov tests (K-S tests) were then performed on the metallicity distributions of various different samples of evolved stars and red giants. Results. All parameters compare well using a line list set, designed specifically for cool and solar-like stars to provide more accurate temperatures. All parameters derived with this line list set are preferred and are thus adopted for future analysis. We find that evolved planet hosts are more metal-poor than dwarf stars with giant planets. However, a bias in giant stellar samples that are searched for planets is present. Because of a colour cut-off, metal-rich low-gravity stars are left out of the samples, making it hard to compare dwarf stars with giant stars. Furthermore, no metallicity enhancement is found for red giants with planets (log g < 3.0 dex) with respect to red giants without planets.