The photospheric iron abundance is important to understanding solar system formation and evolution, but methodological designs in the literature have produced significantly different abundance estimates. 290 original photospheric iron abundances were extracted from 68 primary publications identified through SAO/NASA Astrophysics Data System and Google Scholar. Methodological characteristics corresponding to each abundance estimate were categorised, including the ion studied, local thermodynamic equilibrium (LTE) assumption, the atmospheric model and its dimensions, the source of oscillator strengths, and microturbulence velocity. A linear mixed-effects regression model was implemented to quantify the role these variables play in abundance estimates. The mean photospheric iron abundance across all sources was 7.44 ± 0.02 dex on the astronomical log scale. On the average, abundance estimates from Fe II lines were 0.121+0.069-0.069 dex higher than those from Fe I lines (main effect), after controlling for other factors. Compared to LTE estimates, NLTE estimates were higher by 0.114+0.087-0.087 dex (main effect). The role of LTE was different for neutral and ionised iron analyses. LTE abundance predictions were lower than NLTE predictions for Fe I, but higher than NLTE predictions for Fe II. Other differences are described for various model atmospheres and oscillator strengths. A subgroup analysis of 245 1D and mean 3D model abundances suggests every 0.1 km s −1 increase in the microturbulence velocity parameter results in a 0.04 dex decrease in estimated abundance. These findings explain much of the differences in photospheric iron abundance estimates over time and have implications for future abundance research.