Faraday rotation measure (RM) at radio wavelengths is commonly used to diagnose large-scale magnetic fields. It is argued that the length scales on which the magnetic fields vary can be inferred from correlations in the observed RM. RM is a variable derived from the polarised radiative transfer equations under restrictive conditions. In this talk, we assess the usage of rotation measure fluctuation (RMF) analyses for magnetic field diagnostics in the framework of polarised radiative transfer. We show how density fluctuations could affect the correlation length of magnetic fields inferred from the conventional RMF analyses. In particular, we caution against the interpretations of RMF analyses when a characteristic density is ill-defined. Moreover, the spatial correlations are generally dissimilar along the line-of-sight and across the sky plane, hence the context of RMF must be clarified when inferring from observational data. In complex situations, a covariant polarised radiative transfer calculation is essential to properly track the radiative and transport processes, otherwise the interpretations of magnetism in galaxy clusters and larger scale structures would be ambiguous. Lastly, we discuss the implications of our work on future radio observations, particularly with the Square Kilometre Array (SKA).