Recently, a number of studies have proposed that the dispersion along the star formation rate (SFR) - stellar mass relation (sigma(sSFR)-M-*)-is indicative of variations in star formation history driven by feedback processes. They found a 'U'-shaped dispersion and attribute the increased scatter at lowand high stellarmasses to stellar and active galactic nuclei feedback, respectively. However, measuring sigma(sSFR) and the shape of the sigma(sSFR)-M-* relation is problematic and can vary dramatically depending on the sample selected, chosen separation of passive/star-forming systems, and method of deriving SFRs (i.e. H alpha emission versus spectral energy distribution fitting). As such, any astrophysical conclusions drawn from measurements of sigma(sSFR) must consider these dependencies. Here, we use the Galaxy And Mass Assembly survey to explore how ssSFR varies with SFR indicator for a variety of selections for disc-like 'main-sequence' star-forming galaxies including colour, SFR, visual morphology, bulge-to-total mass ratio, Sersic index, and mixture modelling. We find that irrespective of sample selection and/or SFR indicator, the dispersion along the sSFR-M-* relation does follow a 'U'-shaped distribution. This suggests that the shape is physical and not an artefact of sample selection or method. We then compare the sigma(sSFR)-M-* relation to state-of-the-art hydrodynamical and semi-analytic models and find good agreement with our observed results. Finally, we find that for group satellites this 'U'-shaped distribution is not observed due to additional high scatter population at intermediate stellar masses.