As observers we hope to understand the evolution of the galaxy population by measuring the star-formation rate (SFR) density of the Universe as a function of look-back time, and by studying relationships between SFR and other galaxy properties such as stellar mass. These analyses rely on measuring accurate, unbiased, total SFRs of galaxies at all redshifts, accounting for both escaping ultraviolet (UV) light and reprocessed far-infrared (FIR) emission from dust. As yet we have a very limited understanding of obscuration in typical star-forming galaxies at high redshifts. Galaxy samples at z > 3 become biased towards either FIR-bright or UV-bright tails of the population, as a result of the photometric selection techniques. In this study we strive to bring the two ends of the spectrum together by combining the deepest 450-micron survey with the latest techniques to break through the confusion limit. We show that the UV luminosity is a poor indicator of total SFR, and that rest-frame UV or FIR selected samples respectively probe opposite extremes of obscuration. We compare the efficacy of each as a probe of the star-forming galaxy population from redshift z~0 to z~6, and assess the impact of sample selection on the conclusions drawn about the evolution of the galaxy population.