Yes, randomization is necessary in experimental design, to reduce bias through rendering each sampling unit a fair and equal chance of section or ending up in any condition. It allows us to equalize the effects of unknown or uncontrollable sources of variation.