If there is collective scientific knowledge, then at least some scientific groups have beliefs over and above the personal beliefs of their members. Gilbert's plural-subjects theory makes precise the notion of ‘over and above’ here. Some philosophers have used plural-subjects theory to argue that philosophical, historical and sociological studies of science should take account of collective beliefs of scientific groups. Their claims rest on the premise that our best explanations of scientific change include these collective beliefs. I argue that Gilbert's account of collective scientific belief does not provide a better explanation of scientific change than a non-collective alternative. A different defence of collective scientific belief and knowledge is needed.