The importance of including dynamic social networks when modeling epidemics of airborne infections: does increasing complexity increase accuracy?