Models for the estimation of a ‘no effect concentration’