In the dilute s-state Potts formulation of the site-bond correlated percolation, two current treatments of the single-site interaction in the Migdal-Kadanoff renormalisation group are analysed. It is pointed out that they respect the Ising symmetry and would lead always to a critical site percolation probability p_Sc>or=¹/₂, giving a phase diagram qualitatively wrong for dimensions d>2. It is found that the origin of this defect is traced to the preservation of the Ising symmetry which is absent in the percolation problem. An alternative scheme is proposed which gives the correct behaviour.