I am somewhat confused about how to interpret this mediating effect result:
IV has a strong and significant positive direct effect on DV. However, when I introduce the mediating construct (M), IV has a significant positive effect on M, M has a significant positive effect on DV, but the positive direct effect of IV on DV changes to negative (though bootstrapping results show it is insignificant i.e. t=1.272). I understand this implies perfect mediation but my worry is why the sign should change to negative. Baron and Kenny (1986) implied that the strongest possible mediation effect occurs when the coefficient drops to zero. Mine drops to negative. Please, can someone help me with the interpretation?