Hey guys, I proposed a model in which high scores in the DV were supposed to be explained by high scores in the mediator variable (comparing three different conditions). Now I received some very interesting data through an experiment: 1) In the condition with the lowest scores in the mediator, participants scored highest in the DV. However, overall, group differences for the three conditions (IV) are non-sign.! I employed the PROCESS model and now the findings turn from interesting to troublesome: 2) there is no sign. direct effect (considering the non-sign. group diff. mentioned in (1) reasonable I think), but a sign. indirect effect! If I understood the literature right, than that´s possible and would state, that the mediator fully explains the DV. So in order to further check that, i tested for the correlation between M and DV. They are positively correlated!!! And now I am completely puzzled... Positive correlation would say: high scores in one correlate with high scores in the other. Which is not the case (see (1): lowest scores in M --> highest scores in DV). I´m more than willing to add graphs, specific figures or numbers to further explain my findings, in case someone needs it. Can someone already make sense of these three different analysis taken together? 1) ANOVA: no sign. effect of IV manipulation on DV, however, sign. effect of manipulation on M --> overall: M and DV go in different directions 2) full-mediation of DV through M, however, no direct effect 3) positive correlation between M and DV I know it's a lot to take in, but after two weeks of trying to wrap my head around this, my deadline comes closer and so does my psychotic breakdown. Best, Nils