I'd like to test a 1-1-1 mediation model (x-m-y) with a Level 2 moderator (w) for the m-y path. From what I've read, this should be modeled using a random slope in the WITHIN model for the m-y path, and regressing it on the moderator in the BETWEEN model. But is there a way to do this if I just want fixed and not random slopes? This is what I'm thinking, but I'm not sure if it is correct?
MODEL: %WITHIN% M on CONT; !CONT is a control variable M on X (a1w); Y on M (b1w); Y on X;
%BETWEEN% X M Y W INT; Y on CONT; Y on X; Y on M; Y on W; Y on INT (b3b); M on X; X with W INT; M with W;
MODEL CONSTRAINT:!“ind”refer to !indh/indl = indirect effects when w is high/low NEW(a1 b1 b3 indh indl); a1 = a1w; b1 = b1w; b3 = b3b; indh = a1*(b1 + b3*(1)); indl = a1*(b1 + b3*(0));