In PLS variance based modeling, if we estimate quadratic effect of a mediator on dependent or endogenous variable and find it significant using SPSS ( by adding squared effect in regression). Next, is it recommedned to esimate mediation on this variable using normal mediation tests