To run a mediation?

Home Forums Default Forum To run a mediation?

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
  • #2963
    Em R

    Dear All,

    I have a between-groups research design that is looking at the effect of Rating Condition (uncertain, best and least) on attractions levels, reported thoughts and overall mood. Part of my study is also looking at whether participants’ attachment style and self-esteem mediates this effect.


    However, I have since run my ANOVAs which have yielded all nonsignificant results of Condition on the DVs, except for one. Follow up analysis for the significant effect did not reveal any significant mean difference between conditions. 

    Thus my question is this: Because I have obtained nonsignificant results, am I right in saying that a meditational analysis cannot be justified as there was no significant effect to mediate? Or should I run the meditational analysis anyway? 

    I would very much appreciate any assistance or advice as to how to approach this issue. 


    so… i am somewhat unsure as your design here. so do you have one predictor: Rating Condition and 3 response/dependent variables (attraction, thoughts, mood)?


    assuming that i understood you (and before we take the next step) why aren’t you doing a MANOVA instead of 3 separate ANOVAs? how are you protecting yourself against the increase in experimentwise error rate?

    Em R

    I wish I had a more intelligent answer, but basically this is my first assignment for Research Methods and our teacher advised us to use ANOVA. :S

    And sorry I should have elaborated more upon the IVs and DVs:

    Condition (3 levels)

    Attachment style (4 levels)

    Self-Esteem (2 levels)

    DVs: (between subjects)

    Attraction Levels

    Reported Thoughts

    Within Subjects:

    Mood (time 1 and time 2)


    ok… now, what are you calling a “mediational analysis”? because i think you’re using it somewhat differently from what i’m used to (in the context of path analysis through regression coefficents. but that doesnt have a lot to do with ANOVA)


    so… what you have here, at least from what i read , is a 4-way, mixed-model MANOVA (sorry, many DVs imply the use of multivariate analysis of variance or some sort of experimentwise error correction regardless of who says what)  with 3-between-subjects factors (condition w/ 3 levels, attachment with 4 levels and self-esteem with 2 levels) and one within-subject factors with 2 levels…  so i’m guessing you ran 2 mixed-model ANOVAs and then… so how does the mediational analysis ties into this? (or better, what do you call mediational analysis? is it the same one to specify a path and get the regression coefficients?)

Viewing 4 posts - 1 through 4 (of 4 total)
  • The forum ‘Default Forum’ is closed to new topics and replies.