Multiple Mediation using 3 Mediators

[Rachel_1983]

Hi,

I wonder can you help me out, I think this is a reasonably easy question but I am a novice in the area of multiple mediation and I haven't been able to figure this problem out.

I have three hypothesised mediators M1, M2, and M3 which I argue will mediate the relationship between two independent variables and my dependent variable (N=192). I ran simple mediation models for each of these mediators (without controlling for the effects of the other mediators) and found that M1 fully mediated the effect of IV1 and IV2 on DV. M2 fully mediated the effect of IV1 on DV and partially mediated the effect of IV2 on DV. For M3 - no mediation was found.

When I enter the three mediators into Preacher and Hayes' (2008) INDIRECT, I found that both M1 and M2 mediated the link between IV1 and DV whilst as expected no mediation was found for M3. For IV2 no mediation effect was found for any of the mediators. I am okay with making theoretical conclusions on this findings. However, given that I found full mediation for M1 and partial mediation for M2 on the effect of IV2 on DV when I ran simple mediation models should I report this information also? Or just report the multiple mediation findings using INDIRECT. I would like to show the mediation effect of M1 and M2 for IV2 however is this poor practice... I'm not sure.

Also when I tested the simple mediation for M3 I initially decided to include only M1 and M2 in INDIRECT, however this caused the bootstrap result for M2 to reduce to insignificance. Should I include all three hypothesised mediators on INDIRECT or just the two that had significant mediation in the simple mediation tests?

Best Wishes and Thanks,
Rachel.

[Swank, Paul R]

I would argue to include all three since leaving one out bias other parameter estimates in the model.

Sent from my iPad

On Jun 1, 2012, at 8:19 AM, "Rachel_1983" <[hidden email]> wrote:

> Hi,
>
> I wonder can you help me out, I think this is a reasonably easy question but
> I am a novice in the area of multiple mediation and I haven't been able to
> figure this problem out.
>
> I have three hypothesised mediators M1, M2, and M3 which I argue will
> mediate the relationship between two independent variables and my dependent
> variable (N=192). I ran simple mediation models for each of these mediators
> (without controlling for the effects of the other mediators) and found that
> M1 fully mediated the effect of IV1 and IV2 on DV. M2 fully mediated the
> effect of IV1 on DV and partially mediated the effect of IV2 on DV. For M3 -
> no mediation was found.
>
> When I enter the three mediators into Preacher and Hayes' (2008) INDIRECT, I
> found that both M1 and M2 mediated the link between IV1 and DV whilst as
> expected no mediation was found for M3. For IV2 no mediation effect was
> found for any of the mediators. I am okay with making theoretical
> conclusions on this findings. However, given that I found full mediation for
> M1 and partial mediation for M2 on the effect of IV2 on DV when I ran simple
> mediation models should I report this information also? Or just report the
> multiple mediation findings using INDIRECT. I would like to show the
> mediation effect of M1 and M2 for IV2 however is this poor practice... I'm
> not sure.
>
> Also when I tested the simple mediation for M3 I initially decided to
> include only M1 and M2 in INDIRECT, however this caused the bootstrap result
> for M2 to reduce to insignificance. Should I include all three hypothesised
> mediators on INDIRECT or just the two that had significant mediation in the
> simple mediation tests?
>
> Best Wishes and Thanks,
> Rachel.
>
>
> --
> View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Multiple-Mediation-using-3-Mediators-tp5713461.html
> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD