Resources to interpret Preacher and Hayes (2004) SPSS Macro for Simple Mediation?

[Jaby]

SPSSX List friends-

Have used the wisdom here for sometime, but am having a tough time interpreting the following output from Hayes great add-on to SPSS: 

VARIABLES IN SIMPLE MEDIATION MODEL 

 Y        Com_Eval 

 X        SDO 

 M        ESJ 

 

DESCRIPTIVES STATISTICS AND PEARSON CORRELATIONS 

                 Mean        SD  Com_Eval       SDO       ESJ 

Com_Eval    2.3000     .9232    1.0000     .3830     .4363 

SDO         3.2126     .5155     .3830    1.0000     .4069 

ESJ         3.1935     .4679     .4363     .4069    1.0000 

 

SAMPLE SIZE 

      166 

 

DIRECT AND TOTAL EFFECTS 

                 Coeff      s.e.         t  Sig(two) 

b(YX)       .6858     .1292    5.3092     .0000 

b(MX)       .3693     .0647    5.7050     .0000 

b(YM.X)     .6631     .1474    4.4983     .0000 

b(YX.M)     .4409     .1338    3.2957     .0012 

 

INDIRECT EFFECT AND SIGNIFICANCE USING NORMAL DISTRIBUTION 

           Value      s.e.    LL95CI    UL95CI         Z  Sig(two) 

Effect     .2449     .0700     .1077     .3821    3.4993     .0005

To me, this suggests a significant mediation by between SDO by ESJ to Com_Eval. Am I getting this right? Please confirm and feel free to illuminate! Or if a simple guide to interpretation for this program is available-steer me in the right direction!

Gracias-

Jaby

[Poes, Matthew Joseph]

See my response below.

 

Matthew J Poes

Research Data Specialist

Center for Prevention Research and Development

University of Illinois

510 Devonshire Dr.

Champaign, IL 61820

Phone: [phone]

email: [hidden email]

 

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of New 2it
Sent: Wednesday, December 12, 2012 5:09 PM
To: [hidden email]
Subject: Resources to interpret Preacher and Hayes (2004) SPSS Macro for Simple Mediation?

 

SPSSX List friends-

 

Have used the wisdom here for sometime, but am having a tough time interpreting the following output from Hayes great add-on to SPSS: 

 

VARIABLES IN SIMPLE MEDIATION MODEL 

 Y        Com_Eval 

 X        SDO 

 M        ESJ 

 

DESCRIPTIVES STATISTICS AND PEARSON CORRELATIONS 

                 Mean        SD  Com_Eval       SDO       ESJ 

Com_Eval    2.3000     .9232    1.0000     .3830     .4363 

SDO         3.2126     .5155     .3830    1.0000     .4069 

ESJ         3.1935     .4679     .4363     .4069    1.0000 

 

SAMPLE SIZE 

      166 

 

DIRECT AND TOTAL EFFECTS 

                 Coeff      s.e.         t  Sig(two) 

b(YX)       .6858     .1292    5.3092     .0000 *This is the effect of your predictor X on Y, so it’s the beta coefficient.  For a 1 point increase in X, Y increases by .6858, and this is significant.

b(MX)       .3693     .0647    5.7050     .0000  *This is the effect of your mediator on your predictor.  Remember that you want to establish the relationship between the mediator and both the IV and DV.

b(YM.X)     .6631     .1474    4.4983     .0000  *This is the direct effect of your mediator on the DV, controlling for the effect of your DV.

b(YX.M)     .4409     .1338    3.2957     .0012  *The is the direct effect of your IV on your DV, controlling for your mediator.

 

INDIRECT EFFECT AND SIGNIFICANCE USING NORMAL DISTRIBUTION 

           Value      s.e.    LL95CI    UL95CI         Z  Sig(two) 

Effect     .2449     .0700     .1077     .3821    3.4993     .0005 *This is thus the total indirect effect of your IV on your DV, through your mediator.  This is equal to the last coefficient from above subtracted from the first coefficient above.  This gives you the significance, so again, you can say that the indirect effect is significant.

 

To me, this suggests a significant mediation by between SDO by ESJ to Com_Eval. Am I getting this right? Please confirm and feel free to illuminate! Or if a simple guide to interpretation for this program is available-steer me in the right direction!  The interpretation guide would be the article they wrote.  Please note you are using an older mediation approach, and I would suggest using his bootstrap approach, as the assumption of a normal mediated distribution is not likely true.  Otherwise, yes, you are interpreting this correctly.  The There is a mediated relationship of SDO on Com_Eval through your mediator of ESJ.

 

Gracias-

 

Jaby

[David Marso]

I took a peek at the code in question and my eyes won't stop bleeding!

[Salbod]

Have you tried posting your question to Preacher’s facebook page?

 

https://www.facebook.com/groups/moderation.analysis/

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Poes, Matthew Joseph
Sent: Thursday, December 13, 2012 4:54 PM
To: [hidden email]
Subject: Re: Resources to interpret Preacher and Hayes (2004) SPSS Macro for Simple Mediation?

 

See my response below.

 

Matthew J Poes

Research Data Specialist

Center for Prevention Research and Development

University of Illinois

510 Devonshire Dr.

Champaign, IL 61820

Phone: [phone]

email: [hidden email]

 

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of New 2it
Sent: Wednesday, December 12, 2012 5:09 PM
To: [hidden email]
Subject: Resources to interpret Preacher and Hayes (2004) SPSS Macro for Simple Mediation?

 

SPSSX List friends-

 

Have used the wisdom here for sometime, but am having a tough time interpreting the following output from Hayes great add-on to SPSS: 

 

VARIABLES IN SIMPLE MEDIATION MODEL 

 Y        Com_Eval 

 X        SDO 

 M        ESJ 

 

DESCRIPTIVES STATISTICS AND PEARSON CORRELATIONS 

                 Mean        SD  Com_Eval       SDO       ESJ 

Com_Eval    2.3000     .9232    1.0000     .3830     .4363 

SDO         3.2126     .5155     .3830    1.0000     .4069 

ESJ         3.1935     .4679     .4363     .4069    1.0000 

 

SAMPLE SIZE 

      166 

 

DIRECT AND TOTAL EFFECTS 

                 Coeff      s.e.         t  Sig(two) 

b(YX)       .6858     .1292    5.3092     .0000 *This is the effect of your predictor X on Y, so it’s the beta coefficient.  For a 1 point increase in X, Y increases by .6858, and this is significant.

b(MX)       .3693     .0647    5.7050     .0000  *This is the effect of your mediator on your predictor.  Remember that you want to establish the relationship between the mediator and both the IV and DV.

b(YM.X)     .6631     .1474    4.4983     .0000  *This is the direct effect of your mediator on the DV, controlling for the effect of your DV.

b(YX.M)     .4409     .1338    3.2957     .0012  *The is the direct effect of your IV on your DV, controlling for your mediator.

 

INDIRECT EFFECT AND SIGNIFICANCE USING NORMAL DISTRIBUTION 

           Value      s.e.    LL95CI    UL95CI         Z  Sig(two) 

Effect     .2449     .0700     .1077     .3821    3.4993     .0005 *This is thus the total indirect effect of your IV on your DV, through your mediator.  This is equal to the last coefficient from above subtracted from the first coefficient above.  This gives you the significance, so again, you can say that the indirect effect is significant.

 

To me, this suggests a significant mediation by between SDO by ESJ to Com_Eval. Am I getting this right? Please confirm and feel free to illuminate! Or if a simple guide to interpretation for this program is available-steer me in the right direction!  The interpretation guide would be the article they wrote.  Please note you are using an older mediation approach, and I would suggest using his bootstrap approach, as the assumption of a normal mediated distribution is not likely true.  Otherwise, yes, you are interpreting this correctly.  The There is a mediated relationship of SDO on Com_Eval through your mediator of ESJ.

 

Gracias-

 

Jaby

[Anter]

I could say your model supports partial mediation of ESJ on the relationship between SDO and Com_Eval.

Basically the coefficient of SDO decreases to .45 in the equation with ESJ included as a mediator, but it is still significant and above 0 level (some consider that non-significance, i.e p>.05, in itself is not an indicator of total mediation, but that the coefficient for X must be 0 in the final equation for total mediation, which is not the case here - see for example http://faculty.fuqua.duke.edu/~jglynch/Working%20Papers/Zhao%20Lynch%20Chen%20JCR_Essay_Submission.pdf).

The confidence interval for the mediation does not include 0, which suggests that there is a significant mediated effect on the relationship between SDO and Com_Eval.

Your output shoes normal theory (Sobel) test results as well, which is above the 1.96 level supporting the bootstrap results (significant mediation)

You could try calculating an size effect, for example the indirect effect to the total effect: PM = ab / c, to see how much mediated effect you have.

Have you looked at the BCA's or BC confidence intervals in the output? Are they consistently showing mediation (CIs not including 0)?

---

Andra Toader
E-mail: [hidden email]



--- On Fri, 12/14/12, Salbod, Mr. Stephen <[hidden email]> wrote:

From: Salbod, Mr. Stephen <[hidden email]>
Subject: Re: Resources to interpret Preacher and Hayes (2004) SPSS Macro for Simple Mediation?
To: [hidden email]
Date: Friday, December 14, 2012, 3:15 PM

Have you tried posting your question to Preacher’s facebook page?

 

https://www.facebook.com/groups/moderation.analysis/

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Poes, Matthew Joseph
Sent: Thursday, December 13, 2012 4:54 PM
To: [hidden email]
Subject: Re: Resources to interpret Preacher and Hayes (2004) SPSS Macro for Simple Mediation?

 

See my response below.

 

Matthew J Poes

Research Data Specialist

Center for Prevention Research and Development

University of Illinois

510 Devonshire Dr.

Champaign, IL 61820

Phone: [phone]

email: mpoes@...

 

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of New 2it
Sent: Wednesday, December 12, 2012 5:09 PM
To: [hidden email]
Subject: Resources to interpret Preacher and Hayes (2004) SPSS Macro for Simple Mediation?

 

SPSSX List friends-

 

Have used the wisdom here for sometime, but am having a tough time interpreting the following output from Hayes great add-on to SPSS: 

 

VARIABLES IN SIMPLE MEDIATION MODEL 

 Y        Com_Eval 

 X        SDO 

 M        ESJ 

 

DESCRIPTIVES STATISTICS AND PEARSON CORRELATIONS 

                 Mean        SD  Com_Eval       SDO       ESJ 

Com_Eval    2.3000     .9232    1.0000     .3830     .4363 

SDO         3.2126     .5155     .3830    1.0000     .4069 

ESJ         3.1935     .4679     .4363     .4069    1.0000 

 

SAMPLE SIZE 

      166 

 

DIRECT AND TOTAL EFFECTS 

                 Coeff      s.e.         t  Sig(two) 

b(YX)       .6858     .1292    5.3092     .0000 *This is the effect of your predictor X on Y, so it’s the beta coefficient.  For a 1 point increase in X, Y increases by .6858, and this is significant.

b(MX)       .3693     .0647    5.7050     .0000  *This is the effect of your mediator on your predictor.  Remember that you want to establish the relationship between the mediator and both the IV and DV.

b(YM.X)     .6631     .1474    4.4983     .0000  *This is the direct effect of your mediator on the DV, controlling for the effect of your DV.

b(YX.M)     .4409     .1338    3.2957     .0012  *The is the direct effect of your IV on your DV, controlling for your mediator.

 

INDIRECT EFFECT AND SIGNIFICANCE USING NORMAL DISTRIBUTION 

           Value      s.e.    LL95CI    UL95CI         Z  Sig(two) 

Effect     .2449     .0700     .1077     .3821    3.4993     .0005 *This is thus the total indirect effect of your IV on your DV, through your mediator.  This is equal to the last coefficient from above subtracted from the first coefficient above.  This gives you the significance, so again, you can say that the indirect effect is significant.

 

To me, this suggests a significant mediation by between SDO by ESJ to Com_Eval. Am I getting this right? Please confirm and feel free to illuminate! Or if a simple guide to interpretation for this program is available-steer me in the right direction!  The interpretation guide would be the article they wrote.  Please note you are using an older mediation approach, and I would suggest using his bootstrap approach, as the assumption of a normal mediated distribution is not likely true.  Otherwise, yes, you are interpreting this correctly.  The There is a mediated relationship of SDO on Com_Eval through your mediator of ESJ.

 

Gracias-

 

Jaby