Mediation question

#1
Hello,
Perhaps you can help me. I'm wondering how to interpret this output (pasted below). I am trying to determine if "AVG_PSES" has a mediating effect on the relationship between "grade span" and "TOT_GEN". "Grade span" is a dichotomous variable, where 0=elementary and 1=secondary".

When I compare the betas in the "total effect of x on y" and the "direct effect of x on y", it appears that the beta increases when going from total effect to direct effect, since the sign is negative in "total effect" and positive in "direct effect", indicating that there is NO mediation. But am I supposed to compare the absolute value of the betas instead of the real value? If that's the case, then the beta decreased, and there IS mediation.


Model : 4
Y : TOT_GEN_
X : GRADE_SP
M : AVG_PSES

Sample
Size: 715

**************************************************************************
OUTCOME VARIABLE:
AVG_PSES

Model Summary
R R-sq MSE F df1 df2 p
.1451 .0211 .9137 15.3448 1.0000 713.0000 .0001

Model
coeff se t p LLCI ULCI
constant 6.8017 .0413 164.5819 .0000 6.7205 6.8828
GRADE_SP -.3226 .0824 -3.9172 .0001 -.4844 -.1609

**************************************************************************
OUTCOME VARIABLE:
TOT_GEN_

Model Summary
R R-sq MSE F df1 df2 p
.6329 .4005 86.1646 237.8299 2.0000 712.0000 .0000

Model
coeff se t p LLCI ULCI
constant 23.8523 2.5059 9.5184 .0000 18.9324 28.7722
GRADE_SP .1998 .8084 .2472 .8048 -1.3873 1.7870
AVG_PSES 7.8601 .3637 21.6132 .0000 7.1461 8.5741

************************** TOTAL EFFECT MODEL ****************************
OUTCOME VARIABLE:
TOT_GEN_

Model Summary
R R-sq MSE F df1 df2 p
.0848 .0072 142.4953 5.1589 1.0000 713.0000 .0234

Model
coeff se t p LLCI ULCI
constant 77.3140 .5161 149.8079 .0000 76.3008 78.3273
GRADE_SP -2.3362 1.0286 -2.2713 .0234 -4.3557 -.3168

************** TOTAL, DIRECT, AND INDIRECT EFFECTS OF X ON Y **************

Total effect of X on Y
Effect se t p LLCI ULCI c_ps
-2.3362 1.0286 -2.2713 .0234 -4.3557 -.3168 -.1951

Direct effect of X on Y
Effect se t p LLCI ULCI c'_ps
.1998 .8084 .2472 .8048 -1.3873 1.7870 .0167

Indirect effect(s) of X on Y:
Effect BootSE BootLLCI BootULCI
AVG_PSES -2.5361 .6561 -3.9469 -1.2737

Partially standardized indirect effect(s) of X on Y:
Effect BootSE BootLLCI BootULCI
AVG_PSES -.2118 .0536 -.3222 -.1093
 
#2
You do not have a significant direct effect (LLCI and ULCI cross 0), however there is a significant indirect effect (LLCI and ULCI do not cross 0).

I would encourage you to re-run the same analysis but under 'options' choose to obtain the Sobel Test. Sobel's test is a specialized test that will tell you if there is a statistically-significant mediation.

I am assuming you are using PROCESS for this? Which version of PROCESS are you using? The newest version was released within the past year or so, and I believe some of the models have changed to allow a better integration of dichotomous mediators.

Hope this helps.