|
The short answer is that if
you are using a method such as bootstrapping to
test the Indirect Effect, then you need to only find a significant ab-path.
If you are using a method like The Test of Joint Significance (TJS), then
you would look to see if both a and b are different from 0.
In your example - The test of the indirect effect is under "Total,
Direct,
and Indirect Effects". The CI for the Indirect Effect is [.0009,.1240]
which doesn't include zero, so you would conclude that it is statistically
significant (using bootstrapping criterion). So, you could have an
"a" or
"b" path that is non-significant, but the "ab" path is
significant.
However, the Total Effect is non-significant, indicating that there isn't an
effect to mediate, to begin with. Do you expect a Total Effect (should X
predict Y)? The Direct and Indirect Effects have the same sign, so you
aren't dealing with "Inconsistent Mediation" (a suppression effect).
So,
you could talk about having an Indirect Effect but not Mediation. Also,
you
should look at the size of the Indirect Effect. It seems small - is
this an
important effect or is it trivial?
pl
-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
psy_vm
Sent: Thursday, July 24, 2014 12:17 AM
To: [hidden email]
Subject: [SPSSX-L] Interpreting Mediation results
Hello,
I have a few questions regarding interpretation of my results for Mediation
using Hayes' Process macro. The key point for finding significance is to see
that the Class Interval does not include zero. However, in my mediation
model, although the indirect effect CI doesn't include zero, the initial
Models's CI (for bcope_ma, under Duration) includes zero. Should I still
count my indirect effects as significant even though initial model for
mediation is not significant?
Results pasted below. Please Help! Thanks in advance!
Model = 4
Y = qoltot
X = duration
M = bcope_ma
Sample size
70
**************************************************************************
Outcome: bcope_ma
Model Summary
R R-sq
F df1
df2 p
.1830 .0335 2.3550
1.0000 68.0000 .1295
Model
coeff
se
t p LLCI
ULCI
constant 22.5169 1.3151 17.1223
.0000 19.8927 25.1411
duration -.0388 .0253 -1.5346
.1295 -.0893 .0117
**************************************************************************
Outcome: qoltot
Model Summary
R R-sq
F df1
df2 p
.3234 .1046 3.9140
2.0000 67.0000 .0247
Model
coeff
se t
p LLCI
ULCI
constant 91.2951 9.5536
9.5560 .0000 72.2258 110.3643
bcope_ma -1.0460 .3823 -2.7364
.0079 -1.8091 -.2830
duration .0059 .0811
.0725 .9425 -.1559
.1677
************************** TOTAL EFFECT MODEL ****************************
Outcome: qoltot
Model Summary
R R-sq
F df1
df2 p
.0674 .0045
.3102 1.0000 68.0000
.5794
Model
coeff
se t
p LLCI
ULCI
constant 67.7413 4.3387 15.6134
.0000 59.0836 76.3990
duration .0465 .0834
.5570 .5794 -.1200
.2129
***************** TOTAL, DIRECT, AND INDIRECT EFFECTS ********************
Total effect of X on Y
Effect SE
t p
LLCI ULCI
.0465 .0834
.5570 .5794 -.1200
.2129
Direct effect of X on Y
Effect SE
t p
LLCI ULCI
.0059 .0811
.0725 .9425 -.1559
.1677
Indirect effect of X on Y
Effect
Boot SE BootLLCI BootULCI
bcope_ma .0406 .0311
.0009 .1240
Partially standardized indirect effect of X on Y
Effect
Boot SE BootLLCI BootULCI
bcope_ma .0022 .0016
.0000 .0068
Completely standardized indirect effect of X on Y
Effect
Boot SE BootLLCI BootULCI
bcope_ma .0589 .0411
.0017 .1650
Ratio of indirect to total effect of X on Y
Effect Boot
SE BootLLCI BootULCI
bcope_ma .8736 149.0933 .1603
53.5001
Ratio of indirect to direct effect of X on Y
Effect
Boot SE BootLLCI BootULCI
bcope_ma 6.9093 17.9156 3.4423
275.8767
R-squared mediation effect size (R-sq_med)
Effect
Boot SE BootLLCI BootULCI
bcope_ma .0045 .0173
-.0184 .0543
Preacher and Kelley (2011) Kappa-squared
Effect Boot SE
BootLLCI BootULCI
bcope_ma .0601 .0401
.0066 .1692
Normal theory tests for indirect effect
Effect se
Z p
.0406 .0318 1.2753
.2022
******************** ANALYSIS NOTES AND WARNINGS *************************
Number of bootstrap samples for bias corrected bootstrap confidence
intervals: 1000
Level of confidence for all confidence intervals in output:
95.00
NOTE: Some cases were deleted due to missing data. The number of such
cases
was: 33
--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Interpreting-Mediation-results
-tp5726809.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
|