Online supplemental material

Mediation analyses uncontrolled for symptom levels. Summaries of findings on four mediation analyses uncontrolled for symptom levels, but controlled for loss-related variables that significantly predicted symptom levels after 12 months (i.e., relationship with the deceased and expectedness of the loss), can be found in Table 1 and 2. All four total effects (c paths) from each of the independent variables (grief rumination and trait rumination) on each of the dependent variables (complicated grief and depressive symptoms) were significant. Consequently, it was investigated if these effects could be accounted for by the indirect effects of various mediators, namely thought suppression, experiential avoidance, behavioral avoidance and loss-reality avoidance. All avoidance strategies were entered at the same time in the mediation analyses, so that each indirect effect was corrected for every other indirect effect.

As shown in Table 1, the relationship between grief rumination and complicated grief severity is partly mediated by experiential avoidance and behavioral avoidance. The same avoidance processes fully mediated the relationship between grief rumination and depressive symptoms. In contrast to our predictions, thought suppression and loss-reality avoidance did not significantly mediate these relationships. While grief rumination did significantly predict thought suppression and anxious avoidance (a paths), these avoidance processes failed to predict depressive and complicated grief symptoms (b paths).

Table 2 shows that the results for rumination assessed with the RRQ were highly similar. The link between trait rumination on the one hand and complicated grief and depressive symptoms on the other hand was fully mediated by experiential and behavioral avoidance. Although thought suppression was a significant mediator of the relationship between trait rumination and complicated grief symptoms, it did not mediate the link between grief rumination and depressive symptoms. Loss-reality avoidance was no significant predictor of symptoms of psychopathology and therefore did not mediate these relationships.

Magnitude of mediation effects. Using partial correlation analyses we examined the relative size of the indirect effects. While controlling for the relationship with the deceased and the expectedness of the death, grief rumination explained 39.8% of the variance in complicated grief symptoms, 24.8% (62.3%) of which was accounted for by avoidance processes. Grief rumination also explained 19.3% of the variance in depressive symptoms, 17.6% (91.2%) of which was accounted for by avoidance processes.

Rumination measured with the RRQ explained smaller portions of variance than grief rumination in complicated grief and depressive symptoms, 10.7% and 8.6%, respectively. However, the relative sizes of mediation effects were larger. 10.4% of the 10.7% (97.2%) of variance rumination explained in symptoms of complicated grief was accounted for by avoidance processes. Similarly, 8.3% of the 8.6% (96.5%) variance rumination explained in depressive symptoms was accounted for by avoidance processes.

Another method to assess the effect sizes of indirect effects was proposed by MacKinnon et al. (2007), who use the formula [1-c’/c] to calculate the proportion of the effect of an IV on a DV that is accounted for by the mediators. Using this rule (possible values lie between 0 and 1) the effect sizes of avoidance processes mediating the link between grief rumination on the one hand and symptoms of complicated grief and depression on the other hand were .47 and .75, respectively. The mediation effect of avoidance processes on the relationship between trait rumination on the one hand and complicated grief symptoms and depressive symptoms on the other hand yield the following effect sizes: .82 and .86.

Phase 1: Post-hoc mediation analyses not controlling for symptom levels.

As mentioned above, experiential avoidance (and behavioral avoidance), but not thought suppression or loss-reality avoidance, mediated the proposed relationships between ruminative thinking and symptoms of psychopathology if all avoidance processes were entered simultaneously in the multiple mediation analyses. However, experiential avoidance, broadly defined as avoidance of internal experiences (Hayes at al., 2004), potentially overlaps with avoidance of thoughts as assessed by the WBSI and the loss-reality avoidance as measured with the DAAPGQ, which primarily measures avoidance of thought content related to the loss-reality. Therefore, in order to further investigate the importance of these avoidance processes all four mediation models were rerun using only thought suppression, loss-reality avoidance and behavioral avoidance as mediator variables.

These revised mediation models (see Table 3 and 4) show thought suppression and behavioral avoidance partly mediated the relationship between ruminative thinking and complicated grief symptom severity. However, the link between grief rumination and symptoms of depression was exclusively (but partly) mediated by behavioral avoidance. In similar vein, behavioral avoidance fully explained the relationship between rumination and depressive symptom severity. Again, loss-reality avoidance failed to predict symptoms of complicated grief and depression and therefore was not a significant mediator in any of the proposed mediation models.

Magnitude of post-hoc mediation effects. Again, relative effect sizes were computed using partial correlation analyses. The total portion of variance explained by the avoidance processes was somewhat reduced, because of the exclusion of experiential avoidance as a mediator variable. While controlling for relevant background variables, grief rumination explained 39.8% of the variance in complicated grief symptoms, 21.7% (54.5%) of which was accounted for by avoidance processes. Grief rumination also explained 19.3% of the variance in depressive symptoms, 16.1% (83.4%) of which was accounted for by avoidance processes.

Although trait rumination explained less variance in complicated grief and depressive symptoms than grief rumination, the mediation effects of avoidance processes were more outspoken for this type of ruminative thinking. Of the 10.7% of the variance in complicated grief symptoms explained by trait rumination 7.7% (72.0%) could be attributed to avoidance processes. Of the 8.6% of the variance in depressive symptoms explained by trait rumination 7.6% (88.4%) could be attributed to avoidance processes.

Using the method of MacKinnon and colleagues (2007), the effect sizes of avoidance processes mediating the link between grief rumination on the one hand and symptoms of complicated grief and depression on the other hand were .41 and .67, respectively. The mediation effects of avoidance processes on the relationship between trait rumination on the one hand and complicated grief symptoms and depressive symptoms on the other hand were qualified by the following effect sizes: .67 and .73.


Table 1

Summary of the mediation analysis uncontrolled for symptom levels in which avoidance processes mediate the relationship between grief rumination and symptoms of complicated grief (Model 1) and depression (Model 2).

Model / Mediator / Total
Effect
(c) / Direct
Effect
(c’) / Total indirect effect
(∑a x b) / Unique
indirect
effect
(a x b) / 95% CI
1 / 1.02* / 0.54* / 0.48* / 0.34-0.63
Experiential avoidance / 0.17* / 0.05-0.34
Thought suppression / 0.09 / -0.02-0.21
Behavioral avoidance / 0.20* / 0.09-0.35
Loss-reality avoidance / 0.01 / -0.05-0.09
2 / 0.18* / 0.04 / 0.14* / 0.09-0.19
Experiential avoidance / 0.04* / 0.00-0.09
Thought suppression / 0.00 / -0.03-0.02
Behavioral avoidance / 0.10* / 0.06-0.15
Loss-reality avoidance / 0.00 / -0.02-0.01

Note: * = significant at p < .05.


Table 2

Summary of the mediation analyses uncontrolled for symptom levels in which avoidance processes mediate the relationship between trait rumination and symptoms of complicated grief (Model 1) and depression (Model 2).

Model / Mediator / Total
Effect
(c) / Direct
Effect
(c’) / Total indirect effect
(∑a x b) / Unique
indirect
effect
(a x b) / 95% CI
1 / 0.71* / 0.13 / 0.58* / 0.39-0.79
Experiential avoidance / 0.23* / 0.10-0.41
Thought suppression / 0.14* / 0.01-0.29
Behavioral avoidance / 0.18* / 0.09-0.33
Loss-reality avoidance / 0.02 / -0.05-0.10
2 / 0.15* / 0.02 / 0.13* / 0.07-0.18
Experiential avoidance / 0.04* / 0.01-0.09
Thought suppression / 0.00 / -0.04-0.03
Behavioral avoidance / 0.09* / 0.05-0.13
Loss-reality avoidance / 0.00 / -0.02-0.02

Note: * = significant at p < .05.


Table 3

Summary of the post-hoc mediation analyses uncontrolled for symptom levels in which avoidance processes mediate the relationship between grief rumination and symptoms of complicated grief (Model 1) and depression (Model 2).

Model / Mediator / Total
Effect
(c) / Direct
Effect
(c’) / Total indirect effect
(∑a x b) / Unique
indirect
effect
(a x b) / 95% CI
1 / 1.04* / 0.61* / 0.43* / 0.31-0.57
Thought suppression / 0.15* / 0.06-0.27
Behavioral avoidance / 0.27* / 0.16-0.43
Loss-reality avoidance / -0.01 / -0.08-0.07
2 / 0.19* / 0.06* / 0.13* / 0.08-0.18
Thought suppression / 0.01 / -0.01-0.04
Behavioral avoidance / 0.12* / 0.08-0.17
Loss-reality avoidance / -0.01 / -0.03-0.01

Note: * = significant at p < .05.


Table 4

Summary of the post-hoc mediation analyses uncontrolled for symptom levels in which avoidance processes mediate the relationship between trait rumination and symptoms of complicated grief (Model 1) and depression (Model 2).

Model / Mediator / Total
Effect
(c) / Direct
Effect
(c’) / Total indirect effect
(∑a x b) / Unique
indirect
effect
(a x b) / 95% CI
1 / 0.69* / 0.23* / 0.46* / 0.28-0.65
Thought suppression / 0.21* / 0.10-0.35
Behavioral avoidance / 0.25* / 0.13-0.41
Loss-reality avoidance / 0.00 / -0.06-0.07
2 / 0.14* / 0.04 / 0.10* / 0.05-0.16
Thought suppression / 0.01 / -0.01-0.04
Behavioral avoidance / 0.09* / 0.04-0.14
Loss-reality avoidance / 0.00 / -0.02-0.01

Note: * = significant at p < .05.