Hypothesis 6a predicted that improvements in executive function would mediate the benefits of mindfulness practice on both metacognitive processing and psychological wellbeing. In order to test this, a number of mediating relationships were examined to ascertain whether they met the criteria of Baron & Kenny (1986; see Method section). Table 5 indicates which of the criteria each hypothesised relationship met.
Criterion | |||||
---|---|---|---|---|---|
Proposed Mediating Relationship (IVMVDV) | Correlation between IV and DV | Correlation between MV and IV | Correlation between MV and DV | Partial correlation between IV and DV when MV is controlled for in linear regression | Met all criteria? |
GroupDSBBDI | .45** | -.41** | -.18 | .42 * | |
GroupDSBBAI | .05 | -.41** | -.24T | .-.06 * | |
GroupDSBPANAS+ | -.03 | -.41** | .22T | .06 | |
GroupDSBPANAS- | .32* | -.41** | -.09 | .31 | |
GroupDSBRRS | .23T | -.41** | -.10 | .21 | |
GroupDSBMAAS | -.5*** | -.41** | .31* | -.43 ** | yes |
GroupNeutral Switch Task BDI | .45** | .34* | .41** | .34 * | yes |
GroupNeutral Switch Task BAI | .05 | .34* | -.05 | .07 | |
GroupNeutral Switch Task PANAS+ | -.03 | .34* | -.07 | .002 | |
GroupNeutral Switch Task PANAS- | .32* | .34* | .09 | .32* | |
GroupNeutral Switch Task RRS | .23T | .34* | .09 | .23 | |
GroupNeutral Switch Task MAAS | .45** | .34* | -.08 | -.54*** | |
GroupAffective Switch Task BDI | .05 | .20 | .27T | .41** | |
Group Affective Switch Task BAI | -.03 | .20 | -.07 | .06 | |
Group Affective Switch Task PANAS+ | .32* | .20 | .10 | -.05 | |
Group Affective Switch Task PANAS- | .23T | .20 | .11 | .31* | |
Group Affective Switch Task RRS | -.5*** | .20 | .06 | .24 | |
Group Affective Switch Task MAAS | .001 | .20 | -.21 | -.51*** | |
GroupOverall Switch Task BDI | .002 | .32* | .39* | .36* | yes |
Group Overall Switch Task BAI | .39 | .32* | -.05 | .07 | |
Group Overall Switch Task PANAS+ | .42 | .32* | .04 | -.04 | |
Group Overall Switch Task PANAS- | .02 | .32* | .12 | .31 | |
Group Overall Switch Task RRS | .07 | .32* | .08 | .24 | |
Group Overall Switch Task MAAS | .001 | .32* | -.15 | -.51 |
Note: Values provided Pearson r-values of a bivariate correlation between residual measures of affect,
metacognitive processing, and executive function, except for column 4, which represents Pearson r-values
of a partial correlation between the same measures. All four r-values must be significant for valid calculation
of a mediating relationship.
*p < .05 **p < .01 ***p < .001 Tp < .1 (trend).
As can be seen from Table 5, all conditions were satisfied for the calculation of three mediation tests, each of which demonstrated: (1) a significant correlation between the IV and DV; (2) a significant correlation between the MV and IV; (3) a significant correlation between the MV and DV; and (4) a reduced correlation between the IV and DV when the MV is controlled for. The following are results of Sobel tests (reporting Goodman I values) for each of these mediating relationships:
1. Increased scores on the DSB mediated the relationship between group membership and improved scores on the MAAS (z = 2.06, p = .04).
2. There was a strong trend towards decreased RTs on the neutral condition of the IST mediating the relationship between group membership and improved scores on the BDI (z = 1.77, p = .08).
3. Decreased overall RTs on the IST (collapsed across neutral and affective) were not found to mediate the relationship between group membership and improved scores on the BDI (z = 1.54, p = .12).
Note that the statistics provided by Sobel tests are estimates, and so should be interpreted with caution (Preacher & Leonardelli, 2001).