Exploratory Testing of Treatment Effects on Moral Injury
Bivariate correlations and descriptive statistics of all study variables across time points are presented in Table 2. There were no statistically significant differences on demographic characteristics (except for age, participants were significantly older in the MMMI condition [M= 45.81 years; SD = 10.51] than the ES condition [M = 39.26 years; SD = 8.57]) and baseline scores on all outcomes across treatment conditions, thus ensuring the two analytic treatment groups (MMMI and ES) were largely similar. All results of our repeated measures ANCOVA models are found in Table 3.
Exploratory testing of effects on MMMI on moral injury outcomes. We generally found medium-large effect sizes (.08 < η2 > .26). Regarding those that were statistically significant, we found significant decreases in EMIS-M other-directed (betrayal-based) moral injury collapsing across conditions (η2 = .195); however, this finding was qualified by a significant interaction (p = .023; η2 = .129), such that there was a larger decrease in the MMMI condition (T1: M = 28.57, SE = 1.53; T2: M = 24.71, SE = 1.41; M difference = -3.86, SE = 0.98, p < .001) compared to the ES condition (T1:M = 28.37, SE = 1.61; T2: M = 27.90, SE = 1.48; Mdifference = -0.47, SE = 1.04, p = .650). While no significant interaction was found for EMIS-M self-directed scores (p = .071, η2 = .083), significant decreases in EMIS-M self-directed scores emerged when collapsing across the MMMI and ES conditions (η2 = .100; T1:M = 25.17, SE = 1.29; T2: M = 23.04, SE = 1.10; Mdifference = -2.13, SE = 1.03, p = .046). Similarly, while no significant interaction was found for MISS scores (p = .061, η2 = .089), we found significant decreases in MISS total scores collapsing across the MMMI and ES conditions (η2 = .216; T1: M = 54.83, SE = 2.35; T2: M = 48.69, SE = 2.01; M difference = -6.14, SE = 1.90, p = .003).
Regarding the MIOS measure, we found significant decreases in MIOS total scores collapsing across the MMMI and ES conditions (η2 = .132); however, this outcome was also qualified by a significant interaction (p = .028; η2 = .121), such that there was a larger decrease in the MMMI condition (T1: M = 26.86, SE = 2.15; T2: M = 20.29, SE = 2.10; M difference = -6.57, SE = 1.93, p = .002) compared to the ES condition (T1: M = 28.68, SE = 2.26; T2:M = 28.53, SE = 2.21; M difference = -0.16, SE = 2.03,p = .938). We also found a significant interaction for MIOS – shame-related experiences (p = .042; η2 = .105), such that there was a significant decrease in the MMMI condition (T1: M = 12.24, SE = 1.41; T2: M = 9.57, SE = 1.29; M difference = -2.67, SE = 1.18, p = .030) compared to a marginal increase in the ES condition (T1: M = 12.84, SE = 1.48; T2: M = 13.79, SE = 1.36; M difference = 0.95, SE = 1.24, p = .451).
While no significant interaction was found for MIOS – trust-violation-related scores (p = .050, η2 = .097), significant decreases in MIOS – trust-violation-related scores emerged when collapsing across the MMMI and ES conditions (η2 = .256; T1: M = 15.23, SE = 0.82; T2: M = 12.73, SE = 0.79; M difference = -2.51, SE = 0.69, p = <.001). Finally, we also found significant decreases in MIOS impaired functioning due to moral injury scores collapsing across conditions (η2 = .167); however, this was qualified by a significant interaction (p = .023; η2 = .129), such that there was a larger decrease in the MMMI condition (T1:M = 42.94, SE = 3.74; T2: M = 28.96, SE = 3.53; Mdifference = -13.98, SE = 3.76, p < .001) compared to the ES condition (T1: M = 40.45, SE = 3.74; T2: M = 39.40, SE = 3.72; M difference = -1.05, SE = 3.95, p = .791).