*4.4. Utility*

Most of the ONSO and the OFFSO respondents (84.1% and 77.7%) considered the exercises to be useful during actual emergency work (ONSOs M = 4.45, SD = 0.90; OFFSOs M = 4.20, SD = 1.11). Furthermore, they regarded the inter-organisational exercises as having an impact on their everyday work (ONSOs 61.7%, M = 3.69, SD = 1.16; OFFSOs 44.0% M = 3.26, SD = 1.22). It should be noted here that 27.4% of the OFFOS remained neutral. The exercises were considered to be more valuable for the command officers (ONSOs 58.5%, M = 3.54, SD = 1.30; OFFSOs 50.4%, M = 3.68, SD = 1.18) than for the operative staff in the field (ONSOs 30.8%, M = 2.97, SD = 1.22; OFFSOs 29.3%, M = 3.21, SD = 1.23). The mean for all items within the utility dimension was 3.66 (SD = 1.08) for the ONSOs and 3.59 (SD = 0.71) for the OFFSOs (Figure 1).

**Figure 1.** Mean values for the 94 ONSO and 252 OFFSO emergency personnel answering the collaboration, learning and utility instrument (CLU) scale, distributed in the dimensions utility (four items), learning (five items) and collaboration (eight items).

## *4.5. Bivariate Regressions*

The causal effects of collaboration, learning and utility in the ONSO and OFFSO contexts were tested in a number of bivariate regressions. The collaborative dimension of the exercises was significantly correlated to the mean learning score across most of the items associated with the learning measurements in both exercise contexts. The strongest significant correlation on ONSOs was found between the item 'well-known activities during the exercise' and learning (R = 0.48), with this item explaining a significant proportion of variance in the mean learning score (R<sup>2</sup> = 0.23, F = 27.67, *p* < 0.00). The strongest significant correlation on OFFSOs was the item 'my point of view was regarded' and learning (R = 0.40), with this item explaining a significant proportion of variance in the mean learning score (R<sup>2</sup> = 0.16, F = 48.86, *p* < 0.00) (Table 2).


**Table 2.** Bivariate regression of items in the collaborative dimension of learning (sig. = *p* < 0.05).

Some of the items in the learning dimension were significantly correlated to the usefulness items. The strongest significant correlation of ONSOs was found between the item 'learnt new things' and usefulness (R = 0.47, R<sup>2</sup> = 0.22, F = 26.35, *p* < 0.00). A somewhat weaker correlation was found for the OFFSOs for the same item (R = 0.35, R<sup>2</sup> = 0.11, F = 31.99, *p* < 0.00) (Table 3).


**Table 3.** Bivariate regression of items in the learning dimension of utility (sig. = *p* < 0.05).

#### *4.6. Multiple Regressions*

Significant variables from the bivariate regressions were tested separately for the ONSOs and the OFFSOs in multiple regressions. In the case of ONSOs, the collaborative features together predicted 53% (R<sup>2</sup> = 0.53) of learning and for the OFFSOs, they predicted 25% (R<sup>2</sup> = 0.25) of the variance in learning. This meant that 47% and 75% of the predicted variance was non-unaccounted for in the case of ONSOs and OFFSOs, respectively. In the case of ONSOs, four variables were still significant: 'opportunities to improvise', 'well-known activities during the exercise', 'clear instructions for collaborative practices during the exercises' and 'my points of view were regarded'. The OFFSOs showed one still significant variable: 'my points of view were regarded'. The remaining variables displayed somewhat lower t-values and lacked significance on their own (Table 4).

**Table 4.** Significant variables of a multiple regression of items in the collaboration dimension of learning, (sig. = *p* < 0.05).


In the next multiple regression, it was found that the items of learning predicted 26% (R<sup>2</sup> = 0.26) of the variance in usefulness in the ONSO and the OFFSO context. This meant that 74% of the predicted variance was still missing in the regressions. One variable was still significant, 'learnt new things' in the ONSO and the OFFSO contexts and 'learnt new concepts' in the OFFSO context. The other variables displayed moderate t-values and lacked significance on their own (Table 5).

**Table 5.** Significant variables of a multiple regression of items in the learning dimension of usefulness (sig. = *p* < 0.05).

