**3. Results**

Using the *COM* variable, we performed a series of tests for different groups of respondents at the 0.05 level of confidence. We found a statistically significant *COM* increase of 12% among those who had participated in some energy-saving initiatives (*COM* = 0.64) versus the other group (*COM* = 0.57), see Table 3. It is important to note that the number of those who had participated in energy-saving initiatives is relatively low and does not differ between the years.



Similarly, respondents who reported energy-saving behavior had a statistically significant 9% higher *COM* (average *COM* = 0.59) than their colleagues who do not exhibit energy-saving behavior (average *COM* = 0.53). The results stand well with the COM-B theory where higher capabilities, opportunities and motivation influence a better behavior.

To determine the context of COM-B we measured how an overall satisfaction with the military service is influencing *COM*. The satisfaction (see Figure 4) measured from 1 (lowest) to 10 (highest) has a 0.49 correlation coefficient with the *COM* variable (at the 0.01 significance level). This is not surprising as we expected the satisfaction to be related with the motivation variable.

**Figure 4.** Interdependency between overall satisfaction with the service and the *COM* variable (Capability-Opportunity-Motivation) of: (**a**) professional soldiers, (**b**) conscripts.

Considering the uniqueness of our sample that consists of professional soldiers and conscripts we checked whether there are any differences between these two groups of respondents one year apart. There was no statistically significant difference in average *COM* value indicating independent groups of responders as a large part of the professional soldiers was rotated and all the conscripts were replaced over that year. Meanwhile, the *T*-test of independent samples suggested a statistically significant although quite small difference in energy-saving behavior between the professional soldiers and conscripts: the average *COM* value of professional soldiers (0.60) exceeded the corresponding value of conscripts (0.57) by 5%.

In testing our main statement that *COM* is positively linked not only with individual soldier's behavior, but also with collective energy behavior of the unit, we measured the impact of the behavioral factors on an individual's behavior ("my" behavior) in relation to collective energy behavior of the unit ("our" behavior). For this reason, we split the respondents into 6 different subgroups based on their responses about their own behavior and the collective energy behavior of the unit. The groups and their corresponding *COM* values are presented in Table 4. The first column shows the energy-saving behavior of the unit as reported by a respondent: a plus sign is given when the respondent positively evaluated the collective energy behavior, a minus for the negatively reported behavior, and a zero when the respondent reported a neutral behavior. The second column has two values depending on whether the respondent herself/himself exhibits energy-saving behavior (either a plus or a minus). One-way analysis of variance (ANOVA) showed a significant difference in mean *COM* values between all groups. The results show a much stronger effect of *COM* value on "our" behavior than on "my" behavior.

Individually, each of the behavior variables positively correlates with the *COM* variable—the correlation coefficient is 0.45 for "our" behavior and 0.16 for "my" behavior. This can be observed in Figure 5, where the behavior becomes more positive as the *COM* variable grows.


**Table 4.** *COM* values with "our" and "my" energy-saving behavior.

1 The mean difference is significant at the 0.05 level.

**Figure 5.** Capability-Opportunity-Motivation variable (*COM*) versus the self-reported energy-saving behavior of an individual (white hatched boxes—"my" behavior) and a group (grey boxes —"our" behavior).

## **4. Discussion**

This study fills the gap in the research on military energy behavior and suggests that pro-environmental energy behavior in the military is linked not only with an individual soldier's attitudes and intentions, but also with its military unit. Institutionally favorable environment that forces military transition towards the pro-environmental energy behavior is reflected as a collective action. So far, many studies focused on technical solutions for reducing energy usage [33,39,40] and enabling decision makers [5,14,41] and individual soldiers [10,24] to save energy. It appears that this research for the first-time underlines collectivity and connects individual and collective pro-environmental energy behavior into one measurable construct.

Pro-environmental energy behavior in the military should be considered in a special way as military culture penetrates attitudes and behavior [26,28]. Respectively, our conceptual construct that includes a collective energy behavior goes in line with other studies designed for communality in the military. Transformation from "I" to "We" is unique to the military as it serves as a protective psychological construct in difficult exercises and war situations [35]. According to our findings, in the case of pro-environmental energy behavior, this transformation makes a positive impact on both the collective and the individual behavior. These findings go in line with the findings where macro influences on energy behavior are discussed. According to the comprehensive review of studies about energy behavior in organizations, energy behavior is driven by broad social and institutional considerations [7,42]. In response to this statement, our research findings indicate that in the military unit communality acts as the factor of sustaining pro-environmental behavior, i.e., when the change factors decrease, the unit still exhibits a positive behavior even if individual's pro-environmental energy behavior might falter.

The segregation of collective and individual behavior is a fundamentally new approach in COM-B model testing. These results can be explained by the fact that the COM-B model was not tested before in organizations or communities where communality (strong sense of belonging) is more important than individual contribution. At the same time, the results are consistent with research results from various fields in the military and shows that a collective identity becomes more important than an individual soldier's identity [1,27,43]. For example, the collective identity was found dominating in the research on military innovations [27], where a unique military culture was found manifesting. Similarly, studies on leadership in the military [44] are predominately concerned about the function of military leaders to create subordinates' emotional attraction to the organization which in turn increases commitment and resilience of soldiers.

Despite the segregation of the behavioral element, energy behavior in the military unit follows the classical COM-B model. Just like other researchers, we adopted the construct to the context of the organization. Thus, the results of this study resonate with the evidence of other studies using the COM-B model: the model is constructed with three statistically reliable indicators: capabilities, opportunities and motivation [11,21,45]. In addition, the overall variance of the variables compares to the other energy behavioral models (e.g., [15,46–48]).

Furthermore, our research goes in line with the findings of research that energy-saving behavior is a learned behavior [49]. In our study, those who participated in prior energy-saving initiatives reported a more positive energy-saving behavior. Energy behavior is a learned entity, thus prior experience increases the level of capability and splits energy users into groups according to this component. These findings are in line with the COM-B approach where the capability is one of three behavioral change factors [11,45].

Although the results of this study test behavioral change factors of energy behavior at a military unit, certain limitations need to be considered. Firstly, the data was collected at a unit located in a fixed installation only. To overcome this limitation, data was collected twice for two consecutive years to level out the operational tempo at the unit. The relative consistency of behavioral data suggests that these di fferences are minimal.

Secondly, the behavioral indicators of individual and collective behavior are measured using only a few questions. This is since the questionnaire response time was shortened to 10 min so as not to distract soldiers from their tasks. Therefore, the behavioral indicator was measured only with two measures: one for the individual and one for the collective behavior. Separation of individual and collective behavior was not foreseen from the start of this study—it became evident only during the data analysis. Despite this limitation, the relationship between *COM* and behavioral factors is statistically strong. However, in future research, we sugges<sup>t</sup> using several measures for individual and for collective behavior.

Based on this discussion, the research has a few practical implications:

For practitioners in the military, these results are interesting for gaining statistically tested model for pro-environmental energy behavior change. Our findings point out that pro-environmental energy behavior in the military unit follows the classic COM-B model with one exception: behavior (element B) is segregated into two components: collective and individual behavior. This revised model can be applied in practice and used to measure behavioral change when behavioral interventions are applied in the unit.

For practitioners in other than military organizations, this analysis demonstrates the duality of collective and individual factors in energy behavior. Our findings point to an importance of communality in the organization. Communality makes an impact on capability, opportunity, and motivation in pro-environmental energy behavior in an organization. Specifically, the findings indicate that in the organization with a strong sense of belonging, collective pro-environmental behavior remains active even if an individual behavior indicates low-involvement.
