We did not find self-efficacy to be a general predictor of compliance to specific strength exercises at the workplace. Self-efficacy may be a predictor of compliance to specific strength exercises in some workplaces, but not in others. The present study included a private and a public sector company, and when analyzed separately, self-efficacy was found to be a predictor of compliance during 20 weeks in the private sector company, but not in the public sector company. There was no significant effect of exercises versus reference on changes in self-efficacy over 20 weeks, and self-efficacy did not change over time in any group.
Self-efficacy has been shown to predict physical activity in a series of observational studies where the research did not influence the physical activity level of the study participants [
17,
18,
20,
25] and has also been found to be a predictor in experimental settings where the outcome was compliance to one or more predefined exercise programs [
21,
22]. This study did not find baseline self-efficacy to be a predictor of compliance to training over time in the intervention as a whole. However, we did find self-efficacy to be a predictor of compliance in the private sector company, but not in the public sector company. Workplace-specific differences that need to be taken into account, when considering compliance, might therefore be present. Barriers towards exercising are multidimensional and can be both population and workplace specific [
34]. These barriers might be lack of time, lack of support from company leaders and colleagues, etc. Studies have identified barriers towards exercising at the workplace like work load, limited break time, time wise scheduling, and work conflicts [
35, 38]—factors that might have influenced compliance differently at the two companies in the present study. The limited predictive value of self-efficacy may also indicate that other individual factors and factors other than specific individual psychological factors like self-efficacy may be more important in predicting exercise behavior. The social cognitive theory specifies a set of core determinants for health behavior. These include knowledge of health risks and benefits, perceived self-efficacy, expectancies about the consequences of one's actions, personal goals, and perceived impediments [
15]. Thus, personal psychological factors other than self-efficacy may play a role regarding exercise participation. A specific worksite atmosphere as well as social and cultural factors, like social class, peers, and environment, may better explain barriers or motivation for exercise than initial self-efficacy. Moreover, interactions between participants and social support from group members influence a person's health behavior [
11,
13,
15]. The personality and commitment of the instructor and the contact person of the exercise group may also have influenced compliance. In future research, a more systemic approach could encompass these factors and possibly provide knowledge of important issues to ensure compliance to interventions and, furthermore, to better understand barriers and motivation for exercise and physical activity. In this study, participants with low self-efficacy had a lower compliance than participants with moderate and high self-efficacy. Compliance did not differ between the latter two groups. Thus, self-efficacy to a small extend is not sufficient in securing compliance, which should be taken into account when planning exercise regimens.
The prospective design and an analysis controlling for factors that might be related to compliance strengthen our study. Apart from age and gender, we controlled for baseline neck and shoulder pain considering that the aim of the study was to reduce neck and shoulder pain. In contrast to previous studies, we also took intracluster correlations into account by treating individual observations within the exercise groups as repeated measurements. A weakness of research studies focusing on exercise is the dependence on volunteers. People with extremely low self-efficacy towards exercising would probably not volunteer for an exercise program. Employees who accept participation might therefore have a higher degree of exercise-specific self-efficacy than those who do not volunteer, creating a risk of a selected group with a smaller difference in self-efficacy than in the population as a whole. The low self-efficacy group in the present study (n = 35) was smaller than the moderate (n = 115) and the high self-efficacy (n = 118) groups. Compliance as well as self-efficacy among the participants might have been influenced by the presence of instructors and by the workplace design of the study, which limits the external validity to workplace settings involving training instructors. Thus, the results cannot be generalized to exercise behavior where no instructors are present nor to the population as a whole. A limitation of the study is the use of a self-efficacy measure on generic exercise self-efficacy even though adapted in wording to the workplace context. A workplace-specific measure might have shown another picture of self-efficacy as a predictor of compliance. The measure, though, does reflect an individual's efficacy belief regarding initiation and maintenance of exercise. Even though self-efficacy is an action-specific measure, it might also well reflect a fundamental view on exercise participation and thus on exercising at the workplace.