The aim of this study was to explore the association between MVPA and MH, and between MVPA and WE. Although MVPA was found to reduce the risk for mental disorders in several previous studies, we found no evidence in this study for the aforementioned associations. Hamer & Stamatakis  also found no associations between objectively measured MVPA and well-being, although they did find an association between self-reported MVPA and well-being.
Although associations of PA with the negative side of mental health (i.e. mental disorders) were previously found, it could be that the PA-MH relationship works differently for the negative (i.e. mental disorders) and the positive side (i.e. well-being, work engagement) of MH. For example, it could be that more PA is associated with a reduction of the risk of depression, but this does not necessarily imply that it is associated with an increase of the odds of happiness or comparable mental states such as work engagement. This possible difference should be considered when examining potential mechanisms for the PA- MH relationship. Potential psychosocial mechanisms for the PA-MH and PA-WE relationship (for example: PA enlarges self efficacy and self esteem, which could contribute to MH and WE) could be explored in qualitative research, as this provides insight in the nature of this relationship and possible mechanisms.
A possible explanation for the lack of an association between MVPA and WE, might be that WE is work-related to such an extent, that it is not affected by behaviours outside the work domain (e.g. leisure time MVPA). This might explain why previous studies have found associations with work-related factors, such as job demands and resources [5, 35], financial returns  and job performance . In future studies on associations of behaviours (such as MVPA) and WE, it is recommended to consider the different domains of the PA behaviour. Although this might be the case for WE, this does not explain why there was a lack of associations between MVPA and general MH.
Another explanation for the lack of associations could also be the amount of time spent in MVPA. It is possible that the participants did not perform enough MVPA to show an association with MH or WE. The study population indicated to perceive barriers such as lack of time to engage in leisure time physical activity . When aiming for an increase in physical activity, such barriers should be addressed. Also, it is possible that the data showed no significant results as a consequence of a lack of statistical power. This could be the case for both the objective data, which were available for a subgroup of 100 participants, as for the self-reported data, which were available for the total study population. However, since the b’s were very small, it is considered that greater statistical power would not lead to relevant results.
When comparing the objectively measured data to the subjective data, it appears that in our study, participants tended to overestimate their physical activity. This was also the case in the validation study of the SQUASH: the mean absolute amount spent in each intensity category was consistently higher for the SQUASH than for the accelerometer . This could be due to reporting bias as self-reported physical activity is known to suffer from this type attributable to a combination of social desirability bias and the cognitive challenge associated with estimating frequency and duration of physical activity . In addition, it could be due to some characteristics of the questionnaire; for example one hour of tennis, reported in the SQUASH, is equal to one hour of vigorous activity. An accelerometer does not measure 60 minutes of vigorous activity while playing tennis, but only short bouts of vigorous activity and in-between bouts of moderate or light PA.
Strengths and limitations
A first strength of this study is, that it is the first study exploring the association between MVPA and MH, and the association between MVPA and WE. Another strength of this study is the use of both self-reported and objective measurements to asses PA. While self-reported PA, through questionnaires, is subject to numerous sources of error, few studies have examined subjective well-being in association with objectively measured PA .
Despite the advantages of the objective measurement, the accelerometer has also some limitations. Single waist worn accelerometers are unable to detect specific activity loads, such as load carriage of pulling or pushing and walking on different terrain. Mainly household activities can be substantially underestimated through counts per minute . To classify activity type dual placement, two hip worn or placement around the ankle, should be considered . Moreover, another issue regarding accelerometry data is that activity intensity thresholds, or cut-off points, vary in the literature. Cut-off points should be refined to capture the full range of activity [28, 33, 39]. Also, cut-off points should be age specific. In the literature on calibration, different thresholds are given for children and adults , but no distinction is made in cut-off points for moderate and vigorous for older adults compared to younger adults. However, the calibration study for the cut-off point used in this study was performed with young adults with a mean age of 23 years , while the mean age of the subgroup of the study population wearing the accelerometer was considerably higher (47 years).
Finally, it should be taken into account that the sample consisted of mainly higher educated participants in scientific professions. Barkhuizen and Rothmann  found that higher educated workers were more engaged than their lower educated colleagues. However, our sample of highly educated workers were averagely engaged, with a slightly higher mean 4.1(SD = 0.8) than the a UWES-17 mean of a Dutch sample (n=2313) of 3.8 (SD=1.1) . Differences in mean levels of engagement between various occupational groups might be significant, but relatively small and they almost never exceed the size of one standard deviation . Thus, these results might be representative for other highly educated workers in scientific professions. Nevertheless, it is not recommended to generalize the results to different professions; it could be argued the relationship between physical activity and work engagement is different for professions that require for example more physical activity at work.