3.2.2. Physical Activity and Dietary Behaviors

The number of minutes that participants spent performing recreational physical activity decreased on average in the control groups, but increased in the intervention groups. This difference was non-significant in the unadjusted model, but became significant in the adjusted model (B = 31.03, *p* = 0.03) (Table 4).

There were no observable differences at FU2 between the intervention and control sites for unhealthy diet (i.e., eating fatty or sugary foods) for the unadjusted or adjusted models respectively (Table 4). For healthy diet, we observed no differences between the intervention and control groups in the unadjusted model. However, when accounting for the matched intervention and control pairs within each company, and adjusting for covariates, we found a significant small positive influence on healthier diet behaviors in the intervention compared to the control groups (B = 0.87; *p* = 0.008). Overall, we saw a small improvement in having a more balanced diet nearing significance due to the improvement in healthy eating behaviors (B = 1.05, *p* = 0.054). Thus, hypothesis 1b was partially supported.

**Table 4.** Effects of the ARM intervention on physical activity and dietary behaviors from baseline to FU2 (6 months).


B = regression coefficient; CI = confidence intervals. Results from linear regression models with cluster robust standard errors to account for individual clustering within worksites (\* *p* <0.1, \*\* *p* <0.05). <sup>1</sup> Adjusted with fixed effects for matched pairs and age, sex, race, title, and trade.
