P-443 Employee’s Motivation: A Predictive Model Building Approach Based on 29 Psycho-Social Risk Factors

Introduction Employee’s motivation is an important topic in occupational psychology. The aim of this study was to build a prediction model of motivation based on real world data from psychosocial risk assessments. Methods Mensura, a Belgian occupational health service, developed a psychosocial risk assessment tool, the Sonar, which questions employees on 29 psychosocial risk factors and 7 well-being indicators. Data, collected in 385 companies from 2019 till 2023, were analyzed in this exploratory study. The psychosocial risk factors were rated on a 3-point scale (1-not ok, 2-neutral, 3-ok). Motivation was measured on a 4-point scale (1-not motivated, 4-very motivated). Both stepwise and hierarchical model building approaches were used. Twelve models using linear regression were explored. Criteria for the preferred model included statistical fit, predictive power and parsimony. Specification error and multicollinearity were checked. Results Data were analyzed on 36.264 employees of which 3.8%, 95% CI [3.6, 4.0] reported not being motivated, 11.3%, 95% CI [11.0, 11.7] a little motivated, 41.0%, 95% CI [40.5, 41.5] somewhat motivated and 43.8%, 95% CI [43.3, 44.3] very motivated. Nine models had similar fit statistics. Our preferred model included three predictors: satisfactory jobvariation, affiliation with employers mission and employers policy (R2 = 26.0%, RMSE = 0.69, F(6, 10872) = 622.20, p < 0.001). Discussion There was no overfitting since the model’s predictive performance in the test dataset equaled it’s goodness of fit in the trained dataset. Next, for explanatory purpose a linear mixed effects model should be used. Conclusion This study suggests that immaterial workplace characteristics are paramount for motivation.