Cliff Otieno

1 Practice Multiple Regression Modelling and Interpretation. 25th January 2025 2 In this analysis, a multiple regression model was developed to examine the factors influencing employee productivity, measured by the dependent variable Productivity_Score. The dataset included 1,000 observations and a mix of numerical and categorical predictor variables such as Department, Education, Job_level, Work_Location, Training, Years_Experience, Weekly_Hours, Team_Size, and Projects. As part of the data preparation process, exploratory data analysis (EDA) was conducted to assess data completeness and quality. Missing values were addressed through imputation using the mean value rather than removal, preserving the integrity of the sample. Categorical variables were converted into binary dummy variables to enable regression analysis, and a compound dummy variable was 3 constructed to explore interaction effects specifically, whether employees in technical departments at higher job levels exhibited distinct productivity patterns. An R2 value of 0.857 and an adjusted R2 of 0.856 shows that nearly 86% of the variation in employee productivity is described by the predictors in the model. The F-statistic was highly significant (F = 661.5, p < 0.001), proofing the overall strength and reliability of the model. Upon examining the coefficients, several predictors were identified as statistically significant. Years of experience (β = 2.07, p < 0.001), weekly working hours (β = 1.46, p < 0.001), and the number of projects handled (β = 4.88, p < 0.001) were all positively associated with productivity. Notably, team size had a statistically significant negative impact (β = -0.47, p < 0.001), suggesting that employees in larger teams may experience reduced individual productivity, potentially due to coordination complexity or diluted responsibilities. The predictors like department, education level, job level, work location, and training participation do not show any statistical significance at the 5% level. This indicates that though both organizational context and employee background provide important bases for judgment, actual productivity is more closely tied to inputs in the work and experience. These results show that the focus of the productivity development efforts would, instead, be on more meaningful work assignments, prolonged and more concentrated hours of work, and enhancing employee longevity and experience accumulation. Furthermore, it may also have diminishing returns and operational difficulties with exceedingly large team sizes. In brief, the model proves a rigorous quantitative underpinning to understand what makes productivity in employees. Concentrating on measurable parameters, such as projects, work hours, and experience, organizations can better utilize their human resources to drive higher productivity. The revelations of such a model are particularly useful for decisionmakers who are interested in developing data-based policies to motivate high-performance workplaces. 4 Out of the individual predictors examined, four are statistically significant as contributors to the outcome. Years of Experience produced a positive and highly significant coefficient of 2.07, which implies that every additional year of experience increases a person's outcome by nearly two units, letting everything else remain constant. Weekly Working Hours are significant predictors because of the coefficient of 1.46; this means that longer work hours correlate with higher output values. The number of Projects also reflects a strong positive impact; each additional project corresponds to nearly 4.88 cumulative values to the predicted outcome. Team Size, however, has a strong but negative predictor, -0.47, which shows that high outcome values are linked to large teams. In contrast, variables such as Department, Education, Job Level, Work Location, and Training lack statistical significance effects, as their p-values exceed the common threshold of 0.05. The results suggest that, the model with respect to these factors lack a meaningful impact on the dependent variable. While they may have theoretical relevance, their statistical insignificance indicates that their inclusion in future models may be reconsidered unless additional contextual or interaction effects are explored. Overall, the model highlights the importance of experience, workload, and project involvement, while suggesting that organizational and demographic characteristics may play a lesser role.