A Comparison of Data-Driven and Traditional Approaches to Employee Performance Assessment
Nirvan Sharma, Patrick A. Hosein · 2020
Employee performance monitoring is not new. Traditional systems consider an average of peer and direct supervisor reviews across various subjective criteria in yearly performance appraisals. This leads to problems such as recency and social biases that devalue the credibility of the review process. Modern, data-driven systems have made it easier to automatically track employee data regarding various common performance metrics in useful ways. Statistical measures are derived from these metrics that are collated over a defined period. These measures are used to determine if and when any appropriate action needs to be taken. However, the statistics fail to account for common characteristics of the employee population. Instead of comparing individual performance against rigid idealistic norms, this paper presents an automated and objective approach to determine the collective population norm and calculate abnormal deviations from it. Employees who deviate from the norm are automatically flagged and subject to further review from management. Ideally, a combination of calculated metrics using this approach can be used to accurately reflect real-world situations and highlight anomalies in the employee work force. We compare the results of the proposed statistical approach with traditional human review approaches and evaluate their correlation.