Reducing the Central Tendency Bias in the Performance Appraisal System in SMEs

Abstract

Purpose | This study aims to investigate the persistence and evolution of central tendency bias in performance appraisal systems (PAS) in the small and medium-sized enterprise (SME) sector. Research method | A two-year longitudinal dataset was used, focusing on a homogeneous sample of 125 meter readers at a Polish SME. Due to the non-normal distribution of rating data, non-parametric methods, including the Friedman test, Kendall’s W coefficient, kurtosis, and standard deviation analysis, were employed to measure score dispersion and rank stability. Results | The empirical results confirm a high persistence of central tendency bias, with mean ratings dominating across both cycles (71% and 74%). However, a significant decrease in kurtosis (from 5.36 to 4.5) suggests a gradual shift towards more extreme scores, indicating system maturation. Furthermore, an extremely low Kendall’s W (0.10) indicates low rank stability, suggesting that managers performed more independent reassessments in the next cycle rather than habitually replicating previous scores. Originality / value / implications / recommendations | This paper contributes to the literature by providing rare longitudinal evidence on the dynamics of rating errors, particularly in the context of SMEs. It interprets the observed change in rating distribution through the lens of “bureaucratic enabling”, presenting kurtosis reduction as a signal of organisational learning and a shift towards more transparent and equitable human capital management.