A Comparative MCDM Framework Integrating AHP, SAW and TOPSIS for Robust Public Sector Selection

Abstract

The selection of top employees at DINPMP2KUKM in Bangka Regency faces challenges including subjective evaluation, a lack of clear criteria, and difficulties in assessing employee performance across different fields, which may lead to perceptions of unfairness and decreased work motivation. This study aims to apply a comparative multi-criteria decision-making framework combining the Analytic Hierarchy Process (AHP), Simple Additive Weighting (SAW), and the Technique of Order Preference by Similarity to Ideal Solution (TOPSIS). A mixed-methods approach was employed, involving five respondents (four division heads and one human resources sub-division head) and three employee alternatives. Data were collected through structured interviews and paired comparison questionnaires. AHP was used to determine criterion weights, while SAW and TOPSIS were applied to rank alternatives, followed by sensitivity analysis to test ranking robustness. Unlike prior studies that generally used only a single MCDM method without robustness testing, this study validates ranking consistency across three different MCDM methods with sensitivity-based robustness testing. The results indicate that cooperation (29.2%) is the dominant criterion, followed by performance (22.6%), discipline and innovation (16.8% each), and integrity (14.6%), with a consistency ratio of 0.05 indicating consistent evaluations. All three methods produced identical rankings, with Employee 3 selected as the top employee (43%), excelling in four out of five criteria. Sensitivity analysis confirmed that Employee 3 remained at the top when the weight of cooperation was altered by ±20%, demonstrating decision robustness. Given the limitations in the number of respondents and alternatives, these findings suggest that the combination of AHP, SAW, and TOPSIS with sensitivity analysis can produce more consistent employee selection outcomes compared to single-method approaches. Further research is recommended to develop a web-based decision support system with a larger number of respondents and alternatives.

GS Cited Analysis