The utilization of RF-MEMS, which stands for Microsystem-based (MEMS) Radio Frequency (RF) passive components, is garnering growing attention within the realm of Beyond-5G (B5G) and 6G technologies, despite its longstanding existence. This trend is fueled by the impressive RF characteristics achievable through the judicious exploitation of this technology. However, the complex interplay of various physical phenomena in RF-MEMS, spanning mechanical, electrical, and electromagnetic domains, renders the design and optimization of new configurations challenging. In this study, a modeling approach based on Lumped Element Networks (LEN) is employed to accurately predict the Scattering Parameters (S-parameters) characteristics of multi-state and highly reconfigurable RF-MEMS devices. The device under scrutiny is a multi-state RF step power attenuator, previously fabricated, tested, and documented in literature by the principal author. Although these physical devices exhibit flat attenuation characteristics, they are subject to certain non-idealities inherent to the technology. The refined LEN-based methodology presented herein aims to interpret and incorporate such undesirable parasitic effects to provide precise predictions for real RF-MEMS devices. Two custom metrics, referred to as Percent Magnitude Difference (PMD) and Percent Phase Difference (PPD), are utilized to evaluate the accuracy of the LEN model, revealing differences consistently within 1% and 8%, respectively, across a frequency range spanning from 100 MHz to 13.5 GHz.