Abstract—In academic institutions, merit based promotion & tenure decisions have always been beset with controversy. This paper suggests an agent based model of the decision making process using spectral graph theory, where the voting agents are the vertices of the graph, and edge weights are determined based on the extent of collaborative research between the agents, as well as their estimated levels of social interactions. The model assumes that agents with lower research productivities tend to interact more often with one another. Using the graph theoretic spectrum, the paper applies a multi-dimensional representation that maps the voting agents into points on a low-dimensional grid, where agents that are likely to influence each other more are closely spaced. A multi-agent system model is proposed, where votes are determined based on very small randomly assigned initial values, and the mutual interaction during the decision making process. The model incorporates limited collusive voting within academically inbred agents. The proposed model is able to accurately reproduce a known promotion decision making from a department of a research oriented university which involved a sizable number of voting agents with low research output.
Index Terms—Convex function, graph laplacian, promotion and tenure, multi-agent systems, simulations.
S. Das is with the ECE Department, Kansas State University, Manhattan, KS 66506, USA (e-mail: sdas@ksu.edu).
[PDF]
Cite: Sanjoy Das, "An Agent Based-Model and Equilibrium Analysis of Academic P&T Decisions: The Effects of Inbreeding," International Journal of Modeling and Optimization vol. 8, no. 5, pp. 254-259, 2018.