This work expands on and extends work previously pubished for the Lyra System towards the Ph.D. Computer Science qualifier requirement at NC State University. We conducted a human-subjects study of Lyra to evaluate the generated conversations and affinity groups for their believability and to inform future iterations of the simulation. The conversations were found to be moderately believable, rating 3.3 out of 5 on a Likert Scale. Additionally, survey respondents ascribed humanity to the actions of the virtual agents, associating them with competitiveness, persuasiveness, and intention. We believe this further adds to the believability of our system. In the long run, successful simulation of opinion change in social dynamics provides a foundation for computational recognition, prediction, and interfacing with human social behaviour.