In this paper I describe the effects of gaze on people utilizing humanoid avatars to engage in conversation. The collaborative virtual environment (CVE) chosen for this research is Second Life®. Through the use of Second Life®, I compared responses to dyadic conversations in three different conditions: chat-only, direct gaze avatar condition, and indirect gaze avatar condition. Responses were collected in order to measure the "quality of communication" experienced. The mean questionnaire responses show that an avatar presence does improve the quality of chat communication. It also shows that direct avatar gaze improves upon indirect avatar gaze. However the pair wise comparison of differences between these two specific conditions failed to disprove the null hypothesis. As such, on a significance level of 0.05, the direct avatar gaze condition does not improve upon the indirect gaze condition.