Due to the increased availability of online collaboration tools, individuals are now likely to work together in settings where computers are their primary mode of communication. However, because many communication cues are absent in these settings, unique problems arise, such as deception. Deceptive individuals are difficult to detect over computer-mediated channels because many audio and visual cues to deception are filtered out. This dissertation presents two experiments where groups performed a computer-mediated collaborative task either without deceivers or with confederate deceivers. Task complexity was manipulated in the first experiment, and subject experience and task complexity were manipulated in the second experiment. Results suggest that groups performing a low complexity task were better at detecting deception than were groups performing a high complexity task. Furthermore, experienced groups had higher task performance but did not have higher deception detection accuracy than did groups without experience. These findings should help organizations identify decision-making settings where group performance and deception detection accuracy are at risk and allow them to take actions to minimize the negative impact of deception. These actions might include minimizing the cognitive complexity of a group task and making sure that individuals have had multiple experiences together before performing their task.