Computer-supported collaborative argumentation can improve understanding and problem-solving skills. This study uses WebCT to explore the improvement of argumentation in asynchronous, web-based discussions through goal instructions, which are statements at the end of a discussion prompt indicating what students should achieve. In a previous study (Nussbaum 2005), the goal instruction "generate as many reasons as possible" resulted in more balanced argumentation in an online environment. This study attempts to replicate this finding. It also examines the role of prior attitudes, knowledge and interest, and also the effect of elaborating on possible lines of reasoning in the question prompt. The goal instruction to generate as many reasons as possible (goal/no goal) was crossed with question elaboration (elaborated/unelaborated question) in a 2 x 2 randomized design using 131 undergraduates. Half of the participants were randomly assigned to complete the attitude survey before discussion. Results indicated that, when prior knowledge was controlled, the reason goal instruction positively affected argument development and opposing view exploration, but only for high-issue knowledge students. The sheer volume of notes created by the online environment may have caused cognitive overload for low-issue knowledge students. Question elaboration promoted balanced argumentation for all students, but especially those with low knowledge.