The present work analyzes the effect of argumentative discourse goals on the argumentative quality in pre-service teachers, when writing argumentative texts. As previous research suggests, the argumentative discourse goal has a great impact on learning outcomes, confirmation bias and my-side bias, the potential for building knowledge and the argumentative quality. Seventy-six pre-service teachers were presented a dilemma (thermal vs. nuclear energy) and asked to take position and write two argumentative essays based on a provided corpus of evidence (two graphs pro nuclear energy and two graphs against it) before and after a chat-based dialogue with an opposing partner. Dyads were divided into two conditions: argue to persuade and argue to reach consensus. Results showed that participants in the consensus group decreased their proportion of confirmation bias and justifications, and increased their proportion of meta-statements and counter-arguments. The persuasion group also decreased in confirmation bias and increased in meta-statements (although less significantly than the consensus group) but in the proportion of justifications and counter-arguments showed no change. These results in both conditions are considered as a positive effect of the dialogue itself. However, the better results in the consensus group are believed to have several instructional implications.