This study aims to determine; (1) whether there is an initial ability of the experimental class students with the control class; (2) whether the mathematical reasoning ability of students who were given representation discourse was better than the control class; (3) how is the influence of representation discourse on the mathematical reasoning ability of junior high school students in Serang Regency; (4) how self-confidence is the students' multi-representation discourse. This research uses an experimental type and uses pretest-posttest control group design. This research was conducted in the even semester of 2022 with a sample of two classes: VII B as the experimental class and VII A as the control class. Data collection techniques using tests. At the same time, the data analysis technique used t-test analysis, Effect Size, and percentage of the questionnaire. Based on the study's results, the results obtained (1) The analysis results obtained sig (2 tailed) = 0.834 and the value of sig = 0.05. Because 0.0834 > 0.05, then H0 accepts; (2) The results of the independent t-test obtained sig (2 tailed) = 0.000 and a sig value of 0.05, then H 0 rejects and H1 accepts; (3) The results of the analysis using effect size obtained Cohen's effect size value of 3.494258 where the value is classified as very high; (4) the results of the analysis of the self-confidence test obtained an average of 83.167% which is classified as high. Conclusions (1) There is no difference between the experimental class students and the control class students; (2) The mathematical reasoning ability of students who are given representation discourse is better than conventional classes; (3) the influence of representation discourse on students' mathematical reasoning abilities is classified as very high; (4) self-confidence of students who are given the representation discourse is high