The research aims to describe the underlying cause of students' low Creative Thinking Ability (CTA), and examine the effectiveness of online learning assisted by module in improving CTA in terms of self-efficacy. The research applied a mixed-method. The subjects were 8-grade students. The qualitative research subjects were selected purposively, generating two students for each category of low, medium, and high self-efficacy. While quantitative research used cluster sampling to classify experimental and control classes. The independent variable of the study was self-efficacy, and the dependent variable was CTA. Data collection was conducted by observation, interviews, documents, questionnaire, and test. Data was analyzed using descriptive analysis, statistical regression tests, and t-test. The results showed that the underlying cause of low CTA was in students' low and medium self-efficacy. Students with low and moderate self-efficacy highly depended on teacher help. For students with high self-efficacy, the CTA worked well. The results also revealed that the average CTA in the experimental class reached the minimum criteria of mastery learning; the average CTA of the experimental class was better than the CTA of the control class; and the effect of positive self-efficacy on CTA was 38.50% in the experimental category, showing that this learning was effective.