This study attempts to: (a) investigate whether positive and negative emotions mediate the pathways linking self-efficacy for learning programming with effort and persistence in undergraduates' learning Scratch programming combining with a programmable hardware platform (i.e., Arduino), and (b) assess the effect of academic major (i.e., information technology-related majors vs non-information technology-related majors) on self-efficacy for learning programming and emotions. With the use of responses collected from a sample of 156 undergraduate students, the research model is empirically validated using partial least squares structural equation modeling (PLS-SEM). The findings reveal that undergraduates' effort and persistence while learning programming were significantly predicted by positive emotions, but not by negative emotions. Self-efficacy for learning programming significantly and positively influenced positive emotions and negatively influenced negative emotions. Finally, variations in predicting positive emotions were found between different academic majors: students in information technology-related majors experienced less negative emotional experiences than those in non-information technology-related majors.