The purpose of this study is to investigate and conceptualize the ranks of importance of social cognitive variables on university students' computer programming performances. Spatial ability, working memory, self-efficacy, gender, prior knowledge and the universities students attend were taken as variables to be analyzed. The study has been conducted with 129 2nd year undergraduate students, who have taken Programming Languages-I course from three universities. Spatial ability has been measured through mental rotation and spatial visualization tests; working memory has been attained through the measurement of two sub-dimensions; visual-spatial and verbal working memory. Data were analyzed through Boosted Regression Trees and Random Forests, which are non-parametric predictive data mining techniques. The analyses yielded a user model that would predict students' computer programming performance based on various social and cognitive variables. The results yielded that the variables, which contributed to the programming performance prediction significantly, were spatial orientation skill, spatial memory, mental orientation, self-efficacy perception and verbal memory with equal importance weights. Yet, the effect of prior knowledge and gender on programming performance has not been found to be significant. The importance of ranks of variables and the proportion of predicted variance of programming performance could be used as guidelines when designing instruction and developing curriculum.