Research in the eras has highlighted the significance of noncognitive factors in students’ school-related outcomes. Although many reliable and valid measures to assess noncognitive factors have been extensively developed in past decades, the ensemble of constructs lack a unifying framework to integrate them. Before assessing the predictive power of noncognitive factors, research suggests, it is more important to move the measurement of noncognitive factors toward a clear and integrated system. The action-control theory, established using the human agency perspective, has been noted as a well-validated formulation to connect individuals’ beliefs, attitudes and volitional behaviors. Agency assumes that humans are the agent of their own development. The constructs built in this domain can represent people’s ability to control their actions under the external environment. As such, it provides precise and circumscribed operational characteristics to define and measure noncognitive variables. Using human agency perspective and action-control theory, a multidimensional set of constructs was generated. Primarily, this dissertation aims to verify the psychometrical reliability and validity of this new measure (AME, standing for agent, mean, and ends in action-control theory) in a longitudinal study comparing with the often-used existing measures. Findings from a series of confirmatory factor analysis (CFA) showed that the AME measure had a relative higher reliability, convergent and discriminant validity in current longitudinal data. In the multigroup CFA of longitudinal and gender variation, the equivalency was confirmed in all the AME constructs, suggesting that an equal change in students’ rating to the AME surveyed questions across time and gender was comparable; that is, any comparison made between genders and time points using the AME constructs should be meaningful. In addition, a series of t-test and Cohen’s effect size provided strong evidence of the response-shift bias in self-reported longitudinal data. The presence of response shifts in AME measure was relatively consistent against the often-used existing measures over time, meaning that students were more likely to provide accurate self-reflection in AME items. In other words, the AME reveal the changes and reduce the deleterious effects from a traditional pretest better than the often-used old measures. Taken together, this study developed a coherent evidence-based framework for considering the role of noncognitive factors in students’ performance. Consistent with the previous empirical studies, findings in this dissertation support the use of the action-control theory and the human agency perspective in measuring noncognitive factors.