Technology training in the classroom is critical in preparing students for upper level classes as well as professional careers, especially in fields such as technology. One of the key enablers to this process is computer self-efficacy (CSE), which has an extensive stream of empirical research. Despite this, one of the missing pieces is how CSE actually changes during training, and how such change is related to antecedents and performance outcomes. Measuring change requires repeated data gathering and the use of latent growth modeling, a relatively new statistical technique. This study examines CSE (specifically general CSE or GCSE) growth over time during training, and how this growth is influenced by anxiety and gender and influences performance, using a semester-long lab course covering three applications. The use of GCSE growth more accurately models how students actually learn in a technology classroom. It provides novel clarity in the interaction of gender, anxiety, GCSE, specific CSEs, and performance during training. The study finds that the relationship between anxiety and self-efficacy decreases over time during training, becoming non-significant; it clarifies the significant role gender plays in influencing GCSE at the start of and during training. It finds GCSE influences application performance only through specific CSEs.