This quantitative research study examined the impact of robotics and STEM activities on the STEM self-efficacy of African American high school females from a low socio-economic status enrolled in Early Childhood Education Career and Technical Education classes in a large urban public school system in Jackson, Mississippi using a quasi-experimental 2k full factorial design with a single repeater. Using the Social Cognitive Career Theory (Lent, Brown & Hackett, 1994), STEM self-efficacy was measured using the Self-efficacy in Technology and Science (SETS) instrument developed by Ketelhut (2010). Overall, girls STEM self-efficacy improved significantly regardless of the treatment environment. Overall self-efficacy scores in science inquiry, synchronous chat and computer gaming also showed statistically significant increases. When considering the type of treatment, girls self-efficacy improved greatest as a result of the GoldieBlox only environments compared to all other treatments settings including robotics. An ANCOVA analysis controlling for the type of treatment setting resulted in a statistically significant increase in the videogaming SETS construct.