Expectancy value theory is often evoked by educational psychologists to explain gender differences in Science, Technology, Engineering, and Mathematics (STEM) variables. Yet gender does not operate in isolation. Nor are gender effects likely to be context free. In the current meta-analysis, we explore gender differences in STEM-related expectancy for success, and the task values of intrinsic, utility, attainment, and cost. We find that gender differences were generally small in size. Invoking the concept of intersectionality, we find that heterogeneity in gender effect sizes are large and gender differences are moderated, primarily, by socioeconomic status, ethnic diversity, and somewhat by national gender equality.