Digital learning technologies have grown increasingly important in physics education, partly enforced through the COVID-19 pandemic. During the pandemic, digital technologies allowed for continued teaching and learning of students even when schools were closed. While research in psychology and educational technology has yielded many insights into the effectiveness of e-learning courses, fewer studies have examined the design of e-learning courses. Few studies have empirically investigated the design of learning tasks as a central element of e-learning courses. The present study analyzes how the design of tasks in e-learning courses, specifically with respect to their degree of openness as well as the relevance of their contexts, influences students' behavioral engagement, learning outcomes, and situational interest. Due to the importance of e-learning courses during the COVID-19 pandemic, we also analyzed the extent to which specific learning settings (classroom learning, distance learning) influence the effects of e-learning course design on students' behavioral engagement, learning outcomes, and situational interest. To investigate the research questions, we analyzed a total of N=1060 datasets for 12 different e-learning courses (3 to 5 lessons, middle school physics), of which n=557 were completed before and n=503 during the COVID-19 pandemic. The results suggest that e-learning courses with a high proportion of learning tasks that relate to meaningful real-world contexts appear to be more conducive to behavioral engagement, learning outcomes, and situational interest. Regarding the consideration of open-ended tasks, the results suggest that these appear to be more useful for classroom learning but should be used in a limited way when designing e-learning courses for distance education.