Computation is changing our world. From how we work, to how we communicate and how we relax - few parts of our world have been left unaffected by computation and the technologies that it enables. The field of computer science and the ideas of the discipline are driving these changes, yet relatively little of it is present in contemporary K-12 education. Numerous local and national initiatives are underway to bring the powerful ideas of computing into classrooms around the world. An increasingly popular strategy being employed in this effort is the use of graphical, blocks-based programming environments like Scratch, Snap! and Alice. While these environments have been found to be effective at broadening participation with younger learners, open questions remain about their suitability in high school contexts. This dissertation uses a quasi-experimental, mixed methods design to understand the effects of blocks-based, text-based, and hybrid blocks-text programming environments in high school classrooms. Three custom-designed programming environments were created and used to understand how modality (blocks-based, text-based, and hybrid blocks/text) affects learners’ emerging understandings of core computer science concepts and their attitudes towards and perceptions of the discipline. Additionally, the study investigates if and how the different introductory programming modalities support learners’ transitions to more conventional textbased professional programming languages.
Findings from the study reveal that the modality matters. Differences were found with respect to students’ conceptualizations of programming constructs as well as student performance on content assessments and attitudinal surveys. The data show students in the Blocks condition scoring higher on content evaluations after a five-week curriculum and reporting higher levels of confidence, enjoyment, and interest in the field relative to students using an isomorphic text-based interface. However, these findings did not translate to greater success or better self-reported experiences upon transitioning to a professional, text-based programming language. After ten weeks of learning the Java programming language, students in the text condition showed comparable scores on content assessments and positive trends on attitudinal questions, whereas their blocks-based peers showed decreasingly levels of engagement and enjoyment. This study also demonstrates the potential of hybrid environments that blend features of blocks-based and text-based interfaces for providing the scaffolds and engagement of blocks-based tools with the perceived power and authenticity of text-based introductory environments. Collectively this work contributes to our understanding of the relationship between computational representations and learning programming, and can be used to inform the tools that will train the next generation of computationally literature citizens.