Improving Content Learning and Student Perceptions in CS1 with Scrumage

Uses, Revisions, and the Future of Validated Assessments in Computing Education: A Case Study of the FCS1 and SCS1

“…In a study of 79 novice programmers (defined as having had at most non-Python programming course and never learned or used Python), findings supported the claim that more agency and information can lead to higher student motivation, but not necessarily improved student learning [75]. Java: [17] Python: [37] (mSCS1), [58] MATLAB: [2, 3] (MCS1) German: [66,67] Finnish: [15] With the rise of block-based programming environments, questions have been raised about how block-based programming and block-based programming with text-based programming (dualmodality programming) compare to text-based programming. For a study comparing dual-modality programming environments and instruction with text-based approaches to instruction, participant learning was measured via the SCS1 assessment and course exams for 673 undergraduates in a semester-long CS1 course [5].…”

“…This, combined with the acknowledgment that it is a difficult assessment, can reduce the viability of the SCS1 assessment as a research tool. As such, there have been various efforts to shorten the SCS1 assessment [6,17,45]. Parker and Guzdial, the original creators of the SCS1 assessment, modified the SCS1 themselves, parsing it down to a 12 question test from the original 27 questions [45].…”

“…Despite the goal of creating an assessment that could be used by a multitude of students learning different languages, many research teams have chosen to replicate the SCS1 assessment into specific programming languages [2,3,17,37,58]. In one instance, the motivation for moving away from the pseudocode-based assessment was that the pseudocode was non-intuitive and concerns were raised about burdening students with interpreting a pseudolanguage, which was discussed as a concern with the results of the FCS1 validation studies [37].…”

“…Other research teams have taken a similar approach of translating a subset of the SCS1 assessment into Python, though also adding 127 short, deliberate practice questions, for use in a pre-CS2 assessment [58]. Another case chose to translate some questions of the SCS1 questions into Java for use with 42 students in a CS1 course taught in Java [17]. Duvall et al cited that the pseudocode language was too similar to Python and their students had difficulty understanding the questions.…”