A comparison of Peer Instruction and Process Oriented Guided Inquiry Learning -like pedagogies in teaching Software Testing and DevOps

In this Research Full paper, we present a comparative study of using two pedagogical approaches, Peer Instruction (PI) and Process Oriented Guided Inquiry Learning-like (POGIL-like) in software testing. We discuss our results from multiple studies conducted over a period of four years, both in-person and online, using both PI and POGIL-like in undergraduate software engineering classrooms. Our particular focus is on the topics of unit testing, integration testing, and continuous integration. We compare learning outcomes and shifts in student attitudes compared to pure lecture-based instruction. We included two studies conducted using PI in undergraduate software engineering classes. In these studies, we examined how PI can be used to teach the 3 topics mentioned above. They replicated their own studies over a period of two years and found consistently significant cognitive gains and positive affective shifts in both an in person and online class. For POGIL-like, we included two studies in an undergraduate software engineering class - one in-person, and one online. The intervention design centered around the core tenets of POGIL: no prior information given to students (which is the exact opposite of a flipped classroom), and carefully crafted “models” and “activities”. POGIL-like utilizes “learning cycles”, and activity questions are crafted based on the Explore-Invent-Apply cycle combined with Directed/Convergent/Divergent question types, all aimed at encouraging students to leverage the collective knowledge of the group. These approaches, based on social constructivist concepts, have been well studied in STEM disciplines but are relatively recent arrivals in CS. Our particular contribution is that we analyze both pedagogies in teaching software testing, using results from studies that utilized the same instructor and course materials. We analyze the similarities and differences between both pedagogical approaches, and rate their suitability in teaching software engineering and software testing. Our primary research questions are: •RQl: How do students learn software testing and DevOps using PI and POGIL-like compared to pure lecture format classes? •RQ2: What are the relative advantages and disadvantages for PI and POGIL-like from the instructor's perspective? In each of the studies we analyzed, we collected and classified individual student responses from a pre- and post- course survey for each implementation of the course (lecture, PI or POGIL; in person or online; honors or regular). The pre- and post-surveys each included two components: a technical, content-related survey specifically designed to test student knowledge and synthesis in unit testing, integration testing, and continuous integration, followed by a standard affective instrument aimed at gauging student attitudes and perceptions around the same topics. We adopted within-groups and between-groups Analysis of Variance techniques to determine statistical significance in our correlations. In all 4 studies involving PI and POGIL-like, we found encouraging results showing that students had learned significantly better than pure lecture alone. In this paper, for both PI and POGIL-like, we discuss and compare overall correctness for each group, as well as cognitive and affective gains based on the pre- and post- surveys. We also classify the relative strengths and weaknesses of each pedagogy based on student feedback. We discuss the similarities in these approaches as well as implementation differences, in addition to impediments faced while implementing them, along with ideas to encourage increase more widespread adoption of these evidence-based pedagogies.