This work-in-progress research paper presents the development and initial validation of an instrument to measure adaptability among engineering students. Adaptability is defined as responding quickly and flexibly to changing conditions and situations. Grounded in adaptability literature and theory, and informed by ongoing research, the instrument includes two scales corresponding to adaptability behaviors and mindsets. The online instrument will be piloted with undergraduate engineering students nationwide in the summer and fall of 2023. Exploratory factor analysis, confirmatory factor analysis, and multiple-item response analysis are planned to establish evidence of structural validity. Evidence of internal consistency reliability and stability reliability will also be collected. Findings using the instrument have the potential to influence university and industry stakeholders, engineering students, and researchers interested in studying engineering adaptability, which may, in turn, lead to better preparation and retention of engineers for the workforce.

This research Full Paper conducts a qualitative meta-analysis of hackathon research in IEEE through the lens of Gee's affinity spaces. Hackathon research has resulted in a variety of studies characterizing, adapting, and exploring the hackathon phenomenon in which participants create new projects and technologies to solve problems. Hackathons continue to create valuable learning opportunities within hackathons in the wild, and throughout hackathon adaptations across classrooms, industry, and non-CS fields. Affinity spaces are a community-based framework characterizing spaces in which participants engage at different levels, learn and teach each other regardless of demographics, and are united by a common goal. Affinity spaces resonate deeply with hackathons in the wild for these very attributes as hackathons are spaces where the community gathers to teach and learn from each other alongside their shared aspirations. Additionally, these events encourage participation in a multitude of ways, such as mentorship, presenting, hacking, organizing, and even volunteering. Hackathons have also become a novelty in research, resulting in a growing number of hackathon adaptations. Adaptations leverage a hackathon structure or modify it to meet new goals, not always tech-focused. However, there remains a critical question in preserving the spirit or essence of a hackathon that continues to drive the phenomenon. Do hackathon adaptations still retain these community-supportive attributes, or are they lost in translation? To answer this question, a meta-qualitative analysis is critical. Revisiting past publications is best to identify emerging trends, unite existing work, and encourage research in informed and targeted directions. The meta-qualitative analysis pursued in this work leverages only one database, IEEE Xplore, to briefly explore the literature before engaging in a deeper and longer endeavor across multiple databases. The resultant set of papers was deductively analyzed using codes derived from the affinity spaces framework. Our findings hint at new pathways of study for hackathon research and provide insights into improving hackathon adaptation research.

This innovative practice work in progress (WIP) paper details the development of an observation protocol that was developed to promote pedagogical risk-taking and additive innovation. Our higher level goal is to create a self-sustaining community of pedagogical exploration through formative feedback driven by peer observation. This effort aims to break the mold of status quo professional development protocols with questions specifically designed to ensure educators are receiving targeted, desired feedback, while engaging observers to think about how they might apply something similar in their own classroom(s). Attempts to test this protocol unearthed deeper concerns about observations and faculty reluctance to be observed. This underlying anxiety and concern associated with observations needs to be addressed, particularly during the pandemic. In this paper we share our process of developing this protocol, our pilot testing of the protocol, barriers encountered in further testing of the protocol, and literature that explains faculty reluctance to peer observation protocols that will help guide future efforts.

This work addresses sky segmentation, the task of determining sky and non-sky pixels in images, and improving upon existing state-of-the-art models. Outdoor scene parsing models are often trained on ideal datasets and produce high-quality results. However, this leads to inferior performance when applied to real-world images. The quality of scene parsing, particularly sky segmentation, decreases in night-time images, images involving varying weather conditions, and scene changes due to seasonal weather. We address these challenges using the RefineNet model in conjunction with two datasets: SkyFinder, and a subset of the SUN database containing sky regions (SUN-sky, henceforth). We achieve an improvement of 10-15% in the average MCR compared to prior methods using the SkyFinder dataset, and nearly 36% improvement from an off-the-shelf model in terms of average mIOU score. Employing fully connected conditional random fields as a post processing method demonstrates further enhancement of our results. Furthermore, by analyzing models over images with respect to two aspects, time of day and weather conditions, we find that when facing the same challenges as prior methods, our trained models significantly outperform them.