This year’s Research & Analytics Showcase gathered those working to improve the future of teaching and learning

Mindy Arbaugh, Content Specialist

From studying the effectiveness of simulated medical procedures to building artificial intelligence agentic tools, there was no shortage of illuminating findings at the 2025 Research & Analytics Showcase, held on April 30 at the Center for Academic Innovation.

The event featured research from center staff, Student Fellows, and University of Michigan faculty and students. Attendees listened to a trio of presentations before exploring a poster fair of research projects.

“We want to be a forum for education research here on campus,” said Becky Matz, CAI research scientist lead, as she welcomed visitors to the center. “I really hope you enjoy your time here, being with colleagues who are also committed to enhancing the future of higher education.” 

Tools and Online Learning Experiences to Support Learners

Sharon Jessica, a research associate in Computer Science and Engineering, and Vaibhav Balloli, a PhD candidate at U-M’s Realize Lab, presented first on “Understanding Perspectives and Questions in AI Careers.”

The pair surveyed learners interested in an AI career path and found students wanted to know more about AI applications in other fields like education and healthcare, the technical aspects of AI, and job opportunities. The team then submitted the questions into a large language model (LLM) tool they built using AutoGen, an agentic framework that leverages AI to interact with data and execute tasks.

The results were thorough and aligned closely with responses from actual experts, though Balloli noted some hidden biases in the LLM answers. 

“This is a promising tool, and it outperforms our expectations,” said Balloli. “But the caveat is, to build good systems, we need grounded resources from experts in specific fields to understand how AI can help and how AI can be used.”

Shanley Corvite, a PhD candidate in the U-M School of Information, examined the efficacy of Lettersmith, a digital tool for supporting writing, and how it aided first-year students writing cover letters.

Students were randomly assigned to one of five different experimental conditions. Four conditions included various combinations of the supports available in Lettersmith, such as a checklist of elements to include, example cover letters, and where to include different elements in a letter. The fifth condition provided students with the full Lettersmith experience and all available supports.

Corvite found that students writing cover letters with this full Lettersmith support showed the greatest gains for confidence in their writing.

“This [research] can hopefully inspire more work that explores how we can better support cognitive processes during structured professional writing,” Corvite said. 

Don Peurach, professor of educational policy, leadership, and innovation at Marsal Family School of Education and faculty lead for Marsal’s EdHub for Community and Professional Learning, gave attendees a peek behind the scenes of developing his online course series, “Transforming Education in an Interconnected World.” 

Peurach discussed the challenges of designing the series to teach both the technical capabilities needed to innovate and improve education, as well as the social skills required to nurture those efforts in a learning community. Surveys and observations showed that those studying the social learning components of the series needed additional support, such as discussion templates and team-building toolkits. Yet, while these resources supported meaningful interactions in the course and social media groups, they also drove an increased desire to learn in a face-to-face setting, which is difficult to facilitate in an asynchronous modality.

It’s a conundrum Peurach wants to solve so that learners continue to find value in community-supported learning. 

“They simply want to be in one place at the same time with people to process the experiences they’re having, to process the content they’re engaging, to wrestle with the problems that they’re experiencing,” Peurach said. 

Award-Winning Insights into Learning

The poster fair included a wide range of research projects. Judges reviewed the submitted posters using criteria that included having an innovative topic or methodology, potential impact for higher education, and clarity of findings. Three award winners were named.

Zhonghan ‘Jonas’ Xie, a second-year Master of Public Policy and Master of Science in Information student with the Ford School of Public Policy, won for his ongoing study, “Outcomes of Team Formation Changes in Tandem.” Xie’s research looks at how the removal of gender and race/ethnicity data as a consideration when forming teams using Tandem is impacting team formations and experiences. Tandem is an online tool that helps instructors create, manage, and support student teams. 

Xie found that the elimination of this data led to an increase in the number of “stranded” teams, those with one underrepresented minority or one female. Further, students in gender- and racially-diverse teams reported more positive team experiences overall than those on less diverse teams. 

“Tandem is effective in generating diverse teams,” Xie said. “What is really interesting is how dropping the race and ethnicity information [can] really affect the team’s outcomes.” The team is continuing to investigate these outcomes this summer. 

Esther Suh, a third-year orthodontics resident, based her research on her experience utilizing augmented reality to improve injection skills, as opposed to past methods involving practicing on dental models or fellow students. 

Suh’s prize-winning project showed how using an AR device to guide the placement of the syringe helped novice dental students improve their injection success rate to an advanced level in just four attempts. 

“[It] was interesting to see how quickly they picked it up,” said Suh, who earned her dental degree and completed undergraduate studies in neuroscience and biomolecular science at U-M. “That was really incredible.” 

Vibhavari Vempala, a PhD candidate in Engineering Education Research in the College of Engineering, was awarded a prize for her study of students’ self-efficacy in large STEM courses. She explored how students’ sources of self-efficacy, particularly from social persuasions and vicarious experiences through others, changed over the course of a semester. Her conclusions build on past findings that women pursuing STEM careers are most persuaded by friends and peers, while men draw their confidence more so from instructors and mastery experiences. 

Vempala’s hope is that instructors use support strategies to foster self-efficacy in students and improve STEM persistence.

“If we’re aware of perceptions that already exist, I think maybe the instructors could do something in the beginning of the course, like setting expectations, letting [students] know it’s not as scary,” Vempala said.