Driving STEM Engagement Via NOVA’s Twitch Channel

sisters watching shows on a laptop together in their bedroom

Problem

The National Science Foundation (NSF) wanted to explore whether livestreaming platforms foster more inclusive STEM learning and community building.

In celebration of its 50th anniversary, NOVA launched Building Stuff, a multi-platform initiative designed to highlight the essential role of engineering in shaping the world around us. Central to this initiative was the debut of NOVA’s first-ever Twitch channel, Building Stuff with NOVA, hosted by professional engineer and former NASA researcher Dr. Nehemiah Mabry (“Dr. Nee”). The channel featured weekday livestreams that blended gameplay, expert interviews, virtual field trips, and collaborative challenge—including an engineering-themed escape room designed with audience input.

The initiative aimed to reach new audiences by presenting engineering content in an inclusive, approachable, and interactive format. Through Twitch chat, viewers could ask questions, vote on ideas, and influence the direction of the content in real time. The project culminated in a four-night live escape room event at GBH studios in Boston, where online and in-person participants worked together to solve engineering puzzles.

This innovative approach raised important questions about the potential of livestreaming platforms to support STEM learning, foster community, and promote science identity and belonging. With major support from the NSF, the initiative sought to explore whether digital engagement could serve as a meaningful pathway into STEM for viewers who might not otherwise see themselves reflected in traditional science media.

Solution

NORC analyzed viewer behavior and chat data to assess STEM engagement and community-building.

To assess the impact of NOVA’s Building Stuff Twitch initiative, Slover Linett at NORC and NORC’s Social Data Collaboratory designed and implemented a mixed-methods research study focused on understanding how interactive livestreaming can support STEM engagement, learning, and community-building. The research team collected and analyzed data from 66 Twitch videos and 13 YouTube streams, capturing over 10,000 chat messages from more than 450 unique users.

NORC’s approach combined quantitative and qualitative methodologies. Quantitatively, the team conducted multivariate statistical modeling to identify predictors of high engagement, including chat activity, viewer sentiment, and STEM-related discourse. Qualitatively, researchers coded chat messages to identify patterns in emotional tone, peer support, host interaction, and collaborative learning. The team also examined differences in engagement across content formats—such as interviews, field trips, and collaborations with other streamers—and between audience segments, including NOVA affiliates and organic viewers.

This rigorous, multi-layered approach allowed NORC to explore what kinds of content and behaviors were associated with higher engagement and how those dynamics might support broader goals like science identity, belonging, and inclusive participation in STEM.

Result

We found that interactive formats boosted STEM engagement and fostered a supportive learning community.

NORC’s analysis revealed that Building Stuff with NOVA successfully fostered STEM engagement through emotionally resonant, inclusive content. Streams that featured peer support and collaborative learning moments were strongly associated with higher chat activity and STEM-related discussions. For example, collaborative learning moments tripled the odds of high chat engagement, and peer support messages were linked to 3.1 times higher odds of above-median chat activity.

Content format played a key role in driving engagement. Interview and field trip streams were associated with significantly higher viewership—12.4 and 16.5 times higher odds, respectively, compared to standard gaming streams. Collaborations with other streamers also showed promising potential, with early analyses suggesting a fivefold increase in viewership and chat engagement.

The research also highlighted the strategic role of NOVA affiliates, who, despite comprising only 4.55 percent of chatters, generated nearly a quarter of all chat messages. Their interactions modeled host engagement and helped cultivate a positive, emotionally rich learning environment. Positive sentiment and emotional impact were strongly correlated with peer support and STEM discourse, while even negative messages often coincided with constructive learning moments.

These findings demonstrate the potential of livestreaming platforms to support inclusive STEM learning and community-building. They offer actionable insights for science communicators, educators, and funders seeking to engage diverse audiences in meaningful, participatory ways.