Following the analysis of YouTube videos I need to perform for a study, I found an interesting article that might advance my work. As Science needs to be told, YouTube can be a powerful tool to make it accesible to everybody. But what makes a science video catch attention and spark engagement? A study published in Plos One gives concrete insights based on real analytics.
In this article, researchers analysed over 200 videos from the American Chemical Society’s channel Reactions to see how factors like video length and social media metrics (likes, comments, etc.) influence how people interact with science content online. In this post, I’ll try to walk you through what I understood =)
Yang, S., Brossard, D., Scheufele, D. A., & Xenos, M. A. (2022). The science of YouTube: What factors influence user engagement with online science videos?. Plos one, 17(5), e0267697.
https://doi.org/10.1371/journal.pone.0267697
Why YouTube for Science Communication?
In the past, traditional media was the main channel of communication for “important things” like science, technologies and discoveries. Nowadays, many scientific institutions are turning to digital platforms to reach audiences directly. YouTube is the main channel for that, as it stands our for its reach and capacity for user interaction.
Metrics of engagement (views, likes, comments, shares, watch time, etc.) are key to understanding how audiences respond (just like audience share in traditional media), and to identifying which specific features correlate with those indicators.
The Study:
The researchers used analytics data provided by the American Chemical Society (after having the approval from the Univesity of Wisconsin-Madison Institutional Review Board) from the Reactions channel mentioned above. This channel publishes videos about chemistry in everyday life.
The dataset contained the information of 210 videos published bettween 2014 and mid-2017. They examined multiple variables on a day-to-day basis:
- Video characteristics: title, creation date, a revenue, annotations shown, number of cards shown, etc.
- Audience characteristics: age, gender, country, subscription status, subscription source type, access device, access operating system, traffic source type, use of translation, use of subtitles and closed captioning, sharing service, sharing device, etc.
- Engagement activities: views, average view duration, average percentage viewed, likes, dislikes, shares, comments, subscribing and unsubscribing, adding to and removing from playlists, etc.
To analyse this metrics, the researchers applied several statistical models (e.g., negative binomial regression, or OLS).
Key Findings:
- Video length matters: longer videos tend to receive fewer visits. Each additional minute was associated with a 15% drop in view rate. However, when viewers do stick around, they spend more total time, albeit watching a smaller percentage of the video.
- Likes are powerful: likes show strong positive correlations with nearly all engagement metrics (more views, more comments, more shares, more subscribers, and more watch time).
- Dislikes have mixed effects: dislikes correlated negatively with some metrics (e.g., gained subscribers or percentage viewed), but were associated with more comments. Likely reflecting controversy or the need to express opinions.
- Young, mobile audiences are more active: younger viewers (18-44) and male viewers tended to interact more. Mobile viewers were also more likely to “like” and share than those on other devices.
- Subscribers are more engaged: most views were from non-subscribers (showing strong potential to reach new audiences), but subscribers gave more likes and comments per view, watched longer, and were more likely to take extra actions (e.g., adding to playlists).
Practical Takeaways for Science Communicators:
- Keep videos concise and dynamic, grabbing attention early boosts view counts. User-generated videos performed better than professionally produced ones. Also, short clear and concise video titles drove greater user engagement.
- Encourage “likes” explicitly by asking viewers to interact, this can amplify visibility.
- Viral reach isn’t the same as meaningful engagement. High clicks can come with low retention.
- Design for mobile first, considering pace, visuals, and subtitles.
- Don’t fear the controversy as negative feedback can spark engagement (though it’s important to handle it smartly).
Limitations & What’s left to explore:
The study examined only one specialized channel, so the results may not generalize across all science or educational content, specially since is focused on a single theme (chemistry).
Also, while clear correlations emerged, casuality cannot be inferred: is the video’s quality that drives likes, or do likes amplify its reach?
Additionally, the study didn’t account for content quality, narrative style, or topic, all of which may play major roles.
Further studies could include other disciplines or run controlled experiments to better understand how social cues influence viewer behaviour.
Conclusions:
Science communication (though this applies to anything) isn’t just about crafting great content, it’s also about smart presentation and strategy. The right video length, visible “likes”, and an engaged audience can make the difference between a video that’s ignored and one that resonates deeply.
It is important to think not only what you say, but how you present it. A small change in duration, a subtle call to action, or a tweak for mobile can yield big results.
See you in the next paper =)