Let’s keep digging the YouTube Science Communication rabbit hole. This post analyses the study by Velho, Mendes, and Azevedo (2020), which examines what factors drive differences in the popularity of YouTube videos. While some videos take off, many remain with modest view counts. To explore this, the authors analysed ScienceVlogs Brasil, a network of independent science YouTube channels. The goal of the study was to identify which features (content-related and non-content factors) are correlated with a video’s popularity, defined as views per day since posting.
Velho, R. M., Mendes, A. M. F., & Azevedo, C. L. N. (2020). Communicating science with YouTube videos: How nine factors relate to and affect video views. Frontiers in Communication, 5, 567606.
https://doi.org/10.3389/fcomm.2020.567606
Why do some science videos become popular and others do not?
YouTube plays a particularly powerful role in shaping how audiences encounter and engage with scientific content. Previous research has highlighted the platform’s algorithmic shifts (from prioritising clicks to favouring watch time) and identified a strong “rich–get–richer” dynamic, in which early engagement drives further visibility. Other research has shown that certain content-agnostic factors (e.g., upload timing or channel size) can influence view counts. In the specific field of science communication, typologies have emphasised the prominence of formats like vlogs, animations and short documentaries. However, there has been little empirical evidence on how these formats interact with audience reach.
While some studies have explored the typology and narrative structure of science videos, fewer studies have focused on why some of these videos become popular. This question is particularly relevant in an online environment where misinformation spreads easily, making visibility a critical factor for credible science communication. For these reasons, this study builds on an examination of nine factors to better understand their relationship with science video popularity.
The Method:
The study sampled 441 semi-randomly selected videos from 33 science communication channels within the ScienceVlogs Brasil project (a network of around 60 channels covering a broad variety of themes). The authors categorised each video along nine dimensions:
- Content-related factors:
- Video theme: Earth and Exact Sciences, Biological Sciences, Engineering, Health Sciences, Agricultural Sciences, Applied Social Sciences, Humanities, “Linguistics, Languages & Arts”, Interdisciplinary.
- Video format: Vlog, Interview, Short Documentary or Reportage, Hangout, Video animation, Live conversations, Commented video, and Talk.
- Number of Editing features: [0–9 points] Sound effects, Image effects, Video effects, Logo or Vignette, Filters, Fast-forward technique, Jump-cut technique, Stop-motion technique, and Green-screen.
- Content-agnostic factors:
- Video length: in minutes
- Video age: in days
- Channel productivity: number of videos divided by the number of months of life of the channel
- Number of likes: per video
- Number of comments: per video
- Channel identity.
A combination of descriptive statistics and multiple linear regression analysis was employed to examine the relationship between these factors and video popularity. To control for time effects, the dependent variable was log-transformed views per day rather than absolute view counts.
Key Findings:
Descriptive analysis revealed a few notable patterns:
- Interdisciplinary, Earth and Exact Sciences, and Biological Science videos tended to attract more views per day, though the relationship was not statistically significant.
- Vlogs, Animations, and Group conversations were associated with higher media popularity, while Interviews and Hangouts tended to perform worse.
- Likes showed a moderate positive correlation with views, whereas comments displayed only a weak correlation (this was also highlighted on one of our previous analysis).
- Newer videos tended to collect more daily views than older ones.
- A small number of channels accounted for a disproportionately large share of total views.
Inferential analysis sharpened these findings. The variables with significant predictive power were:
- Number of likes: positive effect
- Channel productivity: positive effect
- Video age: negative effect
- Video format: Animations outperforming Vlogs, and Interviews and Hangouts performing worse.
Key Takeaways:
This study show several practical insights:
- Early engagement matters: videos that attract likes quickly are more likely to be amplified by YouTube’s recommendation system, creating a “rich–get–richer” effect.
- Consistent production pays off: channels with higher productivity are more visible, likely because they generate more opportunities for audience exposure and algorithmic recommendation.
- Format choices influence reach: shorts, animated explainers and vlog-style videos tend to perform better than longer, less dynamic formats such as interviews or hangouts.
- Recency boosts visibility: newer content seems to enjoy a greater algorithmic advantage.
Limitations & What’s left to explore:
The study acknowledges several limitations:
- Only one researcher categorised themes and formats, which could introduce some bias.
- Other potentially revelevant factors (e.g., thumbnail quality, keyword tags, subscriber base) were not included.
- Including very recent videos may have inflated some popularity metrics.
Future research could address these gaps by employing automated content analysis, longitudinal designs, or audience-centred studies exploring how viewers interact with different formats and themes.
Conclusions:
This study provides a valuable empirical contribution to understanding how specific video and channel characteristics shape the popularity of science content on YouTube. Rather than relying solely on content quality or topic, science communicators may need to strategically consider engagement metrics, production regularity, and video format to reach broader audiences.
These insights are especially relevant in the current media landscape, where the ability to make credible science visible can counterbalance the spread of misinformation and help build more informed publics.
See you in the next paper =)