Appendix E: Key Studies Summary

The research that most directly informed this book — summarized accurately, with caveats where appropriate.

Attention and Memory (Part 1 Foundation)

Festinger, L. (1954). A theory of social comparison processes. What it found: Humans evaluate their own opinions and abilities by comparing to others, particularly when objective standards are unavailable. Comparisons with similar others are preferred; upward comparison (to those perceived as better) affects self-assessment more powerfully than downward comparison. Significance: The theoretical foundation for understanding why social media platforms that surface others' highlight reels affect viewer wellbeing through comparison. Still foundational 70 years later. Caveat: Conducted primarily with small groups in laboratory settings; application to mass media social comparison is extrapolated, not directly tested in the original.

Zeigarnik, B. (1927). Das Behalten erledigter und unerledigter Handlungen. What it found: Participants recalled interrupted tasks better than completed ones. The incomplete task creates sustained cognitive activation — a "need" to finish — that the completed task does not. Significance: The Zeigarnik Effect is the psychological mechanism behind open loops in creator content. Starting a question and delaying the answer creates cognitive tension that keeps viewers watching. Caveat: Replication history is mixed — the effect is real but smaller and more conditional than initial reporting suggested. The basic phenomenon is well-established; claims about its magnitude should be moderated.

Paivio, A. (1971). Dual Coding Theory. What it found: Information processed through both verbal and visual channels simultaneously is encoded more deeply and remembered better than information processed through one channel alone. Significance: The theoretical basis for why video outperforms text for information retention, and why graphics reinforcing voiceover improve learning outcomes. Caveat: Well-established and consistently replicated theory; individual differences exist in the degree of benefit.

Virality and Social Spread (Part 2 Foundation)

Granovetter, M. S. (1973). The strength of weak ties. What it found: In a study of job-seekers, people were more likely to have found jobs through acquaintances (weak ties) than through close friends (strong ties). Information travels through weak ties because strong ties connect clusters of people who already know the same things. Significance: The foundational research for why content crosses audience clusters through weak ties (collaboration, sharing beyond immediate network) rather than strong ties (existing friends sharing with their already-connected friends). Caveat: Original research on information spread in social networks; application to content virality is well-supported by subsequent network research but involves extrapolation.

Berger, J., & Milkman, K. L. (2012). What makes online content viral? What it found: Analysis of New York Times articles found that positive content spreads more than negative; practically useful content spreads widely; high-arousal emotions (awe, anger, anxiety) predict virality more than low-arousal emotions (sadness, contentment). Significance: The empirical research behind the viral content frameworks in Chapter 9. The arousal dimension is particularly counterintuitive — viral content is emotionally activating, not just emotionally positive. Caveat: Conducted on NYT content (high-quality news articles) — findings may not transfer directly to short-form video content. The arousal finding has been replicated in other contexts but with varying effect sizes.

Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. What it found: Analysis of 126,000 Twitter cascades from 2006-2017. False news was 70% more likely to be retweeted than true news, reached more people faster, and penetrated deeper into Twitter networks. The effect was driven by human sharing, not bots. Political content showed the strongest false-news advantage. Significance: The most comprehensive empirical study on misinformation spread on social media. The finding that false news has structural advantages over true news (because it's more novel and more emotionally activating) is the basis for Chapter 38's discussion of the epistemic responsibility of creators. Caveat: Conducted on Twitter only; findings may differ on other platforms with different algorithmic structures. The mechanisms identified (novelty, arousal) are well-grounded in prior research.

Social Comparison and Body Image (Chapter 38 Foundation)

Fardouly, J., Diedrichs, P. C., Vartanian, L. R., & Halliwell, E. (2015). Social comparisons on social media: The impact of Facebook on young women's body image concerns and mood. What it found: Women who browsed Facebook for 10 minutes reported higher body image concerns and more negative mood than women who browsed a control website. The effect was mediated by social comparison processes (comparing to Facebook profile photos). Significance: One of the earliest experimental studies demonstrating social media body image effects. Important because it uses an experimental design (randomly assigning participants to conditions) rather than just measuring correlation. Caveat: Small sample (n=112); laboratory condition (10 minutes of exposure) may not represent typical usage; Facebook is not the same as TikTok or Instagram Reels. The finding that social comparison mediates the effect is well-supported; the magnitude in real-world conditions is uncertain.

Orben, A., & Przybylski, A. K. (2019). The association between adolescent well-being and digital technology use. What it found: Reanalysis of three large datasets (over 355,000 adolescents). Found statistically significant but very small negative associations between digital technology use and wellbeing. Effect sizes were comparable to wearing eyeglasses or eating potatoes. Significance: An important corrective to alarmist media coverage of social media and mental health. The effect is real and negative but may be substantially smaller than headlines suggest. Caveat: Also observational/correlational — cannot establish causation. The critique of Twenge's larger effect size claims is methodologically valid, but some researchers dispute Orben and Przybylski's analytical choices as well. The debate continues.

Creator Mental Health and Psychology (Chapter 38 Foundation)

Variable Reinforcement Schedule (B.F. Skinner, multiple studies 1950s-1980s) What the research established: In conditioning experiments, unpredictable reward schedules (sometimes a reward, sometimes not) produce stronger and more persistent behavioral responses than fixed schedules. This finding has been extensively replicated across animal and human studies. Significance: The psychological mechanism behind compulsive analytics checking. The variable nature of content performance (sometimes good, sometimes not, with no reliable pattern) creates exactly the conditions that produce the strongest behavioral conditioning. Caveat: The slot machine analogy is useful but imprecise — social media dynamics are more complex than a simple Skinner box. The basic mechanism (variable reinforcement produces strong conditioning) is not disputed.

Network Science (Chapter 10 Foundation)

Watts, D. J., & Strogatz, S. H. (1998). Collective dynamics of 'small-world' networks. What it found: Mathematical analysis showing that many real-world networks (social networks, power grids, neural networks) share a "small world" property: high clustering (your friends know each other) combined with short path lengths between any two nodes (six degrees of separation). This structure makes information spread fast. Significance: The network science basis for why content can reach far beyond a creator's immediate audience through a small number of sharing steps. The "small world" structure of social networks means even a new creator is only a few hops from millions of people. Caveat: Mathematical/theoretical research; real social networks are more complex than the models. The basic structural findings are well-established.

Reading the Research Landscape

On social media research generally: The field is moving rapidly. Research published in 2015 reflects a different social media landscape than research published in 2023 — platforms have changed dramatically, usage patterns have changed, and the populations being studied have changed. Treat any specific finding from before 2020 with awareness that the platform context has shifted.

On effect sizes: Most effects in social media and wellbeing research are statistically significant but small in magnitude — meaning they're real effects that affect some people meaningfully, but that explaining any individual's experience requires much more than the social media variable alone. The complexity of human psychology resists simple causal narratives.

On what this means for creators: The research doesn't support the narrative that "social media causes depression" as a simple causal claim, nor does it support dismissing any effect entirely. What it supports: creating content with awareness that it enters a system with real, if complex, effects on viewers — and that the creator's choices about what that content amplifies matter.