Case Study 1: The Haidt-Orben Debate — Two Honest Scholars, Two Different Conclusions

Case Study 1: The Haidt-Orben Debate — Two Honest Scholars, Two Different Conclusions

The Public Exchange

The Haidt-Orben debate has played out in journal articles, public lectures, media interviews, and (fittingly) on social media. It is one of the most visible scientific disagreements in contemporary psychology — and it illustrates both the strength and the difficulty of the scientific process.

Haidt's Evidence

Temporal correlation. Haidt emphasizes the striking coincidence between smartphone adoption (2010–2015) and the sharp increase in youth mental health problems. He notes that the increase is steeper and more sudden than what normal social changes produce — suggesting a discrete cause (a new technology) rather than gradual factors (economic changes, cultural shifts).

Gender differences. The increase is steeper among girls, consistent with mechanisms related to social comparison and appearance-based platforms (Instagram). Haidt argues this gender specificity points to social media as the cause, since other factors (the economy, education policy) would affect boys and girls similarly.

Cross-national data. Similar trends appear across multiple countries that adopted smartphones around the same time. Haidt argues this makes country-specific explanations (U.S. politics, UK austerity) insufficient.

Facebook Files. Meta's internal research showing Instagram's harmful effects on teen girls' body image provides "insider" evidence.

Converging evidence. Haidt uses the "converging evidence" argument: no single data point is conclusive, but taken together — timing, mechanisms, gender pattern, cross-national consistency, corporate documents — the case is strong.

Orben/Przybylski's Counterarguments

Effect size. The individual-level association is tiny (r ≈ 0.04 in the largest analyses). At the population level, this means social media explains less than 0.5% of the variance in wellbeing. Orben argues that building policy on such a small effect is premature.

Analytical sensitivity. The specification curve analysis shows the result is highly sensitive to analytical choices. Depending on which measures, covariates, and samples are used, the association ranges from meaningfully negative to slightly positive. This suggests the "finding" is fragile.

Correlation ≠ causation. Temporal correlations are weak evidence for causation. Many things changed between 2010 and 2015 (economic recovery from 2008, opioid crisis acceleration, political polarization, academic pressure increases, COVID precursors). Singling out smartphones requires ruling out alternatives, which hasn't been done.

Reverse causation. Depressed teens may seek out social media more (escape, connection-seeking, passive coping). The few longitudinal studies that examine directionality produce mixed results.

Selective interpretation. Orben argues that Haidt's "converging evidence" approach is vulnerable to cherry-picking — selecting evidence that supports the hypothesis while downweighting contradictory evidence.

What Makes This a Genuine Scientific Disagreement

This debate is NOT a case of one side having clearly better evidence. It is a genuine disagreement about:

How to weigh different types of evidence. Haidt weights temporal trends and convergence; Orben weights individual-level effect sizes and methodological rigor.
What constitutes sufficient evidence for action. Haidt argues that the precautionary principle justifies action despite uncertainty; Orben argues that premature action based on weak evidence may be counterproductive.
What level of risk warrants concern. A small effect (r = 0.04) at the population level might still translate into thousands of affected individuals — but it also means the vast majority of social media users are unaffected.
Who bears the burden of proof. Haidt argues the burden is on those claiming social media is safe; Orben argues the burden is on those claiming it's harmful.

The Meta-Lesson

The Haidt-Orben debate is valuable not because one side is right and the other wrong, but because it demonstrates what honest scientific disagreement looks like:

Both sides cite real data
Both sides have legitimate methodological points
Both sides acknowledge uncertainty (though they resolve it differently)
The disagreement is about interpretation, not fabrication

This is how science is supposed to work. The popular version — "social media is destroying teens, period" — replaces a genuine debate with false certainty. The toolkit's response: sit with the uncertainty, evaluate the evidence on both sides, and make decisions based on your best judgment while acknowledging what isn't known.

Discussion Questions

If you had to advise a policymaker based on the current evidence, would you recommend restrictions on adolescent social media use? What level of evidence should be required for policy action?
Haidt's "converging evidence" approach and Orben's "effect size" approach represent two different philosophies of evidence evaluation. Which do you find more persuasive, and why?
The debate has been conducted partly on social media (Twitter, Substack). Is this ironic? Does the public nature of the debate improve or complicate the scientific process?
If the true effect is very small (r = 0.04), should we spend research resources on social media's effects, or redirect them to factors with larger effects (sleep, exercise, poverty)?