Why depression rates appear to be rising: a methodology problem, partly

Self-reported depression has risen substantially in the past decade and a half, particularly among adolescents and young adults. The U.S. National Survey on Drug Use and Health reports the rate of major depressive episode in 12-17-year-olds went from 8.7% in 2005 to 17.0% in 2020 — roughly doubling (SAMHSA, 2021).

The increase is real. The interpretation is contested, in ways that matter for understanding what's actually happening.

1. What the surveys measure

National surveys of depression don't measure depression directly. They administer screening instruments — questionnaires like the PHQ-9 or the K6 — and apply a cutoff to convert continuous scores into a binary "depressed / not depressed."

The cutoffs are designed to flag probable depression for clinical follow-up. They are not diagnostic. A person who scores above cutoff on the PHQ-9 has elevated probability of meeting clinical criteria but has not been diagnosed by a clinician.

This matters because what changes over time may include:

• Actual incidence of clinical depression
• Willingness to endorse depressive symptoms on surveys
• Cultural fluency with mental-health vocabulary
• Recognition of symptoms previously dismissed as normal adolescence

These can all increase the survey rate without the underlying clinical rate changing identically.

2. The convergent indicators

Some indicators that don't rely on self-report show similar trajectories: hospitalization rates for self-harm, antidepressant prescriptions, completed suicides among adolescents. These have all risen in roughly the same period, supporting the conclusion that the rise is partially real, not entirely a measurement artifact (Twenge et al., 2018).

But the rises in these indicators are smaller than the rises in self-reported symptoms. The doubling of self-reported depression corresponds to perhaps a 30-50% rise in objective indicators. The difference suggests both real increase and measurement-shift.

3. The smartphone hypothesis

Jean Twenge's iGen (2017) and subsequent work argue that smartphone adoption and social media use are causally linked to the adolescent depression rise. The correlational evidence is strong; the causal evidence is contested.

Several counter-arguments:

• The correlation between social media use and depression is modest at the individual level (r ≈ 0.15)
• Cross-country comparisons don't always show the predicted pattern
• The trend lines started before smartphone saturation in some cohorts

The current state of the literature: smartphones probably contribute to the adolescent mental health decline, particularly for heavy users and girls, but the effect is smaller than headline-level claims suggest, and other factors (academic pressure, economic precarity, climate anxiety, COVID disruption) likely also contribute (Orben et al., 2019; Haidt, 2024).

4. The cohort vs. period question

Epidemiologists distinguish cohort effects (something specific to a generation born in a certain era) from period effects (something affecting everyone alive at a certain time). The post-2010 mental health rise appears partly cohort-specific (heaviest in adolescents) and partly period-driven (affecting most age groups, just less).

This complicates causal stories. If it were purely smartphone-driven, we'd expect the effect to track screen time across age groups. If it were purely cultural shift in reporting, we'd expect uniform effects. The actual pattern — sharper in younger cohorts but present across age groups — is consistent with multiple contributing factors operating on different time scales.

5. The honest summary

Depression rates in the U.S. and several other industrialized countries have risen since 2010. The rise is partially real and partially a methodological/cultural shift in reporting. The causal story is contested, but smartphones, social media, academic pressure, and economic conditions all contribute in proportions that aren't yet well-pinned down.

The clinical implications are not affected by the methodological debate. Whether the rise is 30% or 80%, the absolute number of people experiencing depression-like symptoms is large enough to be a major public health concern. The treatment infrastructure has not kept pace. That part is not contested.