Reading a trend
Is U.S. social media use growing?
A growth rate needs at least two observations of the same measure and population.
CDC found no statistically significant change in hourly-or-more social-media use among U.S. high school students between 2023 and 2025. The matching estimates were 37% and 36%. This is a specific frequency comparison, not a universal social-media growth rate.
Source: CDC, Youth Health in Focus technical Table 44, page 47. U.S. students in grades 9–12, valid responses; about once an hour or more often. Source-reported p = 0.4799. Estimates, confidence intervals and question wording.
Source: CDC. Original materials are available free of charge on CDC.gov. Use by SocialUsage does not imply endorsement by CDC, HHS or the U.S. Government.
Keep the result in its survey context
The CDC result is a two-year comparison using the same published threshold. No significant change does not prove exact equality or establish a longer-term trend. Do not substitute an older several-times-daily headline for its hourly measure.
Our NTIA collection contains only November 2023; the Pew briefs concern a different survey in 2025. Those snapshots cannot be joined into a U.S. growth series.
The U.S. explorer compares supported demographic groups and states within NTIA’s single 2023 wave, with source standard errors. The U.S. platform age-gap report compares age groups within Pew’s 2025 survey. Those are differences between populations at a point in time, not changes over time.
Earlier data may exist at the source. Before connecting a historical series, we need to check its question, population, sampling and any method changes. A newer publication date alone is not another observation.
Before calculating a change
1. Check that the statistic means the same thing
Keep the activity rule and time window constant. Ever using a named app, using any social network during the past six months and having an account active during a month are different measures. A familiar label such as “user” is not enough to establish comparability.
2. Align the population and geography
Use the same age range, denominator and boundary in both periods. NTIA’s internet users aged 15+ cannot be replaced with Pew’s all-adult population aged 18+. The same caution applies to company-defined regions and national borders.
A change in the denominator can change a percentage even if the activity you care about has not changed.
3. Check methods, missing values and uncertainty
Inspect question wording, sampling, response modes, weights and source flags before joining observations. Missing is not zero. A break in series can include a measurement change and may prevent a continuous growth calculation.
Retain supplied standard errors or confidence information. A numerical difference alone does not establish statistical significance or explain why a change occurred.
4. Distinguish percentage points from percent change
For compatible percentages, the percentage-point difference is later value − earlier value. Relative percent change is (later value − earlier value) ÷ earlier value × 100. Relative change is undefined when the earlier value is zero.
A population share can change without an equivalent change in the number of people. Estimating a change in people would require a matching population and period, as well as a compatible measure.
A separate European example
See the EU participation series, 2023–2025
The published EU social-network participation estimate is higher in 2025 than in 2023. This describes a share of all EU individuals aged 16–74, not the growth of an individual platform or a U.S. audience.
Source: Eurostat, isoc_ci_ac_i. Private-purpose participation in the three months before the survey; percentage of all individuals aged 16–74.
One indicator, population and geography
Source: Eurostat, EU27_2020 / IND_TOTAL / I_IUSNET / PC_IND, 2023–2025. Chart rounded to one decimal. No break flag is attached to these aggregate observations in this release; country-level methods and breaks still warrant caution.
| Reference year | Share (%) | Flags |
|---|---|---|
| 2023 | 59.32 | None supplied |
| 2024 | 64.82 | None supplied |
| 2025 | 67.26 | None supplied |
What this series cannot tell us
These annual survey estimates do not follow the same individuals through time. They cannot identify why participation changed, how much time people spent online or which platform gained users. Confidence intervals are not supplied in this extract, so the pattern is descriptive.
Country-level comparisons need their own checks. France’s 2024 observations are confidential, and its 2025 observations carry a break in series. A continuous growth calculation is not justified by that extract.
France source example · Eurostat methods. Evidence retrieved and checked 1 October 2026 (UTC); data through 2025. Assisted evidence check; independent human review has not been recorded.
European dataset and immutable releases · Explore European comparisons