U.S. adult survey · 2024 evidence
Health misinformation on social media: what adults think.
A survey can measure how people judge the information they see. That is different from counting verified false claims.
of U.S. adults selected “A lot” or “Some” when asked how much of the health information they see on social media they think is false or misleading.
Civilian noninstitutionalized adults 18+ · 2024 HINTS · All participants, including nonusers and missing answers
Social Usage analysis of NCI HINTS 7 public-use data, B13 / MisleadingHealthInfo, responses 1–2. Sampling SE: 0.98 percentage points. Fieldwork March 25–September 16, 2024. NCI does not endorse this presentation.
This is not the percentage of social-media posts that are false. The survey records respondents’ perceptions of health information.
What people said about the information they see
Exact question: “How much of the health information that you see on social media do you think is false or misleading?”
Source: NCI public codebook, B13, PDF p85 (printed p84), responses 1–5; original Social Usage chart. Accessible table and sampling errors.
Missing answers: 0.7% weighted. Nonuse is shown separately from judgments about accuracy.
“Some” is the largest individual response category. The combined headline uses original survey weights for “A lot” and “Some”; it does not treat those answers as counts of false posts or as evidence that respondents believed a false claim.
View the data table and uncertainty
| Response | Weighted % | SE (pp) | Relative SE (%) | Response n |
|---|---|---|---|---|
| A lot | 32.2 | 1.03 | 3.19 | 2,179 |
| Some | 38.4 | 1.04 | 2.70 | 2,700 |
| A little | 14.6 | 0.68 | 4.64 | 974 |
| None | 1.7 | 0.29 | 16.72 | 137 |
| I do not use social media | 12.4 | 0.51 | 4.12 | 1,235 |
| A lot or Some — combined analysis | 70.6 | 0.98 | 1.39 | 4,879 |
The five response percentages are NCI codebook values. The combined percentage and all standard errors are Social Usage calculations. The combined row overlaps the first two rows; it is not another response category. n is unweighted; percentages use survey weights.
SE is a sampling standard error in percentage points, not a 95% margin of error. No confidence interval or significance test is reported. 53 participants had a missing answer, a weighted 0.7%. They are not coded “None” or nonusers; displayed rounding need not total 100%.
Public codebook B13, PDF p85; Methodology Report §5.5, 50 JK1 replicate weights.
Perception, exposure and belief ask different questions
This finding describes perceived amount of false or misleading health information. It does not measure how often someone encountered a false claim, whether a fact-checker judged a post false, whether they believed it, or whether they shared it. Those require separate evidence.
Why include nonusers in the denominator?
The source codebook reports B13 against all survey participants. “I do not use social media” is an explicit response option, not missing data. Keeping it preserves NCI’s denominator. Excluding it would answer a different question about a selected population. We have not rebased this chart to social-media users or valid answers alone.
The nonuse answer is also different from the separate past-12-month Never-visits measure. Different wording can produce different responses; neither figure establishes account ownership.
What period does the question cover?
B13 has no explicit recall window. It asks for a perception at the time of the survey. The past-12-month wording in the preceding health-information activity questions does not carry into B13.
Can this describe political misinformation or a particular platform?
No. The question is about health information across social media, with no selected platform, political-content, demographic or historical comparison here. It cannot estimate all misinformation online or the accuracy of a particular feed. It supplies no measure of intentional deception, medical outcomes or causal effects.
Source, calculation and review
Credit NCI for the original category estimates and Social Usage for the combined estimate and sampling errors. Keep the 2024 fieldwork, health-information scope and adult denominator in a citation.
Survey design and calculation
HINTS uses an address-based probability sample of U.S. civilian noninstitutionalized adults. One adult per household is selected by the next-birthday method; the survey is postal with a web option. Final person weights account for selection, nonresponse and population calibration.
For “A lot or Some,” sum the final weights for B13 codes 1 and 2 and divide by the weight sum for all 7,278 participants, then multiply by 100. Nonuse and missing responses stay in the denominator. Use original weights, not a sum of rounded published percentages.
Recompute numerator and denominator for each of the 50 replicate weights. SE = √[0.98 × sum of (replicate percentage − full-weight percentage)²]. The combined SE uses the joint indicator, not added category errors.
Our publication screen requires at least 50 selected and 50 complementary responses, total n ≥ 100 and relative SE ≤ 30%. All six estimates pass. These are Social Usage safeguards, not NCI thresholds. Sampling errors do not capture every source of response, interpretation, coverage or nonresponse error.
Annotated English instrument B13, PDF p6 (printed p5) · HINTS methodology, chapter 1 and §5.5 · Analysis Recommendations PDF pp4–9 in the public package.
Data period, source version and dates
- Fieldwork
- March 25–September 16, 2024
- Question window
- No explicit recall period in B13
- Source package update
- August 2025; exact publication day unknown
- Original statistical retrieval
- October 1, 2026
- Assisted analysis review
- 2026-10-07
Retained source data and policy evidence were reused. Preparing this guide does not make the survey current to its review date. Assisted technical checks are not independent human editorial approval.
Citation and reuse
This guide offers no CSV, image export or embed. We publish selected aggregate facts and our own analysis, not respondent records, source artwork or agency logos. NCI does not endorse Social Usage. The original federal source is freely available through the links above.