Methodology & References
Last updated
This page documents the methodology behind the Studies index. It is organized into three sections: 1. Inclusion Criteria (what we include and why), 2. Harm Taxonomy (how we classify what each survey measures), and 3. Source Files (where to find the primary-source citations and files). Per-study citations live on each study’s own sources page, linked from the Studies page.
Inclusion Criteria
This is a living evidence map — we update it as new surveys are published, and we welcome contributions and corrections. The criteria below define what we include. We are working on expanding data, harm and population coverage over time.
What’s included. Population surveys of minors (under 18) that collect primary quantitative data on at least one category in our taxonomy — 7 harms plus 1 risk factor (underage use). We include probability samples and mixed probability/non-probability samples, as well as weighted non-probability samples designed to represent a defined population. We also include surveys of adults that ask them retrospectively about experiences before age 18. We exclude meta-analyses and systematic reviews, qualitative-only studies, and self-selected surveys.
Measurement coverage and column definitions. The harm, offline, and per-app columns in the index summarize what the survey asked, even if the data for these dimensions are not fully available. The per-app count includes distinct branded apps for which the survey measures an included harm or risk factor in the population shown in the index. Categories such as online gaming and text messaging are excluded. N is the sample size for the population shown in each index entry, which may be a subgroup of the full survey sample. The number responding to individual questions may be smaller. Year is the publication year of the survey report or questionnaire. It may differ from the survey’s fielding date.
National index
- Geography: surveys designed to represent defined national populations in the US, UK, or Australia, including population subgroups. Multi-country surveys qualify if they report US/UK/AU estimates.
- Recency: published 2022 or later. For a repeated survey series, once it qualifies we include all available waves (including earlier ones) for trend value.
- Coverage: measures at least one harm or risk factor.
US state index
- Geography: US state-level population surveys.
- Recency: published 2022 or later; for a repeated series, the latest wave only.
- Coverage: measures at least 2 distinct taxonomy categories. States measuring only cyberbullying via the CDC Youth Risk Behavior Survey (YRBS) aren’t shown here.
State surveys are numerous and overwhelmingly bullying-only via the CDC YRBS, so the higher two-category bar surfaces the states that add something distinctive.
Harm Taxonomy
Every study’s measures are classified into one of these 8 categories (7 harms and one risk factor, underage use). The categories draw on cross-sector expert recommendations.
Conduct & contact
Interpersonal harms: things that happen to the respondent, typically due to the directed actions of another person.
BullyingThe respondent is on the receiving end of hostile, humiliating, threatening, or exclusionary behavior online, including name-calling, threats, rumor-spreading, harassment, or doxing. Surveys that ask about only witnessing bullying (vs. being the target of it) are excluded; surveys that ask about both (e.g. "seen or experienced") are included.
Source pattern: "Has anyone ever done X to you online" — insults, threats, rumors spread, exclusion.
Excludes: Receiving sexual messages (→ Contact).
ContactThe respondent is contacted, messaged, solicited, or pressured by another person online in ways they do not want, often of a sexual nature — unsolicited sexual content sent to them, grooming-style approaches.
Source pattern: "Strangers try to contact me…", "had someone send you explicit images you did not ask for", "had someone constantly ask where you were", "asked me to send a nude".
Excludes: Non-consensual creation or sharing of intimate imagery of the respondent, threats to share it, and coercive requests to provide intimate images (→ NCII).
NCIINon-Consensual Intimate Imagery. The respondent is the subject of a nude / sexually explicit image or video that was taken, made, shared, threatened to be shared, or distributed without their consent. Includes AI-generated / deepfake nude imagery of the respondent. Coercive requests to provide nude or sexually explicit images or videos are also included, even if no image was created or shared.
Source pattern: "Has anyone ever shared / threatened to share a sexual image of you…".
Excludes: Receiving someone else's nudes unsolicited (→ Contact).
Content
Respondent saw content that may be harmful, without requiring directed action of another person.
Self-HarmRespondent saw content depicting, encouraging, or providing instructions for suicide or self-harm.
Source pattern: "I see/hear things about suicide or self-harm that are upsetting to me"; "Suicide: people showing this, giving instructions for this".
Eating DisorderRespondent saw content depicting, encouraging, or providing instructions for eating disorders.
Source pattern: "Eating disorder content"; "Showing someone with an eating disorder, or encouraging or providing instructions for one".
ViolenceRespondent saw content depicting real or threatened physical violence against people (or in some studies, animals). Includes weapons, gore, dangerous stunts/challenges, encouragement of violence.
Source pattern: "Someone being seriously violent towards another person"; "Dangerous stunts"; "Real animals being seriously hurt".
Problematic Use
Consequences of use, or perceptions of satisfaction with or control over time spent online.
Problematic UseThe instrument must ask about a consequence of use, such as displacement or behavioral dependence, or about perceptions of use, such as satisfaction with or control over time spent online. Questions about frequency or duration alone are excluded.
Source pattern: "Using it gets in the way of my sleep"; "I feel addicted or end up using it longer than I wanted to"; "Looking at this scale, how happy are you… With the amount of time that you spend online?".
Excludes: Questions about frequency or duration alone, including "almost constantly" or generic hours per day, without consequences or perceptions of use.
Underage Use (risk factor)
Use of age-restricted platforms by those below the minimum age.
Underage UseUsage of applications by those who are younger than allowed by policy or law. Operationalized by combining the respondent's age with whether they hold an account on / actively use a platform whose Terms of Service or minimum-age policy disallows their age group.
Source pattern: Usage rates for ages below the platform's minimum-age policy (e.g., 9–12 use of 13+ platforms). Explicit "underage account ownership" questions.
Source Files
Every field shown for a study on the Studies page is backed by a primary-source citation. Those citations — with the source file, page number, verbatim text, retrieval URL, and SHA-256 hash — live on each study’s own sources page (click a study’s title, marked with a icon, in the Studies table).
To download every cited source file at once: all sources (ZIP).