Social media and health: Evidence from college students

The study implements a randomized survey experiment in which participants are exposed to simulated social media scenarios. Individuals are randomly assigned to either a treatment group, which views short and realistic online interactions with potential emotional content (e.g., lack of social feedback, negative comments, exposure to idealized images, or AI-generated content), or to a placebo group exposed to emotionally neutral online content. Immediately after exposure, participants complete a standardized measure of state anxiety. The intervention is designed to capture the short-term psychological effects of specific digital experiences.

The intervention consists of an individual-level randomized survey experiment administered in person to first-year university students. After completing baseline questions on demographics and social media use, each participant is randomly assigned to one experimental condition and exposed to a single hypothetical but realistic social media scenario presented in written and visual form. Treatment scenarios are designed to mirror common online interactions with potential psychological salience, including: (i) posting content that receives no social feedback, (ii) receiving negative comments on a post, (iii) viewing highly idealized bodies with high engagement metrics, (iv) encountering AI-generated content depicting oneself, and (v) experiencing social exclusion through non-responses from peers. Placebo scenarios involve online content expected to be emotionally neutral, such as simple tutorials or generic advertisements.

Immediately following exposure, participants complete the state subscale of the State-Trait Anxiety Inventory (STAI-20), which captures momentary anxiety. Random assignment ensures comparability between treatment and placebo groups, allowing causal identification of short-term effects through differences in post-exposure anxiety levels. The design also allows for the analysis of heterogeneous effects by gender. Participants displaying severe baseline distress are excluded from the analysis, and mental health support resources are provided at the end of the survey.

2026-02-09

2026-02-16

The primary outcome is participants’ level of state anxiety measured immediately after exposure to the experimental scenario. Anxiety is assessed using the state subscale of the State-Trait Anxiety Inventory (STAI-20), a validated psychometric instrument that captures transitory anxiety. The outcome is constructed from participants’ responses to the STAI-20 items, with higher values indicating higher levels of anxiety.

The primary outcome, state anxiety, will be constructed using responses to the 20-item state subscale of the State-Trait Anxiety Inventory (STAI). Items phrased in a positive or calm direction will be reverse-coded so that higher values consistently indicate higher anxiety. The anxiety measure will be obtained by summing the recoded responses across the 20 items, generating a single composite score, with higher values indicating higher levels of state anxiety. For ease of interpretation, this score may also be standardized using the mean and standard deviation of the placebo group in some analyses.

The study uses an individual-level randomized survey experiment. Participants are randomly assigned to either a treatment group or a placebo group. Treatment participants are exposed to a single simulated social media scenario with potential emotional content, while placebo participants are exposed to an emotionally neutral online scenario. Randomization is conducted at the student level, and each participant is exposed to only one scenario. Outcomes are measured immediately after exposure using a validated instrument for state anxiety. The experimental design allows for causal identification through comparisons of post-exposure outcomes between treatment and placebo groups.

The study is implemented as an in-person, individual-level randomized survey experiment conducted among first-year university students at the Facultad de Ciencias Económicas of the Universidad Nacional de La Plata. Data collection takes place during the first semester of 2026, at the beginning of the academic year. The survey is administered in classrooms during a mandatory introductory course, ensuring access to a large and comparable cohort and minimizing attrition.

After completing baseline questions on demographics, social media use, and general digital access, participants are randomly assigned to one experimental condition. Randomization occurs at the individual student level, and each participant is exposed to a single hypothetical but realistic social media scenario. Treatment scenarios are designed to reflect common online interactions with potential psychological salience, including lack of social feedback, negative peer interactions, exposure to highly idealized content, AI-generated media, and experiences of online social exclusion. Placebo scenarios involve online content expected to be emotionally neutral, such as simple tutorials or generic advertisements.

Immediately following exposure, all participants complete the state subscale of the State-Trait Anxiety Inventory (STAI-20). The primary outcome is constructed by reverse-coding positively worded items and summing responses across the 20 items, generating a composite measure of state anxiety. Random assignment ensures comparability between treatment and placebo groups, allowing causal identification of short-term effects through differences in post-exposure anxiety levels. The design also permits the analysis of heterogeneous effects by gender.

Participants reporting severe psychological distress at baseline are excluded from the analysis, and all participants receive information about available mental health support services at the end of the survey.

Randomization is conducted automatically by the survey platform. Participants are randomly assigned at the individual level to the treatment or placebo group by the platform’s built-in randomization algorithm at the time the survey is administered.

The unit of randomization is the individual participant (student). Randomization is implemented within classrooms, but assignment to treatment or placebo occurs at the individual level through the survey platform, with no group- or classroom-level randomization.

No

Approximately 2,500 individual participants (individual-level units; no clustering).

Approximately 2,500 individual participants, all first-year university students.

Approximately 1,250 individual participants assigned to the treatment group and approximately 1,250 individual participants assigned to the placebo group.