Are placebo effects and expectancy bias overstating adolescent social media interventions?

How big are the real effects of social media on teens?

The headline answer: very small. A large experience-sampling study tracked 387 adolescents six times a day for three weeks—over 34,000 observations—and found that for 88% of teens, social media use had no or negligible effects on self-esteem (effect sizes between -0.10 and +0.10, which is tiny) [2]. Only 4% showed a positive effect and 8% a negative effect, and even those were small [2]. This means that for the vast majority of adolescents, social media use simply doesn't move the needle on well-being in either direction.

A 2023 review of the latest meta-analytic and longitudinal studies confirms this: the more rigorous the study design, the smaller the effects become, to the point of being "inconsequential" [4]. So when you hear dramatic claims about social media harming or helping teens, the actual data says the effects are barely detectable for most individuals.

Where do placebo effects and expectancy bias sneak in?

Placebo effects aren't just for pills—they also happen with control settings on social media. One study found that giving users a sense of control over algorithmic feeds (like toggling what they see) produced a placebo effect: people felt better or more in control even when the settings did nothing [5]. In intervention studies, if teens believe a new feature or program is supposed to help them, that belief alone can create a measurable improvement—muddying whether the intervention itself actually works.

Expectancy bias is even more pervasive. A 2022 study tracked actual social media use across multiple devices (phones, tablets, laptops) for two weeks and then asked participants to self-report their time. The result: most people overestimated their use, and the overestimation was worst on the most popular platforms and among those who used multiple platforms [3]. Females, who tend to report more social media use, were also less accurate [3]. If intervention studies rely on self-reported use—and most do—then any link between "use" and outcomes is built on inflated numbers, making the intervention look more effective than it really is.

What does this mean for studies that claim social media interventions work?

It means we should take many of those claims with a grain of salt. If an intervention is tested using self-reported social media time, the measured "dose" of social media is likely wrong—and the error is systematic (people overreport) [3]. That overreporting can create a false correlation: for example, if heavy users overreport more than light users, an intervention that reduces use might look more effective than it actually is. The same bias can make a useless intervention appear beneficial if participants expect it to help.

Even when studies use objective measures, the effects are tiny [2][4]. So the combination of placebo effects from control settings [5] and systematic overreporting [3] means that many published intervention effects are likely inflated. The real-world benefit for the average teen is probably close to zero.

About These Sources

This answer is built on 5 peer-reviewed studies — published from 2018 to 2023, 2 in Q1 journals, collectively cited 362 times — selected as the most relevant from 5 studies that passed quality screening, drawn from 63 papers retrieved from a database of over 500 million.

Sources used in this answer

Related research questions