While novel antitrust efforts targeting big tech continue to fail, proposals aimed at limiting access to social media are gaining traction. The recent release of The Anxious Generation, by famed social psychologist and author Jonathan Haidt, appears to be another arrow in the quiver for those hoping to place age verification limits on social media. In the book, Haidt likens such limits to those of known harms including smoking cigarettes. However, this comparison drastically overestimates the certainty as to whether social media causes mental health issues. Lawmakers should only regulate based on facts, not popular rhetoric. 

Graphs from Haidt’s book showing the uptick in social media use that coincides with the uptick in mental health may appear compelling at first glance. Haidt asserts that such relationships are causal and introduces examples of random control trials as evidence. However, when promoting his new book Haidt stated that the correlation coefficient, which measures the strength of the relationship between two variables, for mental health and social media is around 0.17 with the exact number varying depending on which gender is the primary focus. However, across all disciplines coefficients below 0.2 were considered weak or poor.

The weakness of this relationship is put on display when directly compared to smoking. The relationship between lung cancer and smoking varies a bit by country, gender, and cancer lag times, but research finds correlation coefficients reaching 0.95, which is a very strong positive correlation.

However, even strong correlations alone are not enough to establish causation, as the publication of Spurious Correlations demonstrates. This site publishes the relationships between such different variables in which a lack of causality is obvious. Examples include the number of movies Russell Crowe appeared in and customer satisfaction with Walgreens. Despite clearly being a coincidence, the correlation coefficient for these two variables is over 0.6.

The reality is that determining the causes of mental health is not straightforward. The link between smoking and cancer is statistically shown by a correlation coefficient, but also by medical evidence on how smoking causes cancer. A simplified explanation is that the chemicals in cigarettes damage DNA; this makes it harder for bodies to repair cells, and such DNA damage can cause cancer.

Explanations offered by proponents of age restrictions such as Haidt, are difficult to verify. Spending hours scrolling through photoshopped images and seeing friends engage in activities without you is one explanation for how social media could cause mental health issues. But Haidt takes it one step further by arguing that the medium is the message. This argument asserts that no matter the content, the act of scrolling and engaging with friends digitally is what drives the trend in mental health.

If the relationship between mental health and social media was related to content, the fix would involve existing options for blocking or limiting content, which could be expanded to help curate a healthy user experience. These actions could take place at either the parental or child level and would allow for a less heavy-handed approach. Another advantage to this approach would be that it is tailorable to specific children, as what is triggering or unhealthy for one child may not have the same effect on a different child.

However, asserting that the cause is the structure of the platform, excludes this less invasive intervention. The assertion also makes demonstrating causality more complicated as the explanation must move beyond problematic content and explain how viewing content, regardless of type, on a screen causes depression and anxiety. That explanation of the causality is a necessary component to the scientific method for hypothesis testing and is something that is missing from this conversation.

Regulation aimed at limiting access should ideally have high levels of correlation and an explanation. Lawmakers can explain the causal relationships between smoking and drinking, social media shouldn’t be an exception.

Smoking, drinking, and adult content are some of the examples used to justify age restrictions on social media. However, most of these relate to and are justified by known and measurable health risks. The exception of adult content still has a stronger correlation coefficient to mental health issues than what currently exists for social media. Additionally, it is difficult to argue that adult content is comparable to non-graphic content or the structure of the platforms. 

Protecting youth online is important and more work should be done to examine recent trends in depression and anxiety. However, parents are overwhelmingly in the best position to make decisions for their children.

Even without determining whether the correlation is statistically significant, having a coefficient of 0.17 means that most of the variance in mental health is explained by factors outside of social media.

When claiming a causality between social media and mental health, policy makers and researchers should be able to support their claims with high levels of confidence. This is missing from the current conversation. Children’s mental health is an important issue, but this importance requires that we get the facts right. Lawmakers should pause before passing legislation potentially based on faulty understanding.

Tirzah Duren is the Vice President of Policy for the American Consumer Institute, a nonprofit educational organization. For more information about the Institute, visitwww.TheAmericanConsumer.org or follow us on Twitter (X) @ConsumerPal.