Social media and suicide: empirical evidence from the quasi-exogenous geographical adoption of Twitter

Alexis Du,Thomas Renault

2024-12-04

Abstract:Social media usage is often cited as a potential driver behind the rising suicide rates. However, distinguishing the causal effect - whether social media increases the risk of suicide - from reverse causality, where individuals already at higher risk of suicide are more likely to use social media, remains a significant challenge. In this paper, we use an instrumental variable approach to study the quasi-exogenous geographical adoption of Twitter and its causal relationship with suicide rates. Our analysis first demonstrates that Twitter's geographical adoption was driven by the presence of certain users at the 2007 SXSW festival, which led to long-term disparities in adoption rates across counties in the United States. Then, using a two-stage least squares (2SLS) regression and controlling for a wide range of geographic, socioeconomic and demographic factors, we find no significant relationship between Twitter adoption and suicide rates.

General Economics

What problem does this paper attempt to address?

The core problem that this paper attempts to solve is to explore whether the use of social media (especially Twitter) has a causal impact on the suicide rate. Specifically, the author aims to distinguish the causal relationship between social media use and the suicide rate, rather than simply observing the correlation between the two.

### Research Background
In recent years, with the rapid development of the Internet and social media, the suicide rate in the United States has also increased. Some studies have shown that people who use social media frequently are more likely to have problems such as anxiety, depression, and sleep disorders. However, most of these studies are based on cross - sectional data or surveys, and it is difficult to determine the causal relationship. In particular, there is the problem of "reverse causality": that is, those who are already at a higher risk of suicide may be more inclined to use social media, which makes it difficult to determine whether social media causes mental health problems or mental health problems prompt people to use social media more.

### Research Methods
To solve this causal identification problem, the author adopts the instrumental variable (IV) method, using the 2007 SXSW (South by Southwest) music festival as an exogenous shock to analyze the adoption of Twitter in different regions. The specific steps are as follows:

1. **First - stage regression**:
- Use the number of new users in March 2007 of SXSW as an instrumental variable to predict the number of Twitter users in each county in 2015.
- The model formula is:
\[
\text{TwitterUser}_{i,2015}=\alpha + \beta_1\text{SXSW}_{i,\text{March 2007}}+\beta_2\text{SXSW}_{i,\text{Pre2007}}+X_i + u_i
\]
- Where \(\text{TwitterUser}_{i,2015}\) represents the logarithm of the number of Twitter users in each county in 2015, \(\text{SXSW}_{i,\text{March 2007}}\) is the number of users who joined Twitter in March 2007 and followed SXSW, \(\text{SXSW}_{i,\text{Pre2007}}\) is the number of users who joined Twitter before 2007 and followed SXSW, and \(X_i\) includes control variables such as population density and age distribution.

2. **Second - stage regression**:
- Use the number of Twitter users predicted in the first stage to analyze its impact on the change in the suicide rate in each county.
- The model formula is:
\[
\% \Delta \text{SuicideRate}_i=\alpha + \beta_1\hat{\text{TwitterUser}}_{i,2015}+\beta_2\text{SXSW}_{i,\text{Pre2007}}+X_i + u_i
\]
- Where \(\% \Delta \text{SuicideRate}_i\) represents the percentage change in the suicide rate per 100,000 people per year, \(\hat{\text{TwitterUser}}_{i,2015}\) is the number of Twitter users predicted from the first - stage regression, and \(X_i\) includes geographical, demographic, and socioeconomic control variables.

### Main Findings
- After controlling for a wide range of geographical, demographic, and socioeconomic factors, the study found that there is no significant relationship between the adoption of Twitter and the suicide rate in each county in the United States.
- This result is consistent in different robustness tests, including different methods of filling in missing values.

### Conclusion
This study provides the first empirical evidence that the expansion of Twitter has no significant impact on the suicide rate in each county in the United States. This conclusion emphasizes the need to interpret the impact of social media on mental health carefully and points out that different platforms may have different effects. Future research should further explore other social media platforms and their impact on specific populations.