In the last decade, disinformation has become a matter of widespread concern. But there is one key question that persists: What is its impact on society? Does it have direct consequences in people’s beliefs or behavior?

The answer isn’t simple. Understanding the causality between the information we consume and our decisions is incredibly challenging – it’s impossible to isolate the effect of disinformation from all other social and cultural factors that shape our beliefs and behaviors. 

However, this doesn’t mean we can’t study the phenomenon. The key lies in understanding the methodological limitations and effects that we are able to document.

Magnitude of the Phenomenon and Measurement Limitations

---

Some researchers question the magnitude of the problem, arguing that exposure to fake content is, on average, limited. However, this claim is based on studies with important methodological constraints.

As we explained in detail in “How much disinformation is being spread?,” these studies usually only measure the exposure to social media posts that include links to websites that have already been classified as disinformation. This leaves out online texts, images and videos that don’t include links to external websites; thousands of items circulating on WhatsApp or Telegram that are difficult to track; claims by political leaders, spokespeople or influencers that are widely trusted and have a massive reach; and all disinformation being currently disseminated in traditional media outlets, like radio, television and newspapers, or in daily conversations and community spaces.

In addition, other studies tend to define disinformation only as false claims, when the seriousness of the problem is also due to misleading arguments with deliberate omissions, half-truths or information that is technically correct, but taken out of context. In fact, some research shows that biased or misinterpreted information has a greater negative impact than information that’s purely false; for instance, on vaccination intentions.

Last, only measuring individual access to certain sites or content ignores the fact that the Internet is a participatory environment where people don’t consume information in a passive and isolated way: they interact, argue, validate and reinforce shared narratives within communities, both online and offline. It is in this social process – rather than in the mere exposure to specific content – where disinformation unfolds most of its impact.

Altogether, these methodological limitations not only underestimate how much disinformation is out there, but also fail to capture how its impact really works: accumulatively, in social interactions and through multiple channels.

Impact on Behavior

---

For socially complex phenomena like forming political opinions, demanding evidence of direct and isolated causality might be inappropriate when assessing relevance. A factor doesn’t need to be the sole cause for it to be important.

As we explained in depth in “What we know (and don’t know) about the real impact of disinformation,” there is consistent evidence linking exposure to disinformation with certain beliefs and behaviors.

In public health, numerous studies find correlations between believing in specific disinformation (like vaccines causing autism) and vaccine hesitancy. During the COVID-19 pandemic, adhering to false or misleading narratives was linked with a lower level of compliance with health measures and a greater mistrust towards health authorities.

In the political spectrum, disinformation rarely changes one’s vote, but it can have other effects that are relevant for democracy, such as electoral demobilization, confusion regarding voting procedures and erosion of trust in the legitimacy of democratic processes. Even relatively small impacts might have significant results when they affect the willingness to participate or undermine institutional trust.

In summary, the fact that the effects of disinformation are hard to isolate or quantify does not make it irrelevant. Acknowledging this complexity is key to design effective responses that protect both information integrity and democratic health.