There Are Two Types of Misinformation

Misinformation of all types continues to affect what you believe.

Posted Aug 11, 2019

PlusLexia is licensed under CC BY 2.0
Source: PlusLexia is licensed under CC BY 2.0

Over the last decade, people have become sensitive to “fake news.” There is real concern that people will hear false information via the internet that they initially believe to be true. The presence of a lot of misinformation is important because of the continued influence effect, which is the observation that the first pieces of information people hear continue to affect what people believe, even when they later find out that information is false.

Most studies of the continued influence effect have focused on situations in which people hear that a particular fact is true. For example, many people have been influenced by reports that vaccines are linked to autism and have chosen not to vaccinate their children. As it turns out, the scientific study that provided evidence for the link between vaccination and autism was shown to be fraudulent. Yet, many people continue to believe that there is a link.

A paper by Andrew Gordon, Ullrich Ecker, and Stephan Lewandowsky in a 2019 issue of the Journal of Memory and Language looked at the opposite case. What happens when the first information you hear is the negation of a fact. You might hear that human activity is not causing climate change. Later, you receive a lot of evidence supporting the assertion that human behavior is in fact causing climate change. Does the continued influence effect work the same way for facts that are initially negated?

 To explore this question, the researchers had participants read a fictitious account of a news story about a crime. The key facts in this story were whether the alleged assailant was Muslim and whether the crime was religiously motivated. 

Participants from the United States read one of five versions of the stories. Then, they read the story a second time. This second version either did or did not change an element from the first version. There was a control condition in which the religion of the assailant was never mentioned.

Two of the versions were intended to provide a baseline. In one version, the first version asserted that the assailant was not Muslim, and there was no change in the second story. This condition provides the baseline for a negated piece of information. A second version asserted that the assailing was Muslim and there was no change in the second story. This condition provides the baseline for an asserted piece of information.

The other two conditions change information from the first version to the second. In one condition, the assailant is first described as Muslim, and the second version corrects that to say that the assailant was not Muslim. The second condition starts with a story stating that the assailant was not Muslim and the second corrects the story to say that the assailant is Muslim.

After reading both versions of the stories, participants answered several questions about the stories including the religion of the assailant.

Because people’s attitudes toward Muslims could influence their likelihood of believing that the assailant was Muslim, all participants took a brief survey during the study assessing their attitudes toward Muslims. Participants overall attitudes toward Muslims was factored into the analysis.

Overall, participants with a more negative attitude toward Muslims were more likely than those with a more positive attitude to answer that the assailant was Muslim—even in the control condition where religion was never mentioned. However, this attitude did not interact with the main manipulation of the study.

As you might expect, when participants heard that the assailant was Muslim and that information was not corrected they generally believed the assailant was Muslim. Similarly, when the participants heard the assailant was not Muslim and that information was not corrected, they generally believed that the assailant was not Muslim.

The two misinformation conditions acted similarly. That is, whether participants first heard that the assailant was Muslim and that information was later corrected or they heard that the assailant was not Muslim and that information was later corrected, their responses came out in between the two baselines. That is, there was a tendency for people to continue to believe the first information they heard, even when it was corrected. A second study in this paper replicated the effect with a somewhat different story in an Australian population.

These studies extend previous research on the continued influence effect by demonstrating that it involves either facts asserted to be true that are later found to be false as well as facts asserted to be false that are later found to be true. This work demonstrates why the first information you hear on a topic has such a profound influence on what you believe. It also suggests that the continued influence effect has some impact on people’s beliefs, even when they are predisposed to disagree with the first facts they hear.


Gordon, A., Ecker, U.K.H., & Lewandowsky, S. (2019). Polarity and attitude effects in the continued influence paradigm. Journal of Memory and Language, 108, 104028