Last month, the House Intelligence Committee released a previously classified section of the 2002 congressional inquiry into the Sept. 11, 2001, attacks known as the missing “28 pages.” The U.S. government allegedly kept this document confidential to protect its relationship with Saudi Arabia. Instead, the move ended up fueling a massive controversy centered on claims of a coverup by the U.S. government intended to suppress evidence of Saudi complicity in the 9/11 attacks.

Our research helps explain why withholding information from government documents like the “28 pages” is so often counterproductive. In a new article (written with co-authors at Dartmouth College) in the Journal of Experimental Political Science, we show how the use of redactions can undermine the effectiveness of documents intended to reduce belief in conspiracy theories. Keeping information from the public can create the perception of a cover-up even when none exists.

In this case, the “28 pages” describe possible Saudi government ties to the 9/11 hijackers and al-Qaeda that have not been corroborated in subsequent investigations. When the document was released, the Obama administration reiterated the 9/11 Commission’s conclusion that it was unable to find evidence that the “Saudi government as an institution or senior Saudi officials individually funded” al-Qaeda.

To demonstrate how withholding information can fuel conspiracy belief in cases like this, we examined a different conspiracy theory that the government has sought to debunk — the unsupported claim that the 1996 explosion of TWA Flight 800 resulted from an accidental Navy missile strike rather than the ignition of a flammable fuel/air mixture.

In our studies, we showed a convenience sample of online participants in the United States a brief article about the controversy over the cause of the crash. Respondents were then assigned to see three documents that supported the official account of the crash, or to a control condition.

Participants in one condition simply saw documents with text, while others were randomly assigned to instead see a version of the documents that appeared to have been redacted. Although no text was actually obscured — people in both conditions were shown the same information — the second group saw black lines over parts of the documents. After reviewing the documents, respondents were then asked about their belief in both the official explanation and the conspiracy theory surrounding the TWA Flight 800 crash.

Across two separate studies, what we found was striking. Compared with the control group that did not see official documents, participants in the unredacted treatment were significantly more likely to believe the official explanation and less likely to agree with the conspiracy theory about the crash. On the other hand, there was a limited or null effect on individuals who were shown what appeared to be redacted versions of the documents. The presence of redactions appears to undermine the effectiveness of corrective information.

The magnitude of these effects was substantial. For instance, we found in our first study that 26 percent of control-group respondents could be described as conspiracy adherents. On average, these respondents did not believe the official explanation and endorsed claims that the government was involved in the crash, that the flight was shot down by a missile and that the government was covering up the true cause of the crash.

Exposure to the unredacted documents significantly reduced the prevalence of conspiracy adherents, which declined to 18 percent. By contrast, 24 percent of respondents exposed to otherwise identical redacted documents were conspiracy adherents — almost one-third more than when the documents were not redacted.

These findings help us understand why the “28 pages” undermined the conclusions of the 2002 congressional inquiry and 9/11 Commission report and took on “near-mythic status.” When the government withholds information, people become suspicious that it has something to hide.

Unfortunately, our study suggests that the release of the 28 pages may itself be ineffective. Although many in Washington hoped that the document’s long-awaited release would put an end to the controversy, the version of the document that was declassified still contains numerous redactions. As a result, it may not only fail to satiate avid conspiracy theorists but could even encourage them.

Sasha Dudding and Heather Szilagyi are 2015 graduates of Dartmouth College, where Brendan Nyhan is a professor of government. Their study in the Journal of Experimental Political Science was co-authored with Franklin Dickinson III, Enxhi Dylgjeri, Eric Neiley, Christopher Pullerits, Minae Seog, Andy Simpson, and Colin Walmsley, who were all undergraduate students at Dartmouth.