Brief 31: How to Handle Rumors

This study explores political rumors surrounding health care reforms in the US. Refuting rumors with statements from unlikely sources can, under certain circumstances, increase the willingness of citizens to reject rumors regardless of their own political predilections. The effect of debunking rumors falls away quickly, while the effect of reinforcing rumors persists.

Link to Full Study

Category: Public Service Provision

Tags: healthcare, reform, misinformation, ACA, Obamacare, partisan

Date of Publication: Tuesday, September 1, 2015

EGAP Researcher: Adam Berinsky

Geographical Region: North America

Research Question:

How do people respond to rumors? What are effective strategies to discredit rumors and correct misinformation?

Preparer: Alicia Cooperman

English

Background:

Rumors are widespread in politics. The internet and news media have only increased the ability to spread false information. For instance, President Obama’s Affordable Care Act (ACA) proposal in 2009 was the subject of rampant rumors about “death panels,” where politicians who were opposed to the bill suggested that sick patients would have to stand before a government panel to get access to health care. Polls done by the Pew Center and YouGov in 2009 and 2010 found that around 30% believed the death panel rumor to be true and around 20% were unsure if it was true.

A common strategy to counter rumors is to confront the “myths” and then debunk them with non-partisan, scientific facts. Unfortunately, this often backfires by reinforcing the rumors and increasing people’s false belief in the long-run. Other research suggests that presenting new information from partisan sources may be effective at correcting misinformation, particularly when the rebuttal comes from politicians who go against their own interests.


Research Design:

To test different strategies of correcting misinformation about the “death panel” rumor regarding the ACA, Berinsky conducted two online survey experiments. The first compared subjects’ beliefs about the death panel rumor after receiving no information, information about the rumor, or information plus corrections. The second experiment studied the effect of reinforcing the rumor.

In the first experiment in May 2010, 1,701 respondents in a national sample were randomly assigned to treatment groups and presented with information from different sources. They were asked if they believed the rumor, their opinion of health care policy, and questions to measure their general attentiveness to the survey. The 699 “attentive” respondents were surveyed again a week later. The four different treatments varied as follows:

  • Rumor-only: a presentation of quotes that warned about the death panels,
  • Rumor + Non-partisan correction (NPC): adds a rebuttal with facts from the American Medical Association (AMA) and American Association of Retired Persons (AARP)
  • Rumor + Republican correction: NPC + a quote from Republican Senator John Isakson that debunks the rumor
  • Rumor + Democratic correction: NPC + a quote from Democratic Representative Earl Blumenauer that debunks the rumor

In the second experiment in October/November 2010, 1,000 respondents in a YouGov study were randomly assigned to three treatment groups (with no control). They were asked if they believed the rumor and were surveyed again a month later. Those who received the Rumor treatments were also randomly assigned to the “long recall” condition and asked two follow-up questions in which the rumor text was presented and the respondent was asked to identify the source, or assigned to the “irrelevant recall” condition and asked one multiple-choice question about the office of a politician quoted in the story.

The treatments were:

  • Rumor-only (same as Study 1)
  • Correction-only: no rumor information, but provides actual information from the ACA.
  • Rumor + Republican correction (same as Study 1)
Results:

The interventions confirm the worry about reinforcing rumors while trying to debunk them, but they also highlight the potential benefits of presenting corrections from ‘unlikely’ politicians whose personal and political interests would otherwise align with the rumor. Still, most of the benefit from ‘unlikely debunking’ falls away after even a week, while the effect of reinforcing the rumor persists.

In the first study, the treatment of the rumor with a correction from a Republican source was most effective in increasing the rejection rate and reducing the “not sure” rate. However, this difference across treatment groups generally disappeared after only a week. The Republican correction continued to be the most effective, but the rumor rejection rate dropped by 10 percentage points and its difference was no longer statistically significant.

While the Republican correction was beneficial for getting people to reject the death panel rumor, it only made a slight difference for increasing support for the ACA health policy itself. In fact, the Democratic correction even backfired, leading to lower support for the ACA.

In the second study, the treatments of information to correct the rumor (54% reject) and the correction in addition to hearing the rumor (60% reject) were more successful than the treatment without any correction (48% reject). Unfortunately, as in the first study, the benefit of the correction with the rumor disappeared after a month.

Asking questions to reinforce the rumor (long recall condition) did make a difference in the rejection and persistence of the rumor over time. In the first wave, the long recall group was more likely to accept the rumor and less likely to reject it. When the rumor was paired with information to correct it, the long recall group was still less likely to reject the rumor (compared to the “irrelevant recall” condition), though the correction did increase rejection rates overall. This stayed true even a month later, though the benefit of the correction fell over time. Broadly, this suggests that repeating the rumor does make it more believable and persistent.

Selected Survey Outcomes:

Study 1: Effect of Treatments on Rumor Belief, May 2010 – Attentive Sample, Wave 1

  Control Rumor Only Rumor + Non-partisan Correction Rumor + Republican Correction Rumor + Democratic Correction
Reject Rumor 57% 46% 60% 69% 60%
Accept Rumor 12 17 14 15 17
Not Sure 31 36 26 16 24
Total 100% 100% 100% 100% 100%

Attentive Sample: N=876 Statistically significant pairwise differences at 95% confidence: Control vs. Rumor + Republican; Rumor Only  vs. Rumor + Republican; Rumor Only vs. Rumor + Democratic

Study 1: Effect of Treatments on Health Care Policy, May 2010 – Attentive Sample, Wave 1

  Control Rumor Only Rumor + Non-partisan Correction Rumor + Republican Correction Rumor + Democratic Correction
Support ACA 51% 42% 46% 48% 37%
Oppose ACA 49 58 54 52 63
Total 100% 100% 100% 100% 100%

Attentive Sample: N=876 Statistically significant pairwise differences at 95% confidence: Rumor Only  vs. Rumor + Republican; Rumor + Republican vs. Rumor + Democratic

Study 2: Effect of Rumor Rehearsal on Euthanasia Rumor Belief, October 2010

  Rumor Only Rumor + Correction
  Irrelevant Recall Long Recall Irrelevant Recall Long Recall
Reject Rumor 54% 42% 62% 58%
Accept Rumor 26 32 18 21
Not Sure 20 26 19 21
Total 100% 100% 100% 100%
Rumor Only: N=350; Rumor + Correction: N=342

 

Policy Implications:

Overall, the studies demonstrate the danger of presenting the rumor and then debunking it. Not only is this strategy generally ineffective, especially in the long run, but it can actually backfire and serve to reinforce the rumor. Asking respondents to directly recall the rumor made them less likely to reject it, even when presented with correct information.

Politicians and organizations must be cautious when trying to correct rumors and misinformation, which run rampant in the news and social media. While many advocacy groups may be inclined to shy away from partisan rhetoric, Berinsky finds that non-partisan corrections do not work and may only reinforce the rumor. Even more worrisome, an attempt to correct the rumor with information from a Democratic source actually led to lower support for the ACA health care policy.

The most effective strategy, though only in the short run, is to present the rumor with a partisan correction: information from a political source who is seen as being unlikely to take that point of view for personal or political reasons. In this case, a quote from a Republican senator against the “death panels” rumor worked, since Republicans were seen to benefit from the backlash against the ACA and were the main source of death panel rumors.