This study examined knowledge of, perceived risk of, attitudes toward, and behavioral intentions related to Ebola soon after the media coverage of the Ebola epidemic in West Africa and the few cases in the United States. News coverage prior to the survey provides context that allows us to better interpret the survey findings. Towers et al.  found that news videos on two major news networks between mid-September and late October 2014 were highly effective at inciting public concern. Each Ebola-related news video inspired over 10,000 Internet searches and tweets. Basch et al.  found that only 4 % of news articles in the three most widely-circulated U.S. daily newspapers between September 17, 2014 and October 17, 2014 included content on precautions the public could take. Although information about the nature of the news coverage prior to our survey provides some indication of the type of information to which the public was exposed, because we used a cross-sectional study design, we cannot conclude that the findings from this survey are a direct result of this media coverage.
Only 19 % of survey respondents believed Ebola would spread to the U.S. This is significantly less than reported in surveys conducted earlier in the fall . If social amplification of risk occurred through heavy media volume, the reduction in media coverage that followed the first week of November  could possibly have contributed to this decline in perceived risk. One report suggests the amount of coverage decreased from 1000 segments in the 4 weeks prior to the mid-term elections to only 50 in the 2 weeks after . The reduction in risk might also be explained by the fact that there were no longer any Ebola cases in U.S. hospitals and transmission to others within the states had been very limited.
With regard to relative risk, participants rated Ebola as less of a threat than many of the other issues, with heart disease, seasonal flu, ISIS, and superstorms rating highest and Ebola rating more similarly to pandemic flu, West Nile Virus, and EV-D68 (though differences between these issues and Ebola were still statistically significant). T-tests for difference in means on perceived threat show differences between all issues and Ebola are statistically significant at p < .001 (except for West Nile Virus, different at p < .01). Twenty-seven percent reported having no knowledge of EV-D68 (compared to only 2 % who had not heard of Ebola). Of those who had heard of EV-D68, the level of self-reported knowledge was significantly lower than for Ebola. This gap is striking, because in 2014 EV-D68 affected significantly more people in the U.S. (1121 people and 12 confirmed deaths) , than Ebloa (seven cases and two deaths). Without minimizing the crisis in West Africa, this knowledge gap suggests that media coverage of these health threats may have been unbalanced.
Despite seemingly reasonable notions of perceived risk of Ebola, slightly more than half of those surveyed said they planned to take some form of action to avoid contracting the illness, including either avoiding public transportation, avoiding those who have traveled to affected areas, or changing hygiene practices. This contrast between perceived susceptibility and behavioral intentions contradicts traditional notions of health behavior theory, which suggest people are most likely to act when perceived susceptibility is high [21–23]. The extreme and graphic nature of an hemorrhagic illness like Ebola may induce fear that motivates the behavior. Literature has shown that fear can be a very compelling motivator of health behavior [23, 24]. Although these behaviors may not have been recommended, they could have positive consequences in some cases. For example, changing hygiene practices can include hand washing, which can also reduce transmission of other communicable diseases.
Only 22 % of respondents said that if someone in their community were to contract the illness they would be likely to die. This finding is not in line with known Ebola case fatality rates (about 50 % in the 2014 outbreak) , but may have been influenced by the high survival rate for healthcare workers treated in the U.S.
Despite the frequent news coverage , the findings regarding knowledge make it clear that many Americans still do not have a clear understanding of how to avoid contracting Ebola. The low correlation between confidence in ability to understand Ebola and actual knowledge is not surprising; this is consistent with other literature [26, 27]. These findings suggest public health officials and the media need to ensure that communication about an outbreak also educates the public about basic issues like transmission and prevention, which should be communicated clearly, in non-technical terms, and repeated often [28, 29]. Public health officials can also use formative research to help identify which channels are most likely to reach the target audience in large numbers.
Responses to the questions about government policies favored conservative choices. Most respondents were in favor of mandatory quarantine for those who have been exposed and in favor of travel bans, which are not currently recommended. Participants were largely split on whether infected American healthcare workers should be brought to the U.S. for care. Demographic differences varied depending on the policy. For example, the same groups who favored quarantine were not most in favor of travel bans. It is unsurprising that those who are older or who have children in the home were more likely to favor travel bans. Older adults may be more conservative in general. Older adults and children could also be perceived as more susceptible to disease in general. Blacks were less in favor of travel bans from affected regions. Some research has found racial bias in ability to empathize with others’ pain . Those in the Midwest and West were less in favor of travel bans than those in the Northeast. Those in the Northeast may have felt more vulnerable to infection because most points of entry from West Africa are located there. Patterns for those favoring quarantine are less clear. These differences should be explored in qualitative research.
Confidence in media, government, and local healthcare was low. This finding may be related to the fact that two nurses who had cared for a patient with Ebola at a Texas hospital contracted the disease just a few months prior to the survey. Because trust is one of the cornerstones of outbreak communication , and the media is the primary communication outlet for updating the public, public health officials and the media must develop concrete communication strategies to build Americans’ trust and confidence in order to be credible and prepared for future outbreaks. The Centers for Disease Control and Prevention have developed numerous materials on Crisis and Emergency Risk Communication (CERC), which include specific guidelines about communication during times of outbreak, such as acknowledging uncertainty, expressing empathy, and avoiding jargon. These can be valuable resources for building that trust . Other strategies include being aware of the preexisting low levels of trust among the American public, being honest and open, providing enough information to make informed personal decisions, and not using euphemisms because they imply a lack of honesty [28, 29].
SteelFisher, Blendon, and Lasala-Blanco  recommend that for future outbreaks, public health officials should establish relationships with independent health professional associations that are trusted by the public and work with them to shape policies and messages with the public . Finally, the media in particular must make sure that news stories provide an accurate portrayal of the risk of an infection or disease and not use gratuitous video footage just to secure high ratings.
Some demographic groups did have more confidence in these institutions than others. Some of these differences were not intuitive. It is not obvious why men might have less confidence than women in the media’s ability to accurately report on an Ebola outbreak, though this is consistent with previous research showing women tend to rate the media as more credible . Differences in confidence in hospitals were somewhat nuanced. The groups who had more confidence in a local hospital’s ability to treat an infected patient did not necessarily have more confidence in the hospital’s ability to prevent healthcare workers from catching Ebola and vice versa. Perhaps tailored communications could help to reach the populations with lowest levels of trust.
Future research should also explore the implications of intentions to avoid those who have traveled to West Africa or foods and beverages produced there. Previous research has found significant social stigma associated with Ebola . These findings could have serious implications for African people living in or traveling to the United States.
As is typical for online surveys using nationally representative panels, the response rate for the survey is less than ideal. However, for comparison, other surveys using GfK’s KnowledgePanel®, such as those conducted by Pew Charitable Trust and the Federal Reserve, have also reported response rates in the single digits [34, 35]. As was mentioned previously, response rates for panel surveys do tend to be lower than for other modes, due to the multiplication of retention, cooperation, profile and completion rates . In addition, weighting helps to alleviate some concerns regarding the sample.
The data are self-reported, and there is likely some social desirability bias, which was not measured. Analysis of travel volume data, point of purchase data (for self-protective gear), and information from other more objective sources can help to shed light on the true impact of the epidemic on specific behaviors.
It is important to point out the variation that exists in the content of media information sources. Effects may have been different for audiences of different types of programming or different channels. It is possible some of the effects of age are due to the fact that younger people get more news online while older people watch more television news . Future research should explore these differences further.
It is possible it was difficult for people to separate the seriousness of Ebola in West Africa from that in the U.S. when answering questions about whether the media had exaggerated its seriousness. Because these were not asked separately for the two regions, we cannot tease that out in this study.