2024 Doctoral Awardee: Kayley D. Estes

Media Coverage of Collective Traumas: Implications for Cognitions, Affect, and Behaviors

Kayley D. Estes, Department of Psychological Science, University of California, Irvine

This dissertation examines the link between media exposure and emotional, cognitive, and behavioral responses during and following three collective traumas using large, representative samples and longitudinal designs. Among a sample of Texas residents following Hurricane Harvey, Study 1 found that trust in media buffers against traumatic stress symptoms over 14 months for people with high amounts of early media exposure to the disaster. Study 2 demonstrated that during Hurricane Irma, pre-hurricane risk perceptions, measured in the days before landfall, mediated the relationship between media exposure and evacuation behaviors among a sample of Florida residents, underscoring the importance of accurate, non-sensationalized information for emergency decisions. Among a national U.S. sample, Study 3 explored how early media consumption patterns during the COVID-19 pandemic were associated with long-term risk perceptions and vaccine uptake over 3 years. Using large, probability-based samples and data collection over time, this research provides insights for policymakers, media outlets, and emergency management personnel, advocating for trustworthy, straightforward reporting to support public health and safety. This work also contributes to understanding the nuanced relationship between media consumption and psychological outcomes during collective traumatic events, contributing to the literature for updated communication strategies and interventions.

Project Summary 


Innovations in 21st century technology have transformed the way people gather information about collective traumas (e.g., natural disasters, public health crises), providing people with minute-by-minute updates and live blogs before, during, and after the disaster (Tanaka, 2021). These advancements allow individuals to retrieve news about collective traumas from multiple media sources (i.e., traditional media, online news platforms, social media) simultaneously, potentially increasing the amount of media consumed in a given timeframe as people often use multiple information sources to remain current on disasters (Simon et al., 2015). Studies show that while media is vital for information dissemination, consuming extensive coverage is frequently associated with negative psychological outcomes in a variety of contexts (Hopwood & Schutte, 2017). In fact, extended exposure to media coverage of disasters may magnify individuals’ anxiety and fear (Holman et al., 2014), which could then trigger attentional bias towards the stimuli, making it challenging for people to disengage (Fox et al., 2001).

Nonetheless, there are many collective trauma contexts in which it is unrealistic – even dangerous – to advise the public to disengage from media. For example, in the context of approaching hurricanes (Studies 1 and 2), access to media allows for a direct line of communication between the public and emergency management personnel. When this communication is current and trusted by the public, it can bolster mitigation behaviors, improve early warning system efficacy, and foster appropriate and timely evacuation behaviors (Monahan & Ettinger, 2018). Indeed, a small but rapidly growing body of literature posits that institutional trust, specifically trust in social and traditional media, may be psychologically beneficial during stressful events characterized by uncertainty, such as natural disasters (Lee, 2022). Additionally, the literature presents inconsistent findings regarding specific factors associated with evacuation behaviors (Thompson et al., 2017), although one of the few consistent factors are risk perceptions. Specifically, subjective perceptions of risk have been robustly correlated with evacuation behaviors (Stein et al., 2013) but are rarely studied as a mediator, nor are the predictors of risk perceptions typically explored. These same issues are often present when considering public health emergencies such as a pandemic (Study 3). Media allows for communication of health protective and mitigation behaviors, which apprises the public of their risks. Consuming this content is likely associated with subjective risk perceptions, which may be connected to behaviors such as vaccine uptake.

A key limitation of collective trauma research, however, is that established research on natural disasters and public health crises often rely on retrospective data collection and employ non-representative samples. Retrospective reports are susceptible to biases due to situational factors (Loftus et al., 1978) and the deterioration of recall accuracy over time (Schmolck et al., 2000). “In the moment” risk perceptions, especially, are likely to be vulnerable to recall bias when measured post-event. In order to use findings from past natural disasters and public health emergencies to prepare and adapt for future incidents, measuring subjective variables in real time using representative, probability-based samples is necessary to ensure the data collected are as unbiased as possible.


Study 1 explored the relationship between hours of media exposure and perceived trust in that media and traumatic stress responses in a probability-based, representative sample of Texans assessed in the early aftermath of Hurricane Harvey in 2017 and followed over time. Harvey was a Category 4 hurricane when it made landfall and stalled over land for four days, severely flooding many areas. An initial online survey using the GfK probability-based preexisting panel of Texas residents was fielded 2 weeks post-landfall (N=1137), with follow-up surveys at 6 weeks (N=1023) and 14 months (N=748) post-landfall. Measures included traumatic stress symptoms (acute stress at Wave 1 and posttraumatic stress at Waves 2 and 3), Harvey-related media exposure, trust in that media, loss and injury due to Harvey, and demographic information. Generalized Estimating Equations (GEE) were used to evaluate longitudinal relationships, with trust in hurricane-related media moderating the relationship between media exposure and traumatic stress symptoms over time. Poststratification weights were used for all analyses to enable population estimates. 

Study 2 examined risk perceptions as a mediator between theoretically derived variables and evacuation behaviors using real-time online pre-event data collection in the GfK probability-based, representative sample of N=1,478 Florida residents as Hurricane Irma approached the U.S. in 2017. Wave 1 was fielded two days prior to Irma’s landfall, measuring real-time hurricane-based risk perceptions and media exposure, in addition to prior hurricane exposures. Wave 2 was fielded one month later, when evacuation behaviors and reasons for not evacuating were assessed (when applicable). Since ~40% of the representative sample incorrectly reported their evacuation status, two identical GSEM mediation analyses with binary evacuation behaviors as the outcome and continuous risk perceptions as the mediator were conducted on subsets of the sample: one among those who were under evacuation orders based on residential address (n=371) and one among participants who self-reported being in an evacuation zone (n=706). 

Respondents for Study 3, begun at the start of the COVID-19 pandemic in March/April 2020, included a national U.S. sample drawn from the probability-based NORC AmeriSpeak Panel. Longitudinal patterns of risk perceptions regarding perceived susceptibility (i.e., the likelihood of falling ill with COVID-19) and perceived severity (i.e., likelihood of dying if ill with COVID-19) were assessed four additional times over the course of the pandemic and endemic periods (September/October 2020, November 2021, May/June 2022, August/September 2023). Results were analyzed using two Growth Mixture Models. Specifically, links between early exposure to COVID-19 media and vaccine uptake three years later were examined with risk perceptions as a mediator using a mediation GSEM model, with vaccine uptake as the outcome and classes of risk perceptions patterns as a categorical mediator. All other exogenous variables were measured at Wave 1, including the key variables of media exposure (average hours of exposure), trust in media, and media dimensions, which were developed by multiple correspondence analyses; information-channel dimensions included journalistic complexity, ideological focus, domestic focus, and non-news (Jones et al., 2023). Final sample size was N=4570; poststratification weights were used for all analyses to enable population estimates. 


Study 1. Among participants reporting high perceived trust in the Hurricane Harvey-related media they consumed before and shortly after the hurricane made landfall, the relationship between average daily hours of hurricane-related media exposure (measured two weeks post-landfall) and traumatic stress symptoms was weaker than for those who perceived low trust in hurricane-related media at both six weeks (ß=-0.35, p=.002, 95%CI[-0.58,-0.13]) and 14 months (ß=-0.45, p=.001, 95%CI[-0.70,-0.19]) post-landfall. Additional statistically significant predictors included Harvey-based loss and injury (ß=.30, p<.001, 95%CI[.22,.39]) at six-weeks post-landfall, and both Harvey-based loss and injury (ß=.31, p<.001, 95%CI[.21,.40]) and self-reported income (ß=-.13, p=.005, 95%CI[-.21,-.04]) at 14-months post-landfall. 

Study 2. Using population weights on the complete N=1,478 sample, 5.84% of participants were unaware of being under evacuation orders, while 33.85% of the sample incorrectly believed themselves to be under evacuation orders when they were not. A total of 17.29% of the sample evacuated unnecessarily. Due to the high percentage of participants incorrect about their evacuation zone status, the self-report subsample was used for interpretation for stronger ecological validity. 

All continuous variables were standardized before being included in the model. Results indicated that more Irma media exposure (β =.22, p=.01, 95%CI[.15,.29]), lower reported income (β=-.11, p=.005, 95%CI[-.19,-.03]), and prior loss or injury due to prior disaster (β=.26, p=.003, 95%CI[.09,.43]) were associated with higher Irma-related risk perceptions. Having higher risk perceptions (OR=1.46, p<.001, 95%CI[1.22,1.73]), being younger (OR=.79, p=.008, 95%CI[.67,.94]), and having prior experience in an evacuation zone regardless if one evacuated or not (OR=1.62, p=.004, 95%CI[1.17,2.24]) were linked with evacuating. Irma-based media exposure (OR=1.09, p=.001, 95%CI[1.03,1.14]), prior loss or injury due to a prior disaster (OR=1.10, p=.02, 95%CI[1.02,1.19]), and income (OR=.96, p=.02, 95%CI[.93,.99]) were indirectly associated with evacuation behaviors through risk perceptions.

Similar results were found for the subset of participants under evacuation orders based on residential address, despite being a smaller sample. 

Study 3. GMM results indicated four longitudinal risk perception trajectories for perceived susceptibility during the COVID-19 pandemic. A majority of the participants (74.50%) fell into the lowest risk perception category in which the average perceived risk was ~19% chance that one would fall ill with COVID-19 at the beginning of the pandemic and reduces to ~5% over time. Other classes of patterns started higher (between 25% and 35%) and decreased or increased only slightly. A small but important group (0.6% of the sample) began with an average perceived risk of 26% and increased to 91%. The patterns were more varied for perceived severity. Again, the majority (74.70%) began and ended with low perceived risk (<10% chance of dying), but 11% of the sample increased over time, while 7.40% of the sample rapidly decreased from 65% chance of dying at the beginning to <5% over time. GSEM results indicate that compared to people who received original and booster vaccines, people who did not receive boosters consumed more early media [RRR=1.03, p<.001, 95%CI[1.02,1.04] and had lower trust in COVID-19 information sources [RRR=.73, p<.001, 95%CI[.64,.83], mediated through risk perception classes. 


 Study 1 demonstrated that perceived trust in hurricane-based media may protect against the negative symptoms associated with early media exposure after a natural disaster. Longitudinally, I show that these results are consistent over time, as trust in hurricane-related media coverage protected against traumatic stress symptoms up to 14 months later among Texans who consumed high daily amounts of Hurricane Harvey-related news during and immediately after Hurricane Harvey’s landfall. Study 1 used both a longitudinal design and a representative sample of Texas residents, providing rigorous data to test this interaction over time. These results further validate the need for media platforms, companies, and government agencies to provide honest and accurate information during natural disasters without sensationalizing the information to promote trust in their reporting.

Study 2 showed that real-time risk perceptions mediate the connection between some predictors with previously mixed results (i.e., income, prior hurricane-based loss and injury) and evacuation behaviors. It also demonstrated that the amount of hurricane-based media exposure is linked with evacuation behaviors indirectly through risk perceptions and should be included in future evacuation behavior models. This study advances the evacuation behavior literature with more comprehensive modeling and novel, real-time data collection as the hurricane approached. Results can benefit emergency management personnel by using data that do not rely on recall bias.

Study 3 highlighted that while a majority of people had low risk perceptions regarding susceptibility and severity of COVID-19 at the start of pandemic, there was a subset of the population who increased in their perceptions of risk over time, even as health protective behaviors, lockdown measures, and vaccines and boosters were communicated and widely distributed. These cognitions appear to be linked with different information-channel characteristics of the media they consumed at the start of the pandemic, as well as vaccine uptake years later. This study advances theory and informs communication professionals by examining the link between early crisis media consumption and long-term risk perceptions and health protective behaviors.

Together, this dissertation demonstrates that many aspects of media (e.g., amount of exposure, subjective trust in one’s media, specific characteristics of the media consumed) are associated with affect (traumatic stress symptoms), cognitions (risk perceptions), and behaviors (evacuation behaviors and vaccine uptake). Its results have implications for the scientific domain, emergency management personnel, media outlets, and the public. Employing probability-based, representative samples in all three studies allowed for generalizing to larger populations of interest, which is essential for emergency personnel and media companies to make policies and mitigation plans that benefit the most people. Additionally, the use of real-time data collection in Studies 2 and 3 bypass the issue of recall bias, especially for subjective risk-perceptions that measure in-the-moment cognitions, improving the quality of data presented in the literature. 

Ultimately, this dissertation provides theoretically rich and methodological rigorous data to inform policymakers, media companies, and the public at large as they navigate access to and delivery of the multitude of collective traumas people experience and consume across traditional and new media in their daily lives.

Impact Statement 

This dissertation highlights the critical need for policies ensuring media integrity and trustworthiness during crises, underscoring the connection between several media factors (e.g., amount of exposure, perceived trust, ideological focus of the information source) and evacuation and health behaviors. It demonstrates that rigorous research with real-time data collection regarding a variety of aspects of media should be incorporated into emergency preparedness and public health communication strategies. The findings, drawn from probability-based, representative samples, offer essential insights for policymakers and media companies to enhance public awareness and promote informed decision-making during collective traumas. Ultimately, this research calls for a collaborative approach between public policymakers, media industry personnel, emergency management personnel, and scientists to foster a media environment that supports public well-being and effective crisis response. 


Fox, E., Russo, R., Bowles, R., & Dutton, K. (2001). Do threatening stimuli draw or hold visual attention in subclinical anxiety? Journal of Experimental Psychology. General, 130(4), 681–700. https://pubmed.ncbi.nlm.nih.gov/11757875

Holman, E. A., Garfin, D. R., & Silver, R. C. (2014). Media’s role in broadcasting acute stress following the Boston Marathon bombings. Proceedings of the National Academy of Sciences, 111(1), 93–98. https://doi.org/10.1073/pnas.1316265110

Hopwood, T. L., & Schutte, N. S. (2017). Psychological outcomes in reaction to media exposure to disasters and large-scale violence: A meta-analysis. Psychology of Violence, 7(2), 316–327. https://doi.org/10.1037/vio0000056

Jones, N. M., Thompson, R. R., Holman, E. A., & Silver, R. C. (2023). Idiosyncratic media exposures during a pandemic and their link to well-being, cognition, and behavior over time. Proceedings of the National Academy of Sciences, 120(26), e2304550120.https://doi.org/10.1073/pnas.2304550120

Lee, S. (2022). Subjective well-being and mental health during the pandemic outbreak: Exploring the role of institutional trust. Research on Aging, 44(1), 10–21. https://doi.org/10.1177/0164027520975145

Loftus, E. F., Miller, D. G., & Burns, H. J. (1978). Semantic integration of verbal information into a visual memory. Journal of experimental psychology: Human learning and memory, 4(1), 19.

Monahan, B., & Ettinger, M. (2018). News media and disasters: Navigating old challenges and new opportunities in the digital age. In H. Rodríguez, W. Donner, & J. E. Trainor (Eds.), Handbook of Disaster Research (pp. 479–495). Springer International Publishing. https://doi.org/10.1007/978-3-319-63254-4_23

Schmolck, H., Buffalo, E. A., & Squire, L. R. (2000). Memory distortions develop over time: Recollections of the OJ Simpson trial verdict after 15 and 32 months. Psychological Science, 11(1), 39-45.

Simon, T., Goldberg, A., & Adini, B. (2015). Socializing in emergencies—A review of the use of social media in emergency situations. International Journal of Information Management, 35(5), 609–619. https://doi.org/10.1016/j.ijinfomgt.2015.07.001

Stein R., Buzcu‐Guven, B., Dueñas‐Osorio, L., Subramanian, D., & Kahle, D. (2013). How risk perceptions influence evacuations from hurricanes and compliance with government directives. Policy Studies Journal, 41(2), 319-342. https://doi.org/10.1111/psj.12019

Tanaka, Y. (2021). Social media technologies and disaster management. In M. Sakurai & R. Shaw (Eds.), Emerging Technologies for Disaster Resilience (pp. 127–143). Springer Singapore. https://doi.org/10.1007/978-981-16-0360-0_8Thompson, R. R., Garfin, D. R., & Silver, R. C. (2017). Evacuation from natural disasters: A systematic review of the literature: Evacuation from natural disasters. Risk Analysis, 37(4), 812–839. https://doi.org/10.1111/risa.12654