Disaster Risk Communication and Digital Vulnerability Among Subsidized Housing Residents

Yan Wang
University of Florida

Haiyan Hao
University of Florida

Seungbeom Kang
Yonsei University

Publication Date: 2023


Emergency officials often use digital media—including webpages, social media posts, and online news—to convey information about disaster risk. However, some groups may be unable to receive digital risk communication due to internet access barriers or lower levels of digital literacy skills. As a result, these groups may lack the information they need to prepare for and respond to disasters. We surveyed 200 subsidized housing residents living in flood-prone areas to investigate the relationship between their internet and social media skills and their acquisition of disaster information and mitigation behaviors. Most survey respondents had low incomes and other socio-economic characteristics that make them vulnerable to adverse disaster outcomes. Survey results underscored the mediating role of social media in disseminating disaster risk information and promoting mitigation behaviors. Risk-aware residents with good digital skills tended to harness social media platforms as additional information channels and prepare for disasters more actively (e.g., purchasing insurance and storing emergency supplies) compared to residents with fewer social media skills. This research suggests that practitioners should be more cautious when using digital tools to disseminate information, as many individuals in subsidized housing, including those with pre-existing vulnerabilities, may have difficulty accessing digital information. Strategies including digital literacy training and targeted communication campaigns should be considered to ensure the effectiveness of risk communication in vulnerable communities.


Urban development historically tends to occur in riverbanks and coastlines, due to the perceived benefits of being close to water resources, fishing and agriculture industries, and convenient transportation. However, proximity to water exposes human settlements to flood hazards (Wing et al., 20181). In the United States, estimates suggest more than 41 million residents are currently living in 100-year flood zones; by 2100 the number of residents at risk is expected to climb to 120 million (Wing et al., 2018). Additionally, the frequency of coastal flooding has doubled during the last three decades due to sea level rise (Sweet et al., 20182). The increased exposure to more frequent and severe flooding has imposed greater risks on human societies than ever before. In response, public agencies have implemented abundant mitigation strategies. The Federal Emergency Management Agency (FEMA) and the Department of Housing and Urban Development (HUD) have administered flood insurance programs to help recover damaged properties and buyout programs to encourage vulnerable residents to retreat from floodplains (Cusick, 20183). At the household level, financial and technical support is provided to reinforce housing structures and subsidize impacted households for recovery (Chatterjee & Mozumder, 20144; FEMA, n.d.5).

Individual residents can also reduce their flood risk by proactively adopting risk mitigation behaviors such as purchasing flood insurance or strengthening housing structures (Haer et al., 20176). Prior to adopting mitigation behaviors, however, residents must be aware of the flood risks (Fielding, 20187). Previous studies found that the risk awareness of residents is influenced by their previous disaster experiences and socioeconomic status (SES) (Chatterjee & Mozumder, 2014, Fielding, 2018; Peacock et al., 20058). Timely and effective communication between public agencies and flood-vulnerable residents can also enhance risk awareness (Terpstra et al., 20099; Kammerbauer & Minnery, 201910).

Risk communication plays a crucial role in preventing deaths, injuries, and property loss during disasters. For example, warnings sent before a disaster can give residents sufficient time to shelter in a safe place or evacuate (Kammerbauer & Minnery, 2019). However, the effectiveness of risk communication is closely associated with individual capacity to access and comprehend risk information. Digital media—including webpages, social media posts, and online news—have become an important way to convey information about disaster risk. As such, uneven access to the internet and different levels of digital skills, which is known colloquially as the “digital divide,” may hamper effective risk communication and, consequently, influence individual disaster preparedness and mitigation behaviors. Previous studies have shown the extent of the digital divide across socio-economic groups (Brodie et al., 200011; Hidalgo et al., 202012; Dargin et al., 202113). Older adults with low income and low educational attainment tend to be less able to access the internet and less proficient in using digital devices, for example. However, no research has studied the relationship between the digital divide and individual flood risk awareness and mitigation behaviors, with the exception of for a few recent studies which demonstrated correlated spatial patterns between Twitter post frequencies, disaster impacts, and sociodemographic characteristics (Stokes & Senkbeil, 201714; Zou et al., 201915; Dargin et al., 2021). Moreover, most extant studies on the digital divide investigate inequities in device ownership and internet accessibility without considering how these inequities affect people’s capacity to acquire credible and actionable information online.

As government agencies increasingly use online platforms to communicate crucial disaster information with the public (Lovari & Bowen, 202016), it is important to understand how these emerging communication tools shape individual disaster risk awareness and adoption of mitigation behaviors. In addition, recent studies in crisis informatics have suggested that user-generated data, such as social media posts, can be leveraged to assist in various disaster management tasks, including early warning, damage assessments, workforce tasking, among others (Wang et al., 201717; Wang & Taylor, 201818, 201919; Yao & Wang, 202020; Hao & Wang, 202121). However, these approaches may further marginalize vulnerable groups that are not online regularly (Yang et al., 202022). Identifying the barriers that vulnerable groups in flood-prone communities face in using online risk information could help address their lack of access to risk information and ultimately contribute to more equitable risk communication.

It has been widely acknowledged that disasters unevenly impact different social groups. Many scholars attributed the uneven disaster impacts to asymmetrical hazard exposures (Walker & Burningham, 201123). For example, low-income subsidized households are more likely to live in hazard-prone areas because of their lower housing costs (Chakraborty et al., 202124). We explore whether the digital divide exacerbates the disaster risks inherent to living in flood-prone areas. We use the term "digital vulnerability" to describe individuals who lack digital capabilities to access online information on disaster risk, preparedness, and mitigation. We tested three hypotheses, outlined in the next section, to investigate whether digital vulnerability interacts with risk perception and mitigation among individuals living in subsidized housing. Our survey respondents had varying socioeconomic and demographic characteristics. Results from this study have the potential to contribute to effective crisis communication and enhance capacities to mitigate flood risks among vulnerable groups.

Literature Review

The digital divide refers to the heterogeneous adoption of the internet and other Information and Communication Technologies (ICTs) across demographic and socioeconomic groups. Three levels of the digital divide have emerged as ICT development has evolved; these are: (1) the diffusion of digital devices and internet services (Wresch, 199625); (2) the skills and abilities required to use ICTs (Hargittai, 200126); and (3) the skills and abilities required to yield tangible outcomes of Internet use (Scheerder et al., 201727). Previous studies have related the digital divide to various socioeconomic and demographic factors including gender, income, age, race, education, and employment status, (Hidalgo et al., 2020, Zou et al., 2019).

Social vulnerability describes the characteristics of social groups that “influence their capacity to anticipate, cope with, resist, and recover from the impact of a natural hazard” (Wisner et al., 2004, p. 1128). Hypothetically, digital media can reduce population vulnerability to flooding in at least three ways. First, it can provide vulnerable residents with additional channels for disaster information (Allaire, 201629). Particularly, the various online information channels can provide up-to-date and engaged communications during disasters (Lovari & Bowen, 2020). For example, residents can learn about imminent disasters and preparedness measures from the websites of official agencies as well as social media posts from online peers which could possibly influence individual decision-making regarding evacuation and return (Yabe et al., 202130). Second, ICTs provide disaster responders with tools to disseminate information about disaster situations and recommended actions (e.g., evacuation and sheltering) during a disaster event. Some emergency managers found an increasing number of followers of official social media accounts during disasters (Lovari and Bowen, 2020; Houston et al., 201531). In this context, the digital divide could degrade the effectiveness of risk communication between public agencies and affected residents. Third, the online social networking platforms enhance social ties among impacted residents and connect them to outside assistance. Individuals can share resources and exchange information on online platforms, which promote collaborations in disaster recovery activities (Metaxa-Kakavouli et al., 201832; Yabe et al., 201933). Increased social connectedness online post-disaster has also been shown to reduce mental stress of affected residents and help them feel more supported and optimistic about the future (Taylor et al., 201234).

Research Questions

The literature review suggests that online platforms may facilitate risk communication and promote mitigating behaviors. However, existing studies on the potential advantages of using social media for disaster management have rarely considered the impact of the digital divide. This study addresses this gap. Specifically, we surveyed residents in subsidized housing living in flood-prone areas. Such residents tend to be more vulnerable to disasters due in part to their limited resources to address flooding risks (Yager & Rosoff, 201735). Subsidized housing residents are also more likely to experience one or more of the multiple dimensions of the digital divide described above (HUD, 201636).

In this study, we tested three hypotheses about the relationship between digital access and skills and flood vulnerability of subsidized residents living in flood-prone areas in Florida:

  • Hypothesis 1. Subsidized housing residents who are age 58 or older, unemployed, racial minorities, and earn incomes below $25,000 are less likely to be proficient in digital skills (Hypothesis 1a) and, also, less likely to use social media (Hypothesis 1b).
  • Hypothesis 2. Subsidized housing residents with higher proficiency in digital skills, indicated by higher scores on the Social Media Skill questionnaire, are more likely to use social media platforms to search for and share disaster-related information (Hypothesis 2a) and to find that information useful (Hypothesis 2b).
  • Hypothesis 3. Residents with more information channels and active social media usage will be more informed about (Hypothesis 3a) and more prepared for disasters (Hypothesis 3b), as indicated by responses to questions about acquiring disaster information and the Disaster Preparedness questionnaire.


Study Area and Survey Distribution

We distributed surveys in 798 Zip Code Tabulation Areas zones in Florida that met the following criteria: (a) More than 25% housing units were located in 100-year floodplains and (b) contained subsidized housing properties. We focused on Florida because the state contains many flood-prone areas and some of the most socio-economically diverse neighborhoods in the United States.

Respondents were recruited by the survey company Qualtrics to fill a panel with a set of nested quotas referring to the census distribution. We recognize that using Qualtrics to recruit research participants is a limitation of this study because Qualtrics recruits individuals online and thus these recruits are more likely to have internet access and higher digital literacy skills. This means our survey was unlikely to include the most “digitally vulnerable” individuals, those with very high access barriers and very low digital skills. However, this research focused more on high-level digital divide about yielding tangible outcomes of Internet use and, as explained in more detail below, our sample did include many individuals with a range of digital skills. This allowed us to investigate whether higher levels of digital literacy enabled individuals to use the internet to inform themselves about disasters and mitigation behaviors more effectively than individuals with lower levels of digital literacy.

Qualtrics emailed potential respondents a description of the study’s purpose and a secure URL to access the survey. An electronic informed consent form explained that their participation was voluntary and their responses would remain anonymous. We set the screening question “Do you live in subsidized housing?” to ensure sampled participants lived in subsidized housing units. These proportional quota sampling techniques are found to produce accurate, valid samples of a population because the samples are drawn from a large panel based on known criteria from the population (Miller et al., 202037; Ansolabehere & Schaffner, 201538). The surveys took respondents 10-15 minutes to complete and Qualtrics provided them with monetary compensation. All data collection procedures were approved by the University of Florida Institutional Review Board (IRB) (IRB202001712). Finally, we obtained 200 survey responses between March 17, 2021, to April 7, 2021.

Survey respondents had varying socioeconomic and demographic characteristics (see Table A1 in the Appendix for a full description). More than half of the respondents were between ages 18 and 28. There were slightly more females than males and 55% of respondents are unmarried. More than half (56.5%) of the respondents were White and 32% were Black. About 48% of respondents have full-time jobs, 23% are part-time employed, and the other 29% include students, retirees, and unemployed people. Overall, survey respondents were more likely to be younger, lower-income, unemployed, and nonwhite than the demographic profile of the state (U.S. Census Bureau, 2021).


Descriptive Statistics

The following subsections summarize the responses that survey participants gave regarding their channels for acquiring disaster information, their social media skills, and the risk mitigation behaviors they have adopted.

Channels for Acquiring Disaster Information

We asked respondents to grade the easiness of acquiring information on disaster preparedness and response. More than one-third (34%) considered it hard to get information about what to do during a disaster while 44% of respondents do not think so (see Figure A1 in the Appendix). More than half of respondents stated that text alerts, TV news, mobile phone weather apps, and acquaintances were reliable channels for acquiring information about disasters. As displayed in Figure 1, social media was the second most common source of information about flooding events in the survey; more than half of respondents (56.5%) said they used this channel. However, most survey respondents did not rely on only one source of information. In fact, 55% used at least four information channels to learn about upcoming or ongoing events (see panel b in Figure 1).

Figure 1. Types and Number of Channels of Information About Flooding Events

Channels of Information

Note. N = 200. Panel (a) displays how many respondents used each channel of information for information about flooding events. Panel (b) shows the percentage of respondents that had between zero and nine channels of information.

We also asked respondents about the specific social media platforms that they used. Instagram, YouTube, Facebook, and Snapchat were the most popular services among the respondents. Most (76.5%) had accounts in at least four different social media platforms. (See Figure A2 in the Appendix for more details). As depicted in Figure 2, nearly three-quarters of those respondents (73%) with social media accounts indicated that they followed social media posts from public officials or the local news. About half of the respondents followed haring al media post from the local news, public health agencies (e.g., Florida Department of Health), state and federal government agencies (e.g., FEMA), or local government websites. Relatively fewer respondents followed non-profit organization accounts (e.g., American Red Cross).

Figure 2. Respondents Following Social Media Posts from Public Officials or Local News

Following Social Media Accounts of Officials

Note. N = 146. Analysis restricted to those respondents with their own social media accounts. Panel (a) displays the percentage of respondents that follow social media accounts of public officials or the local news. Panel (b) shows how many respondents followed each type of public official or local news account.

Respondents also expressed positive views about the usefulness of social media for acquiring disaster information. Seven in ten respondents expressed that they were very likely to search for information about disasters on social media platforms during an event. A similar percentage (72%) found that information useful in helping them to prepare for and respond to disasters. (See Figure A3 in the appendix for more details.)

Proficiencies in Social Media Usages

We asked respondents to evaluate their abilities in a variety of social media skills. As displayed in Figure 3, over half of respondents rated their abilities to share multi-media posts and search for targeted information as extremely proficient. On the other hand, only 10-15% of respondents said that they did not have these basic social media skills. Overall, relatively fewer (35.5%) respondents were confident about their ability to find credible online information sources. Our statistical analysis showed that the answers in Figure 3 were highly correlated and we performed principal component analysis (PCA) for the self-graded scores. We used the first principal factor to create a composite variable called “Social Media Skill” which measures respondent’ overall proficiencies in using social media platforms. The composite variable explains 57.9% variances for the varied self-evaluated proficiencies of social media usage.

Figure 3. Respondent Self-Evaluation of Their Social Media Skills

Social Media Skills

Note. N= 200. Respondents evaluated their skills on a scale of 1 to 5, with 1 representing “do not know at all” and 5 representing “extremely proficient.”

Degree of Disaster Preparation

We asked respondents, all of whom live in either 100- or 500-year flood zones, to identify if they are exposed to flooding hazards. Only 44% of respondents indicated that they lived in flood-prone areas. An additional 19% were uncertain about their flood exposure. More than one-third of respondents falsely believed that they were not living in flood zones. (See Figure A4 in the Appendix.)

We also asked respondents about their preparation or mitigation behaviors in response to flooding risks. Only one-third of respondents said they had flooding insurance. Another one-fourth were unsure if they had flooding insurance. Of those who had no flood insurance, 13% had let their coverage expire and 28% had never purchased flood insurance. Among those with flood insurance, they stated their main reasons for purchasing it was concern about flood risks and the benefits of insurance to reduce flooding costs. Respondents who never enrolled in flooding insurance, on the other hand, believed that their houses were not likely to be flooded. They also stated that they lacked information on flooding insurance and underestimated the costs of flooding damage. Respondents who previously had flooding insurance but later let it lapse said that high insurance costs and lack of awareness of flood risks were their main reasons for not purchasing insurance again. (See Figure A5 in the Appendix for more details.)

We also asked respondents about other emergency preparedness behaviors. Around 80% of respondents said that they stored emergency supplies at home and 61% said they stored emergency supplies outside the home in cars or workplaces. Water, food, flashlights, and first-aid kits were the most common emergency supplies that respondents had stored. A majority (76%) of respondents with disaster supplies replenished their supplies at least once a year. More than three quarters (78%) had family emergency plans, including detailed instructions for household members regarding what to do and where to go during disasters. (See Figure A6 in the Appendix for more details.)

We performed a PCA analysis on responses to the preparedness and mitigation behavior questions and used the first principal factor as the metric “Disaster Preparedness”. This composite variable measures respondent’ overall readiness for flooding events and explains 45.7% variances of the varied flooding preparedness actions.

Hypothesis Testing

We used SAS Analytics Software (www.sas.com) to test the three hypotheses outlined above. The following subsections present the results of the analysis.

Hypothesis 1. Digital Divides Across Demographic Characteristics

We examined how digital skills and social media involvement were distributed across socio-economic and demographic characteristics with one-way ANOVA and t-tests. The results showed that the mean value of Social Media Skill varied significantly across age and race. White respondents under 58 rated themselves as more proficient in using social media platforms than older or non-white respondents. These findings partially confirm Hypothesis 1a that individuals’ digital skills are correlated with their demographic characteristics. However, socioeconomic factors had only a minimal effect on digital skills according to our analysis. One explanation could be the high rate of adopting (i.e., 85%) smartphones in the United States (Pew Research Center, 202139), which makes it challenging to detect nuanced variations in digital skills among residents of different SES with a small sample size. Our analysis also shows that respondents under age 58 have accounts on five to six different social media platforms while respondents older than 58 use fewer platforms. Additionally, respondents who have full or part-time jobs use more social media platforms than unemployed respondents. The presented results support Hypothesis 1 that social media usage and proficiencies are associated with their socioeconomic and demographic background (See Tables A2 and A3 in the Appendix for full results of the statistical tests).

Hypothesis 2. Social Media Platform Usages and Flood Risk Awareness

The Spearman correlation analyses show that respondent’ Social Media Skills are positively correlated with their use of social media for disaster information as well as their perception of the usefulness of social media platforms as disaster information channel. This analysis confirms Hypotheses 2a and 2b. Additionally, a series of t-tests revealed that respondents with more awareness of their flood risks were more likely to follow the social media accounts of public officials or local news agencies, in general, than those who were unaware of their flood exposure. These results confirmed Hypothesis 2c. (See Tables A4 and A5 in the Appendix for the full results of these statistical tests.)

Hypothesis 3. The Digital Divide and Individual’ Flood Preparedness

Our analysis of the relationship between the channels respondents use to search for disaster information and their flood preparedness produced some interesting results. First, it showed that respondents who use conventional media (i.e., TV news and Newspapers) considered it easy to find information about disaster preparedness. In fact, individuals who used conventional media reported less difficulty acquiring information about disaster preparedness than those who did not. On the other hand, individuals who relied on digital information sources, such as online news articles and social media posts, found it more difficult to find information about disaster preparedness. (See Table A6 in the Appendix for full results.)

We used the Spearman correlation to test whether individuals with more channels of disaster information had less difficulty acquiring actionable information, which showed an insignificant correlation. In sum, we did not find evidence to support Hypothesis 3a that more information channels and the use of social media platforms reduced difficulties in acquiring actionable disaster information. The results indicate that residents living in subsidized housing units still tend to rely on conventional media for acquiring credible disaster information.

The t-tests showed that respondents who followed social media accounts of public officials and local news agencies, except for non-profit organizations, were more likely to adopt risk mitigation behaviors (See Table A7 in the Appendix for the details of statistical tests). This correlation is especially significant for emergency management and public health agencies. It is unclear from our analysis what role social media played in contributing to respondents adopting new mitigation behaviors. For example, respondents with higher awareness of flooding risks were also more likely to follow official social media accounts. As a result, we could not say whether respondents increased their disaster preparedness after finding information about the topic on social media platforms or if they did so because of their increased risk awareness, or both. Our analysis, however, did confirm Hypothesis 3b that individuals who follow social media accounts from public officials and local news agencies are more likely to prepare for disasters.


Implications for Practice and Policy

Our findings yield the following action-oriented implications for mitigating the impact of digital divide in vulnerable communities. First, public agencies need to connect with populations at risk via versatile channels to communicate risk information. Especially, flexible, tailored, and targeted communication should be considered to make the crucial information easily accessed and comprehended by different social groups during disasters. Respondents in our study tended to have high rates of social media usage, yet the number of social media platforms they used and the proficiency of their digital skills varied across socio-demographic groups. Older adults and non-whites were less confident about their ability to search for and share disaster information on social media. Additionally, older adults and unemployed people had fewer social media accounts. These findings suggest that older adults, nonwhites, and unemployed people face barriers to accessing online risk information. Public health and emergency management agencies should consider the needs of these vulnerable groups in designing their disaster communication strategies and consider providing trainings to vulnerable communities on basic digital skills and online communication to enhance their capacities to cope with emergency events.

In addition, designers of civil participation portals and data-driven decision-making systems that take user-generated data should anticipate the representativeness bias of their users and take pro-active steps to increase fairness, for example, by curbing the algorithms to balance the sample distribution among different social groups (Yang et al., 2020). We believe that the enhancements of digital equity in risk communication can ultimately turn into more equitable and just disaster responses that contribute to disaster resilience.

Our results also indicate that residents with higher levels of risk-awareness actively seek information about disaster preparedness and mitigation by following social media accounts of public officials and local news agencies. This suggests that social media can serve as an effective information channel for risk-aware individuals. Stakeholders, including emergency management and public health agencies, may consider strategies increasing residents' risk awareness along with campaigns promoting risk communication on digital media.

Lastly, our findings suggest that the inclusion of social media platforms does not lower the perceived difficulties in acquiring actionable disaster information. Respondents still tend to rely on conventional TV and newspapers to search for disaster information. This may be because that online information may lack clear messages or instructions or individuals may be so overwhelmed by the amount and diversity of information they find online that they are unable to act on it, which should be addressed in future risk communication campaign.


This research has several limitations. First, as mentioned above, we used Qualtrics to recruit participants and distribute surveys. This means our survey sample did not include the most “digitally vulnerable” individuals, those with very limited or no internet access. As a result, the “digital vulnerability” of older adults, nonwhites, and unemployed people that we identified may be even more pronounced than the research findings due to recruiting bias. Future studies should consider alternative survey recruitment and distribution approaches, such as telephone or in-person surveys. Second, since we focused on subsidized housing residents, our sample size of 200 respondents was relatively small. Focusing on subsidized housing residents in flood-prone areas allowed us to provide implications for direct public interventions to address the digital divide among this particular subgroup. However, due to the small sample size, many of the results were not statistically significant. Third, due to the cross-sectional survey design, our results revealed associations between many attributes, but it demands further investigations for causal inference. For example, the results indicated that risk-aware respondents tend to actively follow official social media accounts and are more likely to implement more disaster preparation actions. However, it is less clear regarding whether or how the information posted on official social media accounts promotes respondents' disaster preparation. Such knowledge may require in-depth interviews or designed experiments such as randomized controlled trials.


  1. Wing, O. E., Bates, P. D., Smith, A. M., Sampson, C. C., Johnson, K. A., Fargione, J., & Morefield, P. (2018). Estimates of present and future flood risk in the conterminous United States. Environmental Research Letters, 13(3), 034023. 

  2. Sweet, W. V., Marra, J. J., Dusek, G., & Pendleton, M. (2018). 2017 State of US High Tide Flooding with a 2018 Outlook. Supplement to State of the Climate: National Overview for May 2018. Retrieved May 3, 2021, from [https://www.ncdc.noaa.gov/monitoring-content/sotc/national/2018/may/ 2017_State_of_US_High_Tide_Flooding.pdf](https://www.ncdc.noaa.gov/monitoring-content/sotc/national/2018/may/ 2017_State_of_US_High_Tide_Flooding.pdf) 

  3. Cusick, D. (2018, June 7). FEMA Approves Buyout Funds for Houston Homes Flooded by Harvey. Scientific American. https://www.scientificamerican.com/article/fema-approves-buyout-funds-for-houston-homes-flooded-by-harvey/ 

  4. Chatterjee, C., and Mozumder, P. (2014). Understanding household preferences for hurricane risk mitigation information: Evidence from survey responses. Risk analysis, 34(6), 984-996. 

  5. Federal Emergency Management Agency. (n.d.). Individual Assistance. Retrieved April 29, 2021, from https://www.fema.gov/assistance/individual

  6. Haer, T., Botzen, W. W., de Moel, H., & Aerts, J. C. (2017). Integrating household risk mitigation behavior in flood risk analysis: An agent-based model approach. Risk Analysis, 37(10), 1977-1992. 

  7. Fielding, J. L. (2018). Flood risk and inequalities between ethnic groups in the floodplains of England and Wales. Disasters, 42(1), 101-123. 

  8. Peacock, W. G., Brody, S. D., & Highfield, W. (2005). Hurricane risk perceptions among Florida's single-family homeowners. Landscape and Urban Planning, 73(2-3), 120-135. 

  9. Terpstra, T., Lindell, M. K., & Gutteling, J. M. (2009). Does communicating (flood) risk affect (flood) risk perceptions? Results of a quasi‐experimental study. Risk Analysis: An International Journal, 29(8), 1141-1155. 

  10. Kammerbauer, M. & Minnery, J. (2019). Risk communication and risk perception: lessons from the 2011 floods in Brisbane, Australia. Disasters, 43(1), 10-134. 

  11. Brodie, M., Flournoy, R. E., Altman, D. E., Blendon, R. J., Benson, J. M., & Rosenbaum, M. D. (2000). Health Information, the Internet, and the Digital Divide: Despite recent improvements, Americans' access to the Internet—and to the growing body of health information there—remains uneven. Health affairs, 19(6), 255-265. 

  12. Hidalgo, A., Gabaly, S., Morales-Alonso, G., & Urueña, A. (2020). The digital divide in light of sustainable development: An approach through advanced machine learning techniques. Technological Forecasting and Social Change, 150, 119754. 

  13. Dargin, J. S., Fan, C., & Mostafavi, A. (2021). Vulnerable populations and social media use in disasters: Uncovering the digital divide in three major US hurricanes. International Journal of Disaster Risk Reduction, 54. 102043. 

  14. Stokes, C. & Senkbeil, J. C. (2017). Facebook and Twitter, communication and shelter, and the 2011 Tuscaloosa tornado. Disasters, 41(1), 194-208. 

  15. Zou, L., Lam, N.S., Shams, S., Cai, H., Meyer, M.A., Yang, S., Lee, K., Park, S.J. & Reams, M.A. (2019). Social and geographical disparities in Twitter use during Hurricane Harvey. International Journal of Digital Earth, 12(11), 1300-1318. 

  16. Lovari, A. & Bowen, S. A. (2020). Social media in disaster communication: A case study of strategies, barriers, and ethical implications. Journal of Public Affairs, 20(1), e1967. 

  17. Wang, Y., Wang, Q., & Taylor, J. E. (2017). Aggregated responses of human mobility to severe winter storms: An empirical study. PloS One, 12(12), e0188734. 

  18. Wang, Y. & Taylor, J. E. (2018). Coupling sentiment and human mobility in natural disasters: a Twitter-based study of the 2014 South Napa Earthquake. Natural hazards, 92(2), 907-925. 

  19. Wang, Y. & Taylor, J. E. (2019). DUET: Data-driven approach based on latent Dirichlet allocation topic modeling. Journal of Computing in Civil Engineering, 33(3), 04019023. 

  20. Yao, F. & Wang, Y. (2020). Domain-specific sentiment analysis for tweets during hurricanes (DSSA-H): A domain-adversarial neural-network-based approach. Computers, Environment and Urban Systems, 83, 101522. 

  21. Hao, H., &d Wang, Y. (2021). Assessing Disaster Impact in Real Time: Data-Driven System Integrating Humans, Hazards, and the Built Environment. Journal of Computing in Civil Engineering, 35(5), 04021010. 

  22. Yang, Y., Zhang, C., Fan, C., Mostafavi, A., & Hu, X. (2020). Towards Fairness-Aware Disaster Informatics: An Interdisciplinary Perspective. IEEE Access, 8, 201040-201054. 

  23. Walker, G. & Burningham, K. (2011). Flood risk, vulnerability and environmental justice: evidence and evaluation of inequality in a UK context. Critical social policy, 31(2), 216-240. 

  24. Chakraborty, J., McAfee, A. A., Collins, T. W., & Grineski, S. E. (2021). Exposure to Hurricane Harvey flooding for subsidized housing residents of Harris County, Texas. Natural Hazards, 106(3), 2185-2205. 

  25. Wresch, W. (1996). Disconnected: Haves and have-nots in the information age. Rutgers University Press. 

  26. Hargittai, E. (2001). Second-level digital divide: Mapping differences in people's online skills. https://arxiv.org/abs/cs/0109068

  27. Scheerder, A., van Deursen, A., & van Dijk, J. (2017). Determinants of Internet skills, uses and outcomes. A systematic review of the second-and third-level digital divide. Telematics and Informatics, 34(8), 1607-1624. 

  28. Wisner, B., Blaikie, P., Cannon, T., & Davis, I. (2004). At risk: Natural hazards, people's vulnerability and disasters (2nd ed.). Routledge. 

  29. Allaire, M. C. (2016). Disaster loss and social media: Can online information increase flood resilience? Water Resources Research, 52(9), 7408-7423. 

  30. Yabe, T., Rao, P. S. C., & Ukkusuri, S. V. (2021). Modeling the Influence of Online Social Media Information on Post-Disaster Mobility Decisions. Sustainability, 13(9), 5254. 

  31. Houston, J.B., Hawthorne, J., Perreault, M.F., Park, E.H., Goldstein Hode, M., Halliwell, M.R., Turner McGowen, S.E., Davis, R., Vaid, S., McElderry, J.A. & Griffith, S. A. (2015). Social media and disasters: a functional framework for social media use in disaster planning, response, and research. Disasters, 39(1), 1-22. 

  32. Metaxa-Kakavouli, D., Maas, P., & Aldrich, D. P. (2018). How social ties influence hurricane evacuation behavior. Proceedings of the ACM on Human-Computer Interaction, 1-16. 

  33. Yabe, T., Ukkusuri, S. V., & Rao, P. S. C. (2019). Mobile phone data reveals the importance of pre-disaster inter-city social ties for recovery after hurricane maria. Applied Network Science, 4(1), 1-18. 

  34. Taylor, M., Wells, G., Howell, G., & Raphael, B. (2012). The role of social media as psychological first aid as a support to community resilience building. Australian Journal of Emergency Management, 27(1), 20-26. 

  35. Yager, J. & Rosoff, S. (2017). Housing in the U.S. Floodplains. Retrieved May 3, 2021, from https://furmancenter.org/research/publication/housing-in-the-us-floodplains 

  36. U.S. Department of Housing and Urban Development. (2016). Digital Inequality and Low-Income Households. Retrieved May 3, 2021, from https://www.huduser.gov/portal/periodicals/em/fall16/highlight2.html 

  37. Miller, C. A., Guidry, J. P., Dahman, B., & Thomson, M. D. (2020). A tale of two diverse Qualtrics samples: information for online survey researchers. Cancer Epidemiol Biomarkers & Prevention, 29(4), 731-735. 

  38. Ansolabehere, S., & Schaffner, B. F. (2015). Distractions: The incidence and consequences of interruptions for survey respondents. Journal of Survey Statistics and Methodology, 3(2), 216-239. 

  39. Pew Research Center (2021, April 7). Mobile Fact Sheet. https://www.pewresearch.org/internet/fact-sheet/mobile/ 

Suggested Citation:

Wang, Y., Hao, H., & Kang S. (2023). Disaster Risk Communication and Digital Vulnerability Among Subsidized Housing Residents (Natural Hazards Center Mitigation Matters Research Report Series, Report 12). Natural Hazards Center, University of Colorado Boulder. https://hazards.colorado.edu/mitigation-matters-report/disaster-risk-communication-and-digital-vulnerability-among-subsidized-housing-residents

Wang, Y., Hao, H., & Kang S. (2023). Disaster Risk Communication and Digital Vulnerability Among Subsidized Housing Residents (Natural Hazards Center Mitigation Matters Research Report Series, Report 12). Natural Hazards Center, University of Colorado Boulder. https://hazards.colorado.edu/mitigation-matters-report/disaster-risk-communication-and-digital-vulnerability-among-subsidized-housing-residents