The research team, led by Emmanuel Eze of Heidelberg University together with Phindile Tiyiselani Zanele Sabela-Rikhotso of the University of Cape Town and Alexander Siegmund of Heidelberg University of Education, assessed whether warnings contained six critical components long identified in the warning literature: timing, location, magnitude, expected impact, probability, and recommended response. Each affirmative component earned one point on a scale from zero to six. The results revealed a sharp geographic divide. Respondents in the Ethiopian sites reported substantially more complete messages than their South African counterparts, with awareness of warning dissemination, reliable sources, and appropriate responses also significantly higher there.

The most dramatic contrast emerged in Knysna, a comparatively urbanized coastal town on South Africa’s Garden Route. Despite the town’s more developed communication infrastructure, fewer than 30 percent of respondents affirmed any of the three awareness items, and more than two-thirds recorded an aggregate message-completeness score of zero. Most residents there received warnings almost exclusively through intermediaries—community leaders, officials, or neighbors—rather than firsthand. In Desa, by contrast, 86.9 percent of respondents received warnings directly, and message completeness was correspondingly higher. The pattern challenges a common assumption that technological advancement alone guarantees better risk communication.

Statistically, the association between reception mode and message quality was robust. Direct recipients reported an average of 4.0 of the six message components, compared with just 2.1 among those who received warnings only through intermediaries—a difference the authors report as highly significant, with a moderate effect size. The gap held for every individual component, and was widest for expected impact, which 81 percent of direct recipients reported receiving but only 38 percent of indirect-only recipients. Direct recipients were also more likely to report proactive responses to warnings—79.0 percent versus 69.1 percent—though the researchers caution that this association was modest and strongly patterned by site, so it cannot be interpreted as a causal channel effect.

The study’s theoretical framework draws on three complementary perspectives. Public risk communication theory emphasizes that warnings must be clear, specific, consistent, and credible to motivate action. The protective action decision model identifies message content and delivery as key determinants of whether people perceive warnings as personally relevant. Vulnerability theory, meanwhile, highlights how social stratification along lines of gender, age, livelihood, and socioeconomic status shapes who receives, understands, and can act on risk information. Integrating these lenses, the authors argue, reveals that warning effectiveness is contingent not merely on technical accuracy but on the alignment between message design, delivery mechanisms, and the social realities of receiving communities.

The qualitative portion of the survey painted a vivid picture of what is at stake when warnings fail. Respondents described cascading consequences of hazards: in Desa, one wrote that hunger and migration followed economic collapse, adding that their children could not attend school that year because school supplies were unaffordable. Others described waterborne disease, rising child mortality, livestock deaths, and environmental degradation. In Knysna, forest fires were reported to have displaced wildlife and degraded air quality, while mudslides damaged houses. These accounts underscore that hazard impacts ripple through livelihoods, health, education, and social cohesion—chains of consequence that generic warnings rarely articulate.

Perceptions of who was most vulnerable also varied by site, reinforcing the study’s argument that vulnerability is relational and context-bound rather than universal. In Desa and Yayu, women, children, and older people were most frequently identified as most affected, while men with cultivated land were often seen as least vulnerable. In Knysna, respondents pointed to residents of squatter camps and townships, contrasting them with those in suburbs and brick houses. In Phiring, answers ranged from older people and farmers to the whole community. Some respondents simply stated that no one was less affected, reflecting either universal exposure or uncertainty about assigning vulnerability hierarchies.

Perhaps the most actionable finding concerns personal relevance. Across all four sites, 86.9 percent of respondents said they would act more promptly if a warning explained why they were personally at risk. Among direct recipients in Desa and Yayu, that figure exceeded 98 percent. The authors suggest that warnings often convey what will happen but rarely articulate the “so what”—how a hazard intersects with specific livelihoods, assets, or social roles. Without that linkage, even technically complete messages may fail to motivate behavior, a gap most visible in intermediated systems where local messengers may lack the vocabulary to translate probabilistic forecasts into relatable narratives of vulnerability.

The findings carry significant implications for the United Nations’ Early Warnings for All initiative, which aims to achieve universal warning coverage by 2027. Africa currently has the lowest early warning coverage of any world region, at under 40 percent of its population, despite being among the most climate-vulnerable. The World Meteorological Organization estimates that 24 hours of advance notice can reduce disaster losses by up to 30 percent—but only if warnings actually translate into action. The study’s authors argue that infrastructure investment must be paired with equity criteria, such as coverage mapping for informal and rural settlements, and with people-centered dissemination architectures including community liaison networks, trained extension workers, and local leaders serving as communication nodes.

The researchers are careful to note the limitations of their exploratory, cross-sectional design. The four sites were purposively selected rather than statistically representative, convenience and snowball sampling were used, and reception mode was strongly confounded with site, so the associations should not be generalized to either country as a whole. Self-reported data introduce recall and translation biases—illustrated by anomalous reports of cyclones and volcanic activity in inland areas, which may reflect terminological mismatches between expert taxonomies and local risk idioms. Still, the study makes a compelling case that the pathway from forecast to protective action runs through social relations of access and meaning. As the authors conclude, early warning is both a political and technical enterprise, one that determines whose safety information counts and who remains unheard. Closing that gap, they argue, will require standardizing message components, auditing last-mile coverage, and co-designing warnings with the communities they are meant to protect.

Subject of Research: Early warning message completeness, reception mode, and protective action in at-risk communities in Ethiopia and South Africa

Article Title: Associations Between Early Warning Message Completeness, Reception Mode, and Protective Action Across Four At-Risk Communities in Ethiopia and South Africa

Article References: Associations Between Early Warning Message Completeness, Reception Mode, and Protective Action Across Four At-Risk Communities in Ethiopia and South Africa. (n.d.). https://doi.org/10.1007/s13753-026-00774-2

Image Credits: AI Generated

DOI: 10.1007/s13753-026-00774-2

Keywords: early warning systems, disaster risk reduction, risk communication, Ethiopia, South Africa, message completeness, protective action, vulnerability, EW4All, community resilience, impact-based warnings, reception mode