On March 13, 2015, Cyclone Pam, a Category 5 storm, battered the small Pacific island nation of Vanuatu, considered by the United Nations to be one of the world’s least developed countries. Winds of up to 185 miles per hour (300 kilometers per hour) flattened an estimated 90 percent of buildings in the capital. It was described by Oxfam Country Director Colin Collet van Rooyen as “one of the worst disasters ever seen in the Pacific.” But, after a few days’ worth of attention and a few million dollars pledged in aid, even as dozens of islands remained cut off and without food, the world simply moved on, though it will take years for Vanuatu to do the same, especially once the emergency aid runs out.

In 2007, Swedish researchers Thomas Eisensee and David Stromberg published a study in the Quarterly Journal of Economics detailing the results of their analysis of mass media treatment of around 5,000 natural disasters occurring between 1968 and 2002. They found that on average, a disaster in the Pacific must have 91 times as many casualties as an identical disaster in Europe to get the same news coverage. Disasters in Asia and Africa also struggle to make the news, needing figures of 43 and 45 times the casualties of a European disaster. These appalling biases in the way the news media respond to natural disasters are not just tied to a disaster’s location, but also its type. Volcanoes are deemed most newsworthy. To expect the same coverage (with all else, including region, being equal) an earthquake must have twice as many casualties as a volcano; a fire, twelve times as many; and storms, floods, and droughts need 280 times, 674 times, and 2,395 times as many casualties.

These biases are far more dangerous than those usually inherent in the media: human lives are at stake. It is easy to see that those natural disasters ignored by the news media would draw fewer private donations. Far more shockingly though, Eisensee and Strömberg found a strong correlation between whether a disaster is covered by news, and whether it receives aid from the US government (which consistently accounts for around a third of the emergency aid provided by members of the Organisation for Economic Cooperation and Development (OECD)). That is, media biases affect government aid decisions. This link is so strong that a disaster occurring during the Olympics, when the news media is otherwise occupied, must have three times as many casualties as on an ordinary day to have the same chance of receiving relief. And that is not because the US government has been distracted by the political issues that can cloud some Olympics, like the Munich Massacre of the 1972 games, or the massive boycotting of Moscow’s 1980 games. The same effect is found when those games are removed from the study. And if that were not enough, by tracking day-to-day news pressure since 1968, the researchers found that disasters on days with the highest news pressure must have six times as many casualties as those on low news pressure days to have the same chances of aid.

And keep in mind: these are disasters of the same type in the same region. One can only imagine how much more the massive biases of disaster type and region affect aid decisions. Is there a better way to think about international emergency aid?

On average, a disaster in the Pacific must have 91 times as many casualties as an identical disaster in Europe to get the same news coverage.

The biases of disaster type and region stem from two primary responses to natural disasters: shock and empathy. Shock, a short-lived response, is what causes more “dramatic” and “exciting” disasters to get so much news coverage, and can be a powerful motivator driving people to respond to a disaster. That the media can use imagery and statistics and the like to shock is not entirely a negative thing in that it does galvanize people to donate to emergency relief efforts. It is just a thoroughly imperfect motivator in that it is so utterly disconnected from human need, as measured by a disaster’s casualties.

Empathy, on the other hand, comes from identifying with the other, stirring a sense of “that could have been me.” This is where regional biases come in—the predominantly Western news media chooses stories about people and places that their audience identifies with, be that from travelling to the region, having family there, looking like the people there, or sharing a common language or religion. Whatever the commonality that stirs these sentiments, they, along with the aid, disappear with the news stories.

As things are now, places hit by natural disasters that catch the world’s fleeting attention can expect an initial frenzy of this shock and empathy, and the aid it brings with it. But a few weeks or months after the catastrophe, the long, slow work of rebuilding begins, a project often far more expensive than disaster relief. Countries like Japan, following 2011’s earthquake, and the United States, after Hurricane Katrina, have economies that can support such massive rebuilds. Countries like Vanuatu, however, rely absolutely on long-term economic support after a disaster to construct the robust buildings and infrastructure that generate the resiliency necessary for stability and economic recovery. It is this long-term support that is so difficult to generate.

Unlike shock, empathy has the possibility to be sustained over the long term, and, with thoughtful campaigns, can even be developed toward those most “other” if enough similarity can be conveyed. Where traditional media’s monopoly has been diminished in recent years, the internet is increasingly serving as the go-to platform of news and information for many demographics, making it the ideal vehicle for this sort of approach.

So the challenge is: how can the international community provide better (and hopefully more equitable) long-term sources of aid and support in the wake of tragic natural disasters?

The best place to start would be to design ways to retain the attention and commitment of those people compelled, for whatever reason, to make one-time donations. Longer-term campaigns that focus on frequent, specific updates about the needs and achievements of rebuilding efforts, the effectiveness of the aid money, and the stories of locals hit by the disaster would see people who showed some initial support be connected to the progress of reconstruction. Like the child sponsorship model that drives so much long-term development, one could imagine that some sort of “adopt-a-disaster” approach could be impactful. Forming these sorts of relationships could help cultivate longer-term empathy and a sense of engaged responsibility. Donors could follow rebuilding efforts and see how their money helps the region put itself back together. They could be connected to a school, an agency or aid worker in the area, or a particular family or town — anything that turns dry statistics into motivating, empathetic ties. This model of engagement would increase both the sponsors’ giving and the fulfillment they incur from it.

Media biases are powerful and dangerous, and can only really be countered by slow cultural changes. But these problems could be sidestepped, at least in part, by new models of engaging international audiences that transform casual, one-time donors into informed and involved supporters of medium-term recovery efforts. Unlike knee-jerk shock and empathy responses, relational models, more equitably connected to a community’s need, could lead to better (and more) international disaster aid. For countries like Vanuatu, with 90 percent of its crops destroyed, thousands still living in makeshift shelters, and barely a third of what the United Nations estimated as three months’ aid coming in, this could not be more important.