Monica Dashen, Statistics without Borders

Hurricane Matthew pummeled the southern and northwestern portions of Haiti, a Caribbean island, with 140 mph winds and torrential rain storms October 3–4, 2016. Rivers flooded, sweeping away personal belongings, crops, animals, and homes. People sought shelter in nearby schools and churches or with family members. The wind turned nails, tin roofs, and tree branches into daggers, increasing the chance of tetanus. Cholera outbreaks occurred. Medical facilities were destroyed.

Prior to Matthew, many people resided in poorly constructed concrete homes with limited access to latrines and clean water. Haiti is ranked 163 out of 183 countries in the UN Human Development Index, and Matthew made the situation worse.

The team (from left): Chauncy, Franzlande, Ginny, Emilio, Mackenson, and Antoine (Photo by Monica Dashen)

On October 7, the Dominicans were the first to arrive with 5,000 trucks filled with food and sanitation supplies. They repaired roads along the way and unloaded the aid at a distribution site in Port-au-Prince, the Haitian capital.

Other aid organizations followed suit. USAID delivered tarpaulins and sanitation kits. The UN offered food and mapped out Matthew’s path, including the location of closed cholera clinics. The Salvation Army deployed their troops to assess and repair damage to the schools in the hardest hit areas of the south. Dutch volunteer doctors stitched up injuries.

Typically, aid distribution from large organizations is coordinated through the Haitian government, and, in many cases, dropped off at a school or other public facility in a southern commune center. The people residing in the nearby sub-communes some 15 to 120 minutes away by car must find a way to get the aid from the center.

Damage from Hurricane Matthew in the southern part of Haiti

Disasters bring chaos, but data can clarify the situation. Aid organizations can readily report the “whats” and “hows” of the distribution. The UN’s Office of Coordination Humanitarian Affairs (OCHA) provides weekly updates of supplies delivered, as well as assessments. But the “whos” are unknown. There are fairly rural parts of Haiti in the south. Are only those people closest to the distribution site receiving the aid? Here is where Statistics without Borders (SwB) stepped in.

A Survey to Determine the ‘Whos’

Through a survey sponsored by the American Statistical Association, SwB sought to find out who received aid, who did not receive aid, and how Matthew affected them. This type of aid distribution/aftermath survey can be readily applied to an immediate crisis situation like Matthew or more prolonged ones like famines in Ethiopia.

Recognizing that a conclusion of “there was a hurricane and aid arrival and distribution is slow,” would be uninformative, Rachel Green of SwB and I spent time figuring out how to get at the inside story. We designed a survey instrument that assessed home damage, job loss, and family member displacement after Matthew. We also looked at whether people could recognize the danger signs of tetanus and cholera (both treatable diseases) and know where to go for treatment, given that Matthew destroyed many medical clinics.

The impact of the survey results was time sensitive. SwB members needed to be in the field to witness the aftermath well before the locals were back on their feet and before people left the shelters (e.g., schools) and returned to their repaired homes. First responders would be packing up shortly and getting ready to head home. Time was ticking and SwB needed to be on the ground.

On the Ground

I arrived five weeks after Matthew to conduct the survey. Reaching out to respondents in a timely manner would prove to be a challenge in Haiti. SwB members initially thought a mobile phone survey was the best candidate, because phone penetration is high in Haiti. In addition, people did not have to pay to receive incoming calls. Likewise, the survey could be administered and data collected quickly in tightly controlled conditions.

Phone penetration is lower in rural areas than urban areas, and much of Matthew’s damage occurred in the rural areas. Likewise, Haiti does not have area codes, making it difficult to target certain regions.

But phone penetration is lower in rural areas than urban areas, and much of Matthew’s damage occurred in the rural areas. Likewise, Haiti does not have area codes, making it difficult to target certain regions.

As a result, we conducted 100 face-to-face interviews in the south, along with 154 phone interviews throughout the country. Nationwide interviews allowed us to compare the effects of Matthew throughout the country. We expected some areas to be unscathed and others destroyed.

The key to implementing this survey was local knowledge of the area and culture. SwB recruited Antoine Wesner, a university professor who helped conduct a 2010 earthquake survey.

Commune and sub-commune section selection was contingent upon road conditions and lodging (Matthew destroyed many of the guest houses and roads). Unfortunately, detailed maps at the subsection level (4th administration level) and up-to-date population maps are nonexistent in Haiti. To figure out the beginning of one sub-commune and the end of another, Wesner recommended we ask the people to show us their ID cards. In one instance, a subsection was so spread out that the team spent 30 minutes driving from home to home asking about the boundaries before crossing into another subsection.

En route to our guest house in Le Cay, the team and I witnessed trees knocked over by Matthew and tarpaulins on tin roofs. Patches of shiny tin interlaced with tarnished tin could be seen readily from our truck. Had SwB arrived too late? Were these people back on their feet? To allay this fear, I quickly reminded myself of our plan to interview those people residing at the commune center and then those farther away at various sub-commune sections.

In the Field

Our first commune was about a 15-minute drive from the only thoroughfare in the south. We visited a local hospital first and the peoples’ homes afterward. The hospital located next to the major road was still running. The nurse reported an uptick of injuries, rheumatism, and tetanus cases after Matthew.

People residing in the commune center reported injuries from Matthew, as well as loss of their homes and animals. These people, who are farmers, cannot reseed or earn extra money from their crops, since Matthew swept away their seeds and crops. They must buy food or eat from their food stashes. When asked about damage to their homes, the people started to discuss their plans to rebuild, but tears quickly filled their faces. Sadness turned to anger as people reported not being able to pay a dollar or more for the right to stand in line for aid. Relatives and neighbors sitting nearby nodded and at times chimed in with their own stories. The interviewers politely nodded and returned the conversation to the respondent. Private interviews were difficult to obtain.

Lavender-scented antibacterial soap was used as an incentive. It was meant to honor the people while also having a health component, but we quickly learned the soap was not enough. The people viewed it as a tease, and they needed a lot more. A group surrounded us and angrily demanded money while we were packing up the truck and coordinating plans. These people felt that because we took pictures and asked questions about Matthew and aid, we should give them something besides soap. We listened and left. Such a reaction was unanticipated, as team members were in the south just prior to Matthew and did not experience this level of anger. The people were starving.

During the four days in the field, the team heard the same stories, but with twists. Those people who lived farther from the commune center could not afford the transport to the center to pick up the aid or simply did not know about it. Others reported that the sub-commune head brought in aid, but either kept it to himself or offered it at a price. Still others reported that the strongest prevailed when aid was tossed off a truck or their commune head did not have an “in” with the government to receive aid. One respondent thought aid should be delivered door-to-door with an armed guard. Returning to the truck, the team discussed the pros and cons of aid distribution. Door-to-door distribution was too time consuming for a large organization. Bulk deliveries clearly did not work. Micro-credit is a long-term solution. There was no simple short-run solution.

On our last day in the field, the team conducted interviews in communes along the coast. One woman who had a surprisingly positive disposition lost her 13-year-old daughter to Matthew and her husband to the 2010 earthquake. Her home was destroyed, but she remained there as she had nowhere else to go. Two American carpenters were working nearby repairing a neighbor’s home.

Our spirits continued to lift as we saw shiny tin roofs dotted along the hillside of the next sub-commune, which was an hour away. An NGO had delivered the roofs a week or two prior to our arrival. Likewise, a local priest secured beans and rice for his people in a nearby sub-commune. Aid was found in the most unlikely places, some two hours from the nearest commune center.

Our return to Port-au-Prince was delayed due to a road blockage. Here, angry protestors chopped down trees and lit the branches on fire because their children could not return to school. Matthew’s victims, whose homes were destroyed, still sought shelter in the school. Our guide paid the protestors $5.00 to move the branches and let our truck through. We stayed in Le Cay that night and returned to Port-au-Prince the following day to process the data and start the phone survey.

The Call Center

The following day, we sought to obtain 250 completed nationwide phone cases in three-and-a-half days. Our call center was the patio of the Coconut Villa hotel in Port-au-Prince. We were tired from our field experience, so we scaled down to 150 completed cases. The new members became fed up with dialing a string of 25 nonworking numbers before reaching a working one. A sample of randomly selected working phone numbers provided by the phone company was unavailable. Instead SwB produced a list of randomly generated phone numbers.

To reduce interviewer frustration, we tinkered with the randomly generated numbers. The team tried the “last digit up and down” approach and obtained more hits. That is, if the last digit of a working number was a six, a team member dialed a five. Taking this one step further, the team tinkered again with the last digit of known southern phone numbers (those obtained in the field). The thinking was that the phone companies sold blocks of phone numbers to people in various regions. Our first try led us to someone in the less accessible area of the south, but the remaining attempts led us to the north and west departments. These outcomes gave us an indication that the phone numbers were not sold in blocks.

Phone credit was given out at the end of each day to the people we interviewed, but was subject to phone company delays. Interviewers received “reminders” of phone credit delays at all times of the day and night, but they fizzled out toward the third day.

The last and fourth day was devoted to reinterviews. We wanted to make sure enough time had passed for everyone to receive credit, as delays in credit may lead to refusals. This was a mistake, because nearly everyone was at church Sunday morning and did not answer.

Eventually, the team was able to eke out nine reinterviews, but they complained that people changed their answers to some questions in the hopes of garnering more aid. The thinking was that if we were calling back to confirm their answers, we might give them something besides phone credit.

What’s Next?

The next steps involve producing tables of job loss, home damage, and aid distribution for the entire country. Greater home damage and job loss is expected for the Matthew-affected regions. The report generated from the tables will be sent to officials in the Haitian government and at USAID. Likewise, members of the Haitian diaspora who have medical background will be sent a report and asked to help spread the word about the danger signs of cholera, tetanus, and pneumonia. Here, the hope is that by spreading the word, awareness will be raised and long-term changes made.

As for short-term changes, some members of the Haitian diaspora recommend going directly to the people (e.g., delivering food and cash to those displaced people in shelters). Others recommend Christian Aid as an effective organization, along with Catholic Charities. For example, Christian Aid supports Haiti Survive, an NGO, that helps people rebuild their lives by giving them animals.

The old adage that “looks can be deceiving” applies to this situation. On the surface, it looks like the Haitians are rebuilding their lives five weeks after Matthew. Along the main road in the south, I spotted patched roofs, trucks loaded with food, and SUVs filled with western medical workers. These observations were consistent with the foreign aid reports I read. Getting off the main thoroughfare and walking around the communes on the first day, however, I was surprised to see the damaged homes and learn about the lack of aid.

Monica Dashen, who has a PhD in applied psychology and retired early from the Bureau of Labor Statistics, enjoys traveling overseas. Her last Statistics without Borders adventure involved a maternal and infant care survey in Myanmar.