Haiti After the Earthquake: Statistics Without Borders

Haiti after the earthquake Statistics Without Borders

When a major disaster strikes, urgent needs may be food, water, shelter, medicines – and data. Unless you know the numbers of people involved and how their lives have been affected, giving efficient help is impossible. Statistics Without Borders tries to provide the data. The team that worked on a project in Haiti describe one effort.

At approximately 5:00 p.m. on January 12th, Dozens of aftershocks of magnitude 4.5 Mw or • The government of Haiti estimated that 2010, the Caribbean Plate of the Enriquillo–Plan- greater occurred in the ensuing weeks, compli- 60% of the nation’s government, admin- tain Garden fault system moved eastward relative cating initial relief efforts and adding to the istrative, and economic infrastructure was to the North American Plate, resulting in a mag- suffering. destroyed. nitude 7.0 Mw earthquake on the eastern side of Hispaniola, the seismically active Caribbean This occurred in the aftermath of a series of se- island shared by Haiti and the Dominican Repub- The immediate aftermath of the vere hurricanes that hit Haiti in 2008, resulting lic. The epicentre of this earthquake, 18.457°N, earthquake in the deaths of almost 800 people and the de- 72.533°W, near the town of Léogâne, which is struction of approximately 60% of the country’s approximately 25 kilometres southwest of the What were the immediate ramifications when a harvest. Haitian capital and population centre Port- horrific natural disaster struck an area densely au-Prince, could not have occurred in a worse populated by people who were among the least location in Haiti. Over one third (approximately well equipped economically to deal effectively The need for data 2.5 million of approximately 7.2 million) of the with or recover from the resulting calamity? The population of Haiti lived in Port-au-Prince when United Nations Office of the Special Envoy for Each of these estimates is based on data. In the this earthquake struck, and almost half of the Haiti reports that: aftermath of any natural disaster, rapid and reli- country’s population experienced what is consid- able estimation of the extent of injuries, damage ered strong to extreme shaking. Compounding • an estimated 222 000 people (over 3% of to homes, and displacement of people as well as the difficulties created by this dense population the population) have died and an esti- the nature of the displacements (temporary or was the extreme poverty of the nation’s inhabit- mated 300 000 (over 4% of the population) permanent, current living conditions of the dis- ants. According to the Disasters Emergency Com- were injured. (Imagine the ramifications of placed, etc.) is critical. Even under such extreme mittee, before the earthquake: almost 7.5% of the population of any na- and harsh conditions (or perhaps I should say tion dying or being injured – even the best more accurately especially under such extreme prepared and wealthiest of nations would and harsh conditions), data must be collected to struggle mightily to recover.) inform decision-makers and increase the likeli- • 1.5 million people under 18 years old were hood that relief efforts are targeted in the most Disasters bring chaos. affected by the earthquake; approximately useful and helpful manner possible. Misdirected Helping involves finding out half of these were between 6 and 12 years or inappropriate aid helps no one; targeted aid, old. based on knowledge of the needs, helps most. what people need • The Haitian Ministry of Education esti- And this is how and where Statistics Without mates that 23% of all schools in Haiti (ap- Borders (SWB) enters the story. proximately 5000 schools) were affected by SWB (http://community.amstat.org/ the earthquake, and 80% of these schools statisticswithoutborders/home/) is an (almost 4000) were closed as a result of outreach group of the American Statistical As- • Haiti ranked 145th of 169 countries in earthquake damage. Again, imagine how sociation. It consists entirely of volunteers. It the UN Human Development Index (the any nation could deal with the abrupt loss provides pro bono (free!) consulting to organi- lowest rank of any nation in the Western of over 18% of its schools. sations and government agencies (particularly Hemisphere). • Estimated total damages and losses result- from developing nations) that do not have the • Over 70% of the population of Haiti was ing from the earthquake exceeded 120% of resources for statistical services and are working living on less than $2.00 per day. Haiti’s 2009 gross domestic product. on not-for-profit projects. In support of non-par- • 86% of the population of Port-au-Prince • Almost 50% of the nation’s population tisan and secular activities, SWB promotes the lived in deplorable conditions (in poorly could not live in their homes at the peak of use of statistics to improve the health and well- constructed concrete buildings), 50% had displacement, and the Haitian government being of all people. The organisation’s vision is no access to latrines, and 67% had no ac- estimated that over 100 000 houses were to achieve and implement the best statistical cess to tap water. completely destroyed. practice in the service of others.

SWB was established late in 2008 by Gary • to work with Haitian authorities and rep- had occurred overnight. Two calls per day were Shapiro of Westat, 2005 ASA President Fritz resentatives of other relief organisations necessary because the situation in Haiti was Scheuren, ASA Director of Science Policy Steve in Haiti to assess the situation and the extremely fluid and circumstances were still Pierson, and James J. Cochran of Louisiana Tech potential difficulties of data collection; changing rapidly. University. Initially SWB projects were pro bono • to coordinate with the SWB Project Direc- The Project Director would use the informa- projects that had been undertaken by the found- tor and other SWB volunteers on the design tion gathered by the SWB team on the ground ers prior to the establishment of the organisa- of a questionnaire and the plan for execut- in Haiti to revise SWB’s strategy and confer with tion. However, SWB’s membership grew quickly, ing the study; other SWB volunteers who were working on the and in late 2009 the organisation began focusing • to assist with cognitive testing, field test- project remotely. The team in Haiti had access to on finding partner organisations with projects to ing, and focus group testing of the draft cell phone signals and email at a small café in the which SWB volunteers could contribute. questionnaire; hotel where they stayed (SWB checked on this In late January 2010, soon after the earth- • to assist with translation and back-transla- well in advance of the trip), so documents such quake, Jean Orelien, a native Haitian and the tion of the questionnaire and development as drafts of the survey and instructions for the President of SciMetrika (a firm that focuses on of instructions for data collectors and their survey takers (as well as back-translations) could providing solutions to advancing human health), supervisors; be sent back and forth. Finally, Orelien identi- was directing an effort to collect data in Haiti to • to assist with a pilot test; fied two local professors in Haiti with graduate aid in the assessment of post-earthquake con- • to advise on revisions to the questionnaire training in statistics and demography, Robert ditions. Orelien contacted Sastry Pantula, who and data collection plan based on what Philippe and Wesner Antoine, who immediately was current ASA President, and was referred to was learned from the pilot test. became integral members of the team. They SWB. He asked the organization for help with the proved invaluable in the study design phase, de- design and execution of a survey in Haiti. Ore- The team needed time to get the necessary vac- velopment of the questionnaire, recruiting of the lien planned to use the data collected to assess cinations and make other arrangements. It flew interviewers and overseeing the data collection. the impact of the quake. Three SWB members into Port-au-Prince on March 12th and returned Having the participation of local researchers and (Jim Ashley, Justin Fisher, and Scheuren) were on March 19th. While in Haiti, the team would students served two purposes: their knowledge selected from several SWB members who quickly telephone a US-based SWB member every even- of the country and culture was instrumental in volunteered to go to Haiti. ing to give a progress report, and again every developing the questionnaire and at the same The group identified six purposes for trip: morning to provide updates on changes that led to institutionalizing the knowhow locally on

30 april2013 how to conduct a similar study. As an example of Data could be collected conveniently from requests. This was an added incidental benefit to how their local knowledge contributed, for the respondents throughout the country. The cost the SWB survey – it provided a humanitarian ser- question on marital status they told us that we of collecting data would be relatively low. The vice by getting information on available sources should include a category called “placeage” for time required to collect data would be greatly of aid to the survey respondents. those who were living together in a committed reduced. The safety of the data collectors would While the means for administering the survey relationship like spouses but were not married. not be at risk. Data collection from a centralised were being determined, the initial draft of the location would be possible, reducing the risk of questionnaire (which now reflected the objective lost data and enabling a much higher degree of of assessing the economic impact of the earth- Execution of the Project quality control in administering the ultimate quake) was under development, as was a manual questionnaire. And surprisingly, the response rate of instructions for the survey takers and their As the project progressed, Orelien and the SWB was high – much higher than would have been supervisors. These documents, prepared in con- volunteers came to realise that the broad initial expected in a door-to-door household survey. sultation with Orelien and several Haitian gov- goal of the project , to estimate mortality and The willing cooperation of those receiving the ernment officials and university members, were morbidity due to the earthquake, was already calls astonished and pleased us all. Even more originally written in English. Once a draft was being addressed by several other groups. Such surprisingly, the response rate was sent even complete, it was translated into Haitian Creole discoveries of duplication were to be expected. higher due to the willingness of those whose and back-translated to English repeatedly until Disasters are by their nature chaotic in their telephones were turned off when receiving the everyone was satisfied that reliable documents, early unfoldings; what is important is to dis- calls to use their caller ID functions and return cover duplicated efforts as soon as possible, and the survey taker’s calls and complete the survey! to adapt accordingly. As a result, the general Respondents were offered an inducement (credit goal was revised; the new aim was to estimate on their cell phone accounts) for completing the the economic impact of the earthquake – the survey. As with all other expenses associated Respondents initially believed changes in work status, household composition, with this project, the cost of providing these they were receiving a call from living conditions, and so on that it brought. This inducements was borne by SciMetrika. was an objective that, while not as dramatic, While the administration of the survey by a relief worker who could help was certainly important and one which could telephone greatly simplified many aspects of the with housing, medical care or potentially inform concrete and helpful actions. project, the approach was not a panacea. Cover- food and water The team also realised that a household sur- age was incomplete, non-response was still an vey of face-to-face interviews was not plausible, issue, and the scope of the survey objectives had and so one of the tasks of the team visiting to be reduced because of pragmatic limits to the Haiti would be to find a way to take a reliable length of a telephone interview. It is difficult to random sample of all segments of the Haitian expect someone to spend 20 minutes or more population. We were initially very concerned that on the phone responding to dozens of questions written in Haitian Creole, had been developed. finding a way to take such a sample would be under the best of circumstances; the tolerance of While differences in language make the need problematic, but once in Haiti the team found potential respondents is understandably far less for back translation obvious, a more subtle but that cell phone penetration was extraordinarily in the aftermath of a natural disaster such as equally critical reason for using this technique high (SWB later found that a household survey this, and we had to be sensitive to this issue as is to identify cross cultural differences. Orelien conducted by FAFO International in late 2009 we designed the questionnaire. Our judgment led and the SWB team on the ground in Haiti then estimated that two thirds of households in Haiti us to limit the questionnaire to no more than 25 worked with the survey takers and their supervi- had access to a mobile phone). Furthermore, questions in a format that could be administered sors to train them in the proper administration upon arrival the team found that one of the in less than 10 minutes. of the survey and recoding of the responses. first post-earthquake cottage industries that Many users own multiple phones, so there Once everyone had settled on the design and developed was providing means of recharging was a relatively high probability of contacting means of administering the questionnaire, we cell phones (one entrepreneur used an old movie a respondent multiple times. But the most gut- discovered that problems still existed despite projector to project movies onto a large sheet wrenching issue arose early in the data collection the diligence and best efforts of all involved. so his customers would be entertained while step. Several of the survey takers whom SWB Soon discrepancies in definitions assumed by they waited for their cell phones to receive full had trained reported that respondents initially the volunteers and those used by the Haitian charges). In addition, Haiti had only three major believed they were receiving a call from a relief respondents arose. Differences in interpretations providers of cell phone service (Voilà, Digicel, worker who could offer assistance with housing, of the concepts of household, employed, and and Haitel), and so a frame of potential respond- medical care, and food and water. Other survey and marital status (as explained earlier) were ents could be created by gaining the cooperation takers reported that while the potential respond- particularly troublesome; the definition of these of relatively few organisations. Thus, cell phones ents generally understood the nature of the call, concepts were not as straightforward as antici- were a viable means for collecting data from a they still asked the survey taker for help in find- pated. After the earthquake several families lived random sample of a vast majority of the Haitian ing sources of assistance. The SWB volunteers and together under the same roof or shared resources population. Orelien saw the emotional toll that being unable with neighbors. Households needed to be re-de- While this alone was welcome news, there to respond to these requests was extracting on the fined as those who eat from the same stew pot were many other potential benefits of this ap- survey takers (most of whom were young college and could easily include 25 or more people. The proach to sampling and data collection. Cell students). Information on how to get in touch nature of the Haitian economy leads Haitians phone users in Haiti are charged only for outgo- with organisations that were engaged in relief ef- to use a drastically different definition of em- ing calls, so responding to the survey would not forts was gathered and given to the survey takers, ployed given the lack of formal employment. The place any financial burden on the respondents. who were then instructed on how to handle these questionnaire was quickly revised to reflect the

april2013 31 common Haitian culture interpretations of these legal, economic, cultural, business, and govern- lesson; the discipline of statistics encompasses and other concepts. mental environments can create when working far more than analysis of data. Study design, We found ourselves racing against the clock. in a foreign country. The volunteers also realised collection of data, and presentation of data are It did not initially occur to any of us that by that the challenges associated with these differ- also key components of the discipline, and in a early May Haiti would be facing the rainy season, ences are generally greater when working in a difficult environment such as post-earthquake and once that occurred it would be very difficult developing nation. Still the team was surprised Haiti, proper study design and data collection to get a response because the people would be by some of the obstacles that arose, but not are the most important issues; if these steps otherwise occupied. We were also told that if surprised by being surprised – the team was re- are not done well, no worthwhile or meaningful we did not get this done before the World Cup alistic and expected some surprises. The first les- analysis can be done. started, we would have trouble getting people to son that one learns every time one is involved in answer our survey! In the event data collection was conducted from June 7th to June 11th – five Further work months after the earthquake – and, as we have said, the Haitian people put aside their other As well as working in Haiti, SWB has worked with concerns to respond to our questionnaires. When homes had been destroyed, representatives from Madagascar to improve Respondents were called between 9:00 a.m. health and education, helped American statis- and 4:00 p.m.; this was due to the availability the definition of “household” had tical students plan a survey on bottled water of offices. We had hired office space in Port-au- to change: it became “people who usage in Mexico, and worked with UNICEF to Prince for the ten interviewers to phone from. eat out of the same stew pot” determine the impact of community health work- Port-au-Prince was still a dangerous place, and ers treating childhood illnesses in Sierra Leone. security concerns would have prevented extend- We are always seeking further projects. We would ing interviews beyond 5:00 p.m. Our interviewers like to hear of organisations or government had to get safely home. Most interviewers were agencies (particularly in developing nations), university students in the latter part of their an endeavour such as this is humility. Although working on not-for-profit projects to improve studies whose university had not yet reopened the subject of this article is SWB’s role in these the health and well-being of people, that do (most universities and schools did not reopen efforts, we must emphasize that SWB supports not have the resources for statistical services until the following fall); they were extremely the efforts and initiatives of other individuals they need to effectively complete the project. dedicated and did a consistently outstand- and organizations, and SWB does not deserve or If you have a suitable suggestion please con- ing job, ultimately dialling 2649 numbers and wish to take credit for this project. Credit for the tact SWB at statisticswithoutborders@ completing 769 interviews. The response rate, project belongs to Jean Orelien, whose inspira- gmail.com or visit the SWB website after adjusting for nonworking numbers, was tion, perseverance, and commitment were the (http://community.amstat.org/ approximately 50 percent, and the cooperation driving force behind the genesis, development, statisticswithoutborders/home/). rate was in the high 90s; these outstanding and completion of this survey. We would also be numbers can be attributed to the conscientious remiss if we did not also acknowledge Robert References efforts of the interviewers and their supervi- Philippe and Antwone Wesner, who despite fac- 1. sors. The responses were recorded on paper by ing extreme hardship still provided invaluable http://www.theworld.org/2010/ 06/calculating-aid-for-haiti/. the interviewers, and the recorded results were guidance and insight into the culture and the 2. Sherlock, T. (2010) Statisticians help brought to the US by Orelien, who then arranged post-earthquake conditions in Haiti. Finally, the needy countries collect accurate data. Vancouver for the entry of these data into an electronic data collection supervisors and telephone inter- Sun, July 28th. http://www2.canada.com/ file. The anonymised Haitian post-earthquake viewers should all be commended for the quality vancouversun/news/westcoastnews/ cell phone survey data set is now available as of their work, which would have been impressive story.html?id=b2269163-17eb-49f2- part of a zip file that contains documentation even under the best of circumstances. The brav- bc7a-ea2520c12f28&p=1. and the data in both SAS and SPSS formats; ery and steadfastness in the face of incredible 3. www.amstat.org/sections/srms/ . this zip file can be downloaded from the SWB adversity shown by each of these individuals was jsmprg/jsm2010.htm 4. Ashley, J. D. and Scheuren, F. (2010) website at http://community.amstat. tremendously inspirational to the SWB members Considerations in the study design of a mobile phone org/StatisticsWithoutBorders/ who assisted with this project. survey of the Haitian population. JSM Proceedings. SWB_Projects/CurrentProjects/ Our experience on this project underscored Alexandria, VA: American Statistical Association. Haiti1. our belief that there is a need for more research 5. Fisher, J. (2010) Survey administration Stories on SWB’s efforts in Haiti also received on the effectiveness of cell phone surveys, es- in the wake of a natural disaster. JSM Proceedings. a great deal of coverage in the media. For exam- pecially in developing countries and especially Alexandria, VA: American Statistical Association. ple, the work was featured on Public Radio In- in the immediate aftermath of natural disasters ternational (the transcript and a recording of the (which disproportionately affect those living This article was written by members of the SWB team broadcast are both available1,2. The effort was in developing countries). A need for testing in Haiti. Jean Orelien is President of SciMetrika; Rob- also featured in a late-breaking session at the the effectiveness of other means of electronic ert Philippe is Strategic Information Advisor with 2010 Joint Statistical Meetings in Vancouver3. communication can also be seen. More detailed Futures Group International; Antwone Wesner is a discussions of these issues can be found in the professor with the State University of Haiti School references4,5. of Agriculture; James Ashley is Senior Mathematical Statistician with the US Government Accountability What did we learn You might be surprised to find that an ar- Office; Justin Fisher is Senior Statistician with the ticle on statistics applied to a natural disaster US Government Accountability Office; Fritz Scheuren SWB volunteers went into this project cognisant includes no discussion of any analyses of the is Senior Fellow and Vice President at NORC in the of the difficulties that differences in social, data that have been collected. That is the final Center for Excellence in Survey Research.

32 april2013