EDITORIALS
Bombarding people with questions: a reconsideration of survey ethics
Hirotsugu Aiga1
Japan International Cooperation Agency, Tokyo, Japan
"A survey, again? You are the third survey team who visited us during these couple of months. I am fed up with ..."
These words greeted me and my assistant at a house in Zalingei Internally Displaced Persons (IDPs) camp in Darfur in 2004. Health surveys play an increasingly critical role in responses to humanitarian crises and in monitoring progress towards the Millennium Development Goals, yet survey ethics are rarely discussed.
A survey provides an opportunity for vulnerable people to speak of the obstacles they are encountering and coping with. Participation in a survey allows them to debrief,1 yet interviewers may ask questions that respondents are reluctant to answer or even think about. Sensitive questions can increase psychological stress and compromise the accuracy of responses.2 Talking about traumatic experiences can sometimes be positive, but often it is not.3 Participation in a survey also deprives respondents of economic opportunities, e.g. time for farming, trading or gathering water and firewood.
Because of the risk of emotional and economic opportunity costs, individuals should be given the choice of whether to participate in surveys. To ensure informed consent, questionnaires often include protocol phrases.4 Unfortunately obtaining consent by stating prescribed phrases becomes mechanical as survey fieldwork continues. Also, not all participation is willing.5 Some respondents believe that refusal to be interviewed could be interpreted as refusal of assistance. Thus, informed consent is not necessarily built on mutual cooperation between respondents as possible beneficiaries and the survey team as a possible assistance provider.
When the target area is an emergency "hot spot" that draws international attention, more agencies rush in to conduct surveys. Limited access to the specific areas (e.g. due to security situations) allows a smaller sampling frame to be made available for a larger number of researchers. These conditions increase the probability that the same communities and households will be selected.
For instance, 44 surveys were conducted in Darfur between February and September 2004.610 Of the 107 communities covered, 33 (31%) underwent two or more surveys, and two (2%) underwent five or more surveys. Even taking into account the volatile situation, which required frequent monitoring, arent five surveys in the same community over eight months too many?
There are also development project "hot spots"11 where it is not very unusual for two agencies to independently implement similar health projects and conduct similar surveys to gather baseline data, to monitor and to evaluate the projects. Under the circumstances, the same communities and households can be selected and repeatedly contacted.
A joint survey, one of the most feasible solutions to avoid duplicated efforts, not only relieves the respondents psychic burden and opportunity costs but also saves survey resources. Another advantage of conducting a joint survey is to allow agencies to agree on methods and interpretation of results. Even when two surveys are conducted in the same area to measure the same indicators during the same period, the results could be significantly different due to differences in survey methods.
For instance, WHO concluded that the crude mortality rate in Darfur exceeded the emergency threshold in its mortality survey in JuneAugust 2004,12 while the World Food Programme (WFP) maintained that it fell below the threshold.6 This was interpreted as contradictory and led to political debate.13 Epidemiologists should have reviewed the survey reports more carefully and guided stakeholders to the correct interpretation. The reports make clear that the WHO survey targeted only IDPs12 while the WFP survey targeted both IDPs and residents,6 so the different results are reasonable, not contradictory. Survey practitioners and epidemiologists should avoid creating unnecessary political debate that stems from survey results; their professional ethics should not be limited to conducting surveys but also extend to taking responsibility for the impacts of survey results.
According to the informed consent protocol phrases of the demographic and health survey questionnaire, 40 (minimum) to 75 (maximum) minutes were allocated for completing an interview with a household with one child under five years of age and one reproductive-age woman, by asking 583 questions.4 The mean of the designated time per question ranges from 4.1 (40 x 60/583) to 7.7 seconds (75 x 60/583). It is highly doubtful whether the entire process for each question (i.e. asking a question, receiving an answer, verifying and recording it) is completed correctly in such a short time. It must have taken a much longer time to complete an interview. Other surveys are likely to share similar flaws.14,15
Two ethical issues are raised in this situation. First, there are too many questions to complete the interview within the allotted time. Questions without a clear analysis plan should be dropped, particularly when a survey has key questions and hypotheses. Second, it is a violation of informed consent to interview a respondent for a much longer time than projected.
Surveys have been conducted in a practitioner-driven manner. Survey ethics need to be reconsidered more from the respondents viewpoint. The frequency and content of surveys must be reasonable and justified, or they will lack accurate data and, ultimately, significance. 
References
Available at: http://www.who.int/ bulletin/volumes/ 85/11/07-047381/en/index.html
References
1. Dyregrov K, Dregrov A, Raundalen M. Refugee families experience of research participation. J Trauma Stress 2000; 13: 413-26.
2. Adey LA. Designing and conducting health surveys: 2nd ed. San Francisco: John Wiley & Sons; 1996: 182-4.
3. Chaitin J. "I wish he hadnt told me that": methodological and ethical issues in social trauma and conflict research. Qual Health Res 2003; 13: 1145-54.
4. ORC Macro. Model questionnaire with commentary: measure demographic and health survey basic documentation number 2. Calverton: ORC Macro; 2006.
5. Leaning J. Ethics of research in refugee populations. Lancet 2001; 357: 1432-3.
6. Emergency food security and nutrition assessment in Darfur, Sudan, 2004. Rome: World Food Programme; 2004.
7. Depoortere E, Checchi F, Broillet F, Gerstl S, Minetti A, Gayraud O. Violence and mortality in West Darfur, Sudan (2003-04): epidemiological evidence from four surveys. Lancet. 2004; 364: 1315-20.
8. CEDAT brief Darfur, Sudan: analysis of mortality and nutrition surveys from North and West Darfur (Update 2). Brussels: Centre for Research on Epidemiology of Disasters (CEDAT); 2004.
9. Mortality survey among the internally displaced persons and other affected populations in Greater Darfur, Sudan, September 2004. Geneva: WHO; 2004.
10. Persecution, intimidation and failure of assistance in Darfur. Amsterdam: MSF-Holland; 2004.
11. Van de Walle N. Overcoming stagnation in aid-development countries. Washington: Center for Global Development; 2005.
12. European Programme for Intervention Epidemiology Training (EPIET). Retrospective mortality survey among the internally displaced population, Greater Darfur, Sudan, August 2004. Geneva: WHO; 2004.
13. Thieren M. Health information systems in humanitarian emergencies. Bull World Health Organ 2005; 83: 584-9.
14. MICS questionnaire. New York, UNICEF; 2005.
15. World health survey: B-individual questionnaire (Rotation A). Geneva: WHO; 2002.
1 Correspondence to Hirotsugu Aiga (e-mail: aiga.hirotsugu@jica.go.jp)