One house, two doors: Making sure survey numbers represent the people
published 10 August 2015
published 10 August 2015
What a compelling fact! We need facts like this. They help us to understand the health needs of a community and how we should be addressing them. But where do such facts come from, and how do we know if they reflect reality?
In June I joined the data collectors contracted by the IDEAS project’s Nigerian research partner, Data Research and Mapping Consult Ltd, in Gombe state as they carried out the IDEAS follow up survey to our 2012 baseline survey.
The realities of data collection made me realise there is still plenty of room for improvement so that the numbers truly reflect reality. Just as a few years ago it was common to pay people to take part in a survey or not ask for written constant, today we face new challenges from ghost households to bending the truth about money and possessions…
Politicians distributing money house-to-house to canvas for votes and malaria programmes giving out free insecticide treated bed nets have resulted in some Nigerians installing more than one door to their house. Two doors means being counted twice and therefore twice the amount of handouts! During the IDEAS survey household listing, each door was counted as a single, new household. It was only when interviewers finished with one household, left, and went through the second door that they discovered they were in the house they’d just left!
During the IDEAS survey, some husbands refused to leave their wives alone with the interviewer, especially when the interviewer was a man. In this situation, when a male interviewer asked a woman how much she spent on her last health facility visit, the woman would often stutter or dodge the question. In one instance, the interviewer asked the husband for water for ablution (ceremonial washing) to prepare for afternoon prayer. As soon as the husband left the room the woman quickly told the interviewer how much she spent on her last health facility visit and then promptly chastised him for trying to break up her marriage! Missing or inaccurate data on health service costs to families may result in conclusions that don’t reflect reality. Having only female data collectors could help to solve this challenge because husbands would be more likely to leave their wives to answer questions alone, who in turn would be more likely to respond truthfully.
Some interviewees were not keen on disclosing how many assets they have (e.g. number of pigs, owning a radio). Some thought if their assets are recorded they will miss out on future handouts, and others felt uncomfortable bragging about their possessions. Without accurate data on assets we cannot assess whether how wealthy or poor a family is affects whether they get health services or not, i.e. how equitable the health services are.
One village had people from two major tribes – tribe A and tribe B – dispersed over a piece of land but within eye sight of each other. The village head belonged to tribe A. The households to be interviewed for our survey were chosen at random and tribe B households fell outside the cluster to be surveyed. After the interviewers had finished in the village members of tribe B confronted the interviewers to ask: why had they been excluded? They started accusing the village head of partiality towards his tribe (they believed the interviews will lead to benefits for tribe A) and threatened to burn down the village head’s house. The interviewers had to quickly explain what the survey was all about and provide a credible explanation of how the households were selected.
Being aware of and considering issues such as these when planning surveys will help us to improve data collection in the future so that the numbers truly represent the people.