Impacts and determinants of panel survey attrition: The case of Northern Uganda survey 2004-2008

The paper analyses the impact of household attrition in the Northern Uganda Survey panel of 2004 and 2008. These surveys were designed to evaluate the performance of the first phase of the Northern Uganda Social Action Fund (NUSAF). The first survey was conducted in 2004 when the region faced heightened levels or rebel insurgency and the subsequent survey in 2008 when rebel hostilities had ceased. As such, the panel survey was plagued by a high level of attrition—at least 25 percent of the households could not be resurveyed in 2008. The paper examines the impacts of attrition on determinants of household welfare as well as household experience of insecurity shocks. The pattern of attrition is not random with households in urban areas and those that were resident in internally displaced person camps (IDPs) were more likely to be lost during the follow-up survey. Furthermore, residence in West Nile and Acholi sub-regions were key determinants of household attrition. Within these sub-regions, households with younger heads were more likely to be lost in Acholi while households with teenage children are more likely to be lost in West Nile. Finally, the attrition tests confirm that the regression coefficients differ significantly between households resurveyed and lost during the resurvey.