Much of the data we rely on is collected by UN agencies such as OCHA, UNHCR and IOM and their NGO partners. In some cases, it is collected by an institution or consortium mandated with that single primary function, such as DRC’s Commission on Population Movements, the Task Force on Population Movement in Yemen and IOM’s numerous DTM operations. where both institutions published displacement figures for the same region, these differences resulted widely disparate estimates, generating even more confusion and casting some doubts on the accuracy and reliability of both datasets. For example, for the month of July 2016 IOM estimated that there were 2,444 IDPs in Rumonge province compared to 13,095 reported by OCHA. Much of the time, however, data on IDPs is collected by institutions working under broader mandates, such as the UN’s humanitarian profiles, humanitarian needs overviews and humanitarian response plans. In these cases, the data is updated only a few times a year, and often lapses once the humanitarian phase of a crisis has ended, even if the displacement has not. Longitudinal data collection also ends, or is interrupted, based on government policies. This can involve who is counted as an IDP or where data collection is undertaken or permitted. These are common challenges when a government has adopted a policy that specifically aims to reduce the number of IDPs, as in Kenya, or when it wants to shift attention away from a particular crisis. As noted in part 1, our source of data on internal displacement associated with conflict in Colombia comes from the government’s registry of victims (Registro Único de Víctimas, RUV). The purpose of the RUV is to account for all victims of the conflict. This involves identifying people who are or were internally displaced, but it does not necessitate tracking them over time. Once someone has been registered they remain so, meaning there is little or no follow-up information with which to determine whether or not they are still displaced. As in last year’s report, we have included these “decaying” figures, which were last updated prior to 2015 (see figure 3.6). This year, these figures account for only 6.5 per cent of the global total. We publish information about the age of the data for two reasons. It allows readers to draw their own conclusions about the figures, and by depicting them in this way we hope to encourage anyone with more recent data to come forward, or to help follow up on the situations in question if no more recent data is available. Gathering time-series data systematically can be costly and sometimes a lack of funding means collection falls off before a crisis is resolved. When various crises compete for attention and resources, some inevitably lose out and become neglected, which translates into less funding and political will to stay on top of them. This can occur even when the number of IDPs is significant, as has been the case with Burundi, where more than 141,000 people were displaced at the end of 2016.277 For most of the year, IOM’s DTM covered only three of 18 provinces, excluding Bujumbura Mairie, one of the locations most affected by internal displacement in the country. In September, IOM’s DTM coverage expanded to seven provinces and in December to 11 (although still excluding Bujumbura). At the end of the year OCHA published its annual Humanitarian Needs Overview for Burundi which also included IDP estimates for Burundi. OCHA’s figures differed from IOM’s in that they covered all provinces in the country, were collected at different intervals and were based on a different estimation methodology that placed more emphasis on expert opinion. In the few cases 78 GRID 2017 Next year, we plan to remove the following figures from our global total unless we receive updated information (see table 3.2). For a more comprehensive and transparent assessment of our confidence in the data we have provided, please see the methodological annex at the end of this report, where the age of the data, its geographical reach and other factors are further discussed and evaluated.