Even within the UN and coordinated international humanitarian reporting mechanisms there is inconsistency in the way different populations are described and counted, with some estimates based on “people affected” and others on “people in need” or “people targeted”. Many terms and expressions are specific to internal displacement, and our database captures the most common ones (see table A.9). They may refer to individuals, groups of people such as families or households, or housing. We use the number of houses destroyed as a proxy because it shows that at least one household has been left homeless. We calculate the number of individuals by applying the AHHS available for each country. Housing information Housing information is important in estimating displacement associated with disasters. To produce our 2016 estimates, we analysed more than 300 reports that mentioned housing damage or destruction rather than the number of people displaced. In order to use housing data as a valid proxy, we only consider figures for homes that have been damaged to the extent they are no longer habitable. Terms that indicate the extent of damage include “houses at risk [of collapse]”, “houses severely affected/damaged” and “houses destroyed”. We consider housing to be any place where people have established a habitual residence. We include hospitals if the information provided suggests that long-term patients have been displaced. We also include shelters in refugee and displacement camps. “Collapsed tents” in Jordan’s Zaatari refugee camp, for example, are counted as uninhabitable housing. Such cases constitute multiple displacement, in which people have already fled once, only to become displaced again when their camp is flooded. Evacuation data We often use data on mandatory evacuations and people staying in official evacuation centres to estimate event-based displacement. This was the case for 8.4 million of the new displacements we reported on in 2016. On the one hand, the number of people counted in evacuation centres may underestimate the 108 GRID 2017 total number of evacuees, as others may take refuge elsewhere. On the other, the number of people ordered to evacuate may overstate the true number, given that some are likely not to heed the order. The potential for such discrepancies is much greater when authorities advise rather than order evacuation, and as a result we do not incorporate such figures into our estimates. Quality assurance and independent peer review As in previous years, and in order to improve our methodology, we submitted this year’s estimates to a quality assurance process to verify the data. The verification stage is as important as the data collection itself, because it allows possible discrepancies to be identified, and the data to be refined before it is finalised. This year’s process was mainly led in-house, and all of our entries have been double-checked, through rigorous analysis by experts previously not involved in the data collection and analysis for each of the events. Colleagues were assigned each country with displacement associated with conflict and disasters involving more than 500 people. They dug through all the data collected and collated by others, asking questions and highlighting potential gaps, and so ensuring the highest possible level of transparency and clarity. As an example of an entry having undergone changes following the internal review process, reports of displacements associated with violence in India were questioned, leading to a rigorous follow-up process with existing and new sources. This allowed us to solidify our data and present it with a much higher level of confidence in its accuracy and value. Our data on the huge volumes of historical displacement in Colombia also underwent intense scrutiny, including exchanges with OCHA, the government’s victims’ registry and NGOs. The review unearthed previously unknown information on the primary source’s methodology and data treatment processes, which led to significant changes (see spotlight, p.29). The quality assurance process for displacement associated with conflict was supported by external advice. We presented our figures and methodology to NRC country offices, IOM teams,