particularly when governments lack the capacity or will to collect the data or when their estimates are unreliable. Different sources gather different data for different purposes, with different methodologies and for different objectives. These include operational planning, which is influenced by considerations of timely funding. Divergent objectives often affect the way in which data gatherers estimate target populations or beneficiaries. We are aware that some sources may also have an interest in manipulating or tweaking the number of IDPs. They may choose to do so in order to call international attention to a crisis, maximise the amount of external assistance received or downplay the scale of a conflict or disaster if the government is held accountable. In order to mitigate this potential bias, whenever possible we triangulate the data by using several sources and prioritising those we have historically deemed to have been most objective. Particularly for displacement associated with disasters, we monitor the different stages of the humanitarian response cycle, from the emergency to the reconstruction and recovery phase, by identifying the different organisations and indicators that report on displacement over the time. Language bias also affects our ability to source displacement data comprehensively. We can only obtain and analyse information in the languages we speak and read. Our staff and partners speak most of the required languages, but we inevitably fail to capture some information, particularly for parts of Asia. Disaggregated data We seek to obtain not only quantitative data from our sources on possible increases and decreases in the number of IDPs, but also more specific information such as data disaggregated by sex and age (SADD). This is vital in guiding an appropriate and effective response to IDPs’ protection and assistance needs. Relatively little SADD is available for displacement associated with either conflict or disasters. This is mainly because information on IDPs’ sex, age and disabilities tends only to be captured in organised settings such as relief camps, while in many cases a significant majority of IDPs live in dispersed settings among host families and communities. We also aim to gather and report disaggregated information by geographical area and time period in order to paint the most comprehensive and dynamic picture of displacement possible and provide a sound basis for more complex research and analysis. Even when disaggregated data is available, however, it tends not to represent a statistically significant portion of the overall data collected. More is vital if we are to accurately inform the identification of, and respond to the specific needs of different groups of IDPs. Methodological challenges particular to displacement associated with conflict We gather data from primary and secondary sources on the number of people displaced by international and non-international armed conflict and other situations of violence. We aim to include all people forcibly displaced in such contexts. Sources tend to be numerous during humanitarian crises and visible emergencies, when they compile information to target assistance, as in Syria. During protracted and neglected crises, displacement data tends to be unavailable or outof-date, as in Armenia, Cyprus, Georgia, Togo and Turkey. Sources often do not use the same definition of an IDP as the Guiding Principles. Nor do they use the same methodologies, which creates a serious challenge when compiling our estimates. In several countries, including Afghanistan, Bosnia and Herzegovina, DRC, Georgia, Pakistan and Ukraine, only IDPs who have been officially registered with the authorities are counted. In some countries only one data source is available, while in others there may be several. For each country listed in the 2017 GRID dataset, we systematically looked for several sources. We always strive to identify new data sources, even for countries and situations where others already exist. This enables us to crosscheck, but it may also create confusion because sources rarely explain their methodologies. When different sources are available, or when a new source provides information, we may still decide to base our estimate on only one METHODOLOGICAL ANNEX 103