need to be able to identify the start of displacement if we are to gauge its duration, which is vital for measuring and addressing protracted situations. If the causes of displacement are to be addressed, it is also crucial to know when, where, how and why new, repeated and secondary displacements occur. Sometimes we receive information about the number of displacements that have occurred during the course of a particular year. This was the case for Algeria, Burkina Faso, El Salvador, Iraq, Syria and Turkey in 2016. In other cases, such as Mozambique, we inferred the new displacements from the simple fact that there were 15,000 IDPs accounted for in December 2016 against none in 2015. Without flow data about specific individuals or cases, we use the term “new displacements” to cover the following: || People being displaced for the first time from their home or place of habitual residence || People previously displaced who had returned or settled elsewhere being displaced again || IDPs being displaced from their place of temporary shelter or residence Given the way data is collected and reported, “new displacements” often mask secondary, tertiary and repeated displacement. In the absence of specific data on each inflow and outflow, we are forced to infer these processes from the contextual analysis of changes in stock data using a consistently applied set of decision rules. In order to avoid miscounting IDPs’ voluntary movements as incidents of displacement we take a conservative line, inferring new displacements associated with conflict when the total number of IDPs in a country increases from one point in time to another, and when the increase is not the result of a change in measurement or methodology (see figure 3.2). This is an imperfect approach which, depending on the specifics of a given situation, involves varying degrees of uncertainty. For example, based on additional contextual information we have obtained we have reason to believe our estimate of new displacements in CAR to be an underestimate. In the absence of credible quantitative data on the number of new displacements, however, we opt to err on the side of caution and indicate that there were “at least” 46,000 new displacements in 2016 instead of the 60,000 to 80,000 displacements that have been reported to us anecdotally. Difficulties detecting repeated, secondary and onward movements We also confront numerous gaps when it comes to covering all flows in and out of displacement. In only a few situations do we receive direct observational data and information about new, secondary or repeated displacement, returns and other processes. Sometimes we receive information about children born to IDPs, which is not the same as new displacement, and deaths in displacement. Figure 3.2: Internal displacement figures from the Central African Republic Source: Commission Mouvements de Population INSIDE THE GRID: Overcoming data shortfalls 73