It is not surprising to anyone that population-environment dynamics research involves using data on population and environment. When starting such research, one of the first things you should be looking for is census data, right? That’s what I did, and I found it at the directorate of Statistics in Duhok city, in the shape of a pile of papers with Arabic text and (Arabic) numbers. Since I was interested in migration (which requires population data from at least two different times), didn’t put much effort into digitizing the data, but I instead used my 1200 households survey data.
To get to the problem: There is no reliable population data in Iraqi Kurdistan. But, you might ask, did you not acquire data from the directorate of Statistics in Duhok? Indeed I did, but I also was told that these data were estimations based on an older census, where they had simply multiplied the old population of each village and city, with the estimated population growth in the country (if that country is Iraq or Kurdistan Region, I’m not sure).
I will here explain the problems with census data in Iraqi Kurdistan, and suggest some alternative solutions that might be more reliable.
Extapolations, i.e. using a general population growth factor to estimate the population, makes the assumption that population is changing linearly with time. It hence assumes that the only factors determining the population size of a community are birth or death. This would not hold true even for isolated societies where no migration was possible, and definitely not for societies that have been shaped by conflicts (conflict is a common factor leading to population changes, e.g. death or flight) and other political calamities targeting certain ethnic and religious groups. In the Kurdistan Region as well as Iraq, wars, droughts, and economic changes have led to both forced and voluntary migration. People have been killed in conflicts and extermination campaigns, and the situation have probably also affected the birth rates. The poison gas attacks used during the Anfal genocide have affected peoples health, and most likely the birth rates and mortality. As you can see, population is not something that can be easily modeled. For another example, click here.
The politics of demography
So why do we even have extrapolations instead of regular census data from every five or ten year period?
Information about population size is really sensitive in Iraq since it affects the power allocated to different ethnic groups. During the late 1980’s a prohibited zone was declared in the Kurdish areas and anyone living there was relocated or killed. Participation in the 1987 census would therefore mean relocation to a camp, while people who “failed to participate” in the census were no longer considered Iraqi citizens and were therefore “destroyed” in the genocide. For more information about the connection between Al-Anfal and the 1987 census, see Hardi (2011).
Two decades before the Anfal Campaign, an “arabization” process had started in lands that were considered strategic to the government, for example the oil producing area of Kirkuk. The Kurdish population in this governorate were during this process diluted by Arab migrants who were given economic incentives to move from the south.
Since the 1987 census, no other complete census has been conducted in Iraq. There was a census in Iraq in 1997 but it excluded the Kurdish governorates, and the census that would be the first nationwide census since 1987, planned for 2007, has been delayed several times due to political tensions and has not produced any published results yet (Adnan, 2010; Hiltermann, 2010; Ryan & Kami, 2009).
Two keys to understanding political aspect of population are the Kirkuk status referendum and the Article 140 of the Iraqi Constitution from 2005. The Kirkuk status referendum is part of a popular vote where a decision will be made on whether some Kurdish majority regions of Iraq will become part of the KRG. Article 140, however, states that this referendum cannot be carried out until two measures have been taken, normalization and census. Normalization means assisting the return and/or compensating the people who were displaced from (and to) Kirkuk, and the census data would then be used to decide which areas can be considered Kurdish, i.e. having a mainly Kurdish population. The Kirkuk Status referendum was supposed to resolve the issue of Kirkuk by the end of 2007, but has since been delayed. It is important to mention that Kirkuk is not only coveted for heritage reasons, but perhaps mainly for its vast oil resources. Critics say that the article is not working, that compensation is happening to slowly, and that the compensation to Kurdish families is not equal to the compensation to Arab families. However, not only money is the obstacle for Kurdish families wanting to return to Kirkuk, as security is a big issue in the Kirkuk area since the fall of the Iraqi regime in 2003.
So now we have established that the latest census covering all governorates in Iraq was in 1987, during the Iran-Iraq war and in connection with the Anfal campaign, and therefore unreliable. Earlier censuses were also likely to have been politically motivated, where the Ba’ath party changed the shapes of governorates by aggregating districts differently, but it seems that the government of Iraq, to some extent, relies on the 1957 census for decision making.
In conclusion, census data is a strongly political issue that will determine the future of Kirkuk and thereby the future of Iraq. Even though data exist, its reliability needs to be questioned as extrapolations and modified governorate boundaries are not comparable to reality in such a complex region.
There are a few ways to overcome the lack of population data, but neither method is perfect. Since I could not find any reliable census data, I decided to conduct large-n samples instead to give me the information I needed. I was then able to obtain the exact information I wanted, but on the downside, I could not cover the whole population living in my study area. This approach is also time and money consuming, and might not be a viable option for all researchers. Satellite imagery can give some idea of where people live, especially the size of larger towns and cities. This does not give an exact measure of how many people are living there, and satellite data of sufficient resolution can be expensive to acquire. A third option that I recently came across was using the food distribution cards, which has information about all people receiving food aid from the government. How this works, and where to find that data and metadata, however, has yet to be explored. By triangulating these data, we might get a decent idea of how many people are living in the area.