The Center for Climate & Security

Home » climate and security » The Streetlight Effect in Climate-Conflict Research on Africa

The Streetlight Effect in Climate-Conflict Research on Africa

somaliaBy Cullen Hendrix, Senior Research Advisor, The Center for Climate and Security

Climate change research on Africa has a streetlight problem: researchers tend to invest more attention on former British colonies and countries with relatively open, stable political systems than other countries, with these factors emerging as more important than objective indicators of “need” like physical exposure to climate change or adaptive capacity. That is, our research seems less guided by objective need and more guided by convenience/safety concerns.

The logic is straightforward: natural and social scientists alike pick cases and field sites for a variety of reasons that have very little to do with objective need or scientific criterion: ease of travel, safety, predictability, familiarity with language, access to professional networks, data availability and an existing literature to which to respond. Given that the Brits kept the most comprehensive colonial records and English has become the lingua franca of scientific communication, all these factors bias case selection toward English-speaking, comparatively politically stable countries like Kenya and South Africa. As I demonstrate in a just published article at Global Environmental Change (abstract below), we know more about those cases than 29 other countries comprising nearly half the African land mass combined:

The streetlight effect is the tendency for researchers to focus on particular questions, cases and variables for reasons of convenience or data availability rather than broader relevance, policy import, or construct validity. To what extent does the streetlight effect condition the state of knowledge about climate change in Africa? Analysis of Google Scholar search results, both general and within leading climate change-related journals, reveals that two proxies for objective need, population and land mass, are associated with a higher volume of scholarly attention. Countries with greater exposure to the negative effects of climate change and countries with less adaptive capacity do not receive more scholarly attention. Rather, I find evidence that factors like British colonial history, strong civil liberties, and to a lesser extent political stability − factors not directly related to risks from climate change − affect scholarly attention. The streetlight effect is evident in climate change research on Africa.

This implies our understanding of climate impacts on natural and human systems – and adaptations thereto – in non-Anglophone, less politically open and stable countries is comparatively limited.

Is this problem evident in research on the climate-security nexus in Africa as well? Africa has been a locus of attention regarding the security implications of climate change, both due to the centrality of rural livelihoods there and the prevalence of conflict in recent decades. The US Department of Defense Minerva Initiative awarded a $7.6 million grant to study the subject (the author was a co-PI), and foundational contributions to the environmental security literature – in political science, economics, and geography – have focused on the continent.

At first cut, there seem to be pretty significant gaps in scholarly attention across cases. I used the search methodology described in the linked Global Environmental Change article to search the Journal of Peace Research (JPR) – a Europe-based conflict journal that has been prominent in the environmental security literature and on whose editorial committee I serve – for references to African countries and climate change. I then correlated those references with the Uppsala Armed Conflict Database data on conflict-related deaths since 1989. The Uppsala data include not just conventional armed conflicts but also non-state conflicts, such as violence between tribal militias, and one-sided violence against civilians.

The results are telling: while references in JPR correlate relatively highly (r = 0.47) with conflict deaths for the period 1989-2015 – countries with more violence receive more attention – there are some seeming outlier cases. Tanzania, a former British colony with a history of relative political stability and an absence of violence (61 battle deaths for the period) receives almost as much attention as Somalia, where the death toll from armed conflict has been almost 700 times higher.

Kenya is featured in 25 climate change and conflict-related articles, mostly discussing “range wars” between pastoral groups and recurrent bouts of ethnic rioting. However, the intensity of conflict there has paled in comparison to that in neighboring Ethiopia, which is more violent and more exposed to the physical effects of climate change but has received less scholarly attention than its comparatively peaceful neighbor.

This figure plots the general relationship between two measures of “need”: ND-GAIN’s physical exposure measure, which captures the magnitude of expected physical effects of climate change for a country (higher values correspond to more exposure) and conflict deaths, with the circles representing the number of scholarly references to the particular country in JPR – larger circles correspond to more mentions. If scholarly references tracked perfectly with physical exposure and past conflict prevalence, we would expect the larger circles to predominate in the upper right; instead, we see them peppered throughout.

Climate exposure conflict deaths

Climate Change Exposure, Conflict Deaths, and References in Journal of Peace Research for African Cases.

Using the base model in the Global Environmental Change article, I estimate former British colonies have received about 73% more scholarly interest than non-former colonies, even controlling for the magnitude and frequency of conflict therein. This difference is large, roughly proportional to the change in scholarly interest we would expect going from ~1,000 to ~60,000 conflict deaths during the period 1989-2015.

One major caveat is in order: Not all – or even most – conflict is linked to environmental factors. Most conflicts are driven by factors like exclusionary rule and repression of ethnic, religious or peripheral minorities. If conflict drivers are different across countries, then it is possible that environmentally-linked conflict is more prevalent in former British colonies. However, there’s no systematic research to support this claim.

These gaps matter for policymakers. Our evidence-based advice is only as good or comprehensive as the cases from which they are drawn. For instance, if the IPCC thinks community-level approaches to adaptation are crucial in areas of weak governance – the areas most prone to conflict – it probably matters that 75% of the evidence for this claim comes from former British colonies. If our knowledge comes disproportionately from less violent yet less accessible cases, like Kenya and Tanzania, how can we know that our proposed interventions will work in Central African Republic or Burundi? If we want to expand our knowledge in this area, more effort needs to be tasked to non-Anglophone African countries.

The next IPCC report will be released in 2022. Hopefully by then, funders, researchers and translators will have helped to close the gaps in our knowledge created by looking where the light is.

1 Comment

  1. Alex Randall says:

    Hi there, really interesting work. It would be interesting to see how number of country references relates to number of conflict deaths. It may be that scholars tend to focus on the places that have the most wars. There is also a strong connection between colonial history and conflict.(e.g see here So academics may be picking places because of the high number of conflict deaths – which skews their focus towards former colonies.

Leave a Reply

Featured Project

Follow Blog via Email

Enter your email address to follow this blog and receive notifications of new posts by email.

Follow us on Twitter

%d bloggers like this: