Research design challenges in investigating climate-conflict linkage include difficulties in specifying direct causal relationships and measuring climate exposures or events. Even with the proposition of indirect causal pathways, numerous intervening variables often obscure the mechanisms underpinning observed interactions, rendering conclusions from research tentative at best. Scholars often apply aggregate-level variables to measure individual-level experiences of climate and conflict, leading to non-equivalence in the units of analysis used in theorization and statistical analysis. Researchers have addressed these challenges by using proxies, particularly for measuring climate effects.
Mixed-methods approaches have been applied by merging event data with innovative statistical techniques. However, secondary research design challenges persist, such as the under-theorization of the relationship between impacted populations, perpetrators of conflict, and climate. The non-linkage of event data used for hypothesis generation to the causal graph analyzed statistically remains a significant gap. Poor specification of causal pathways derived from event data often results in marginal explanations for mechanisms underpinning indirect causal loops between climate and conflict. This paper begins with a discussion on research design challenges in the climate-conflict relationship, detailing researchers' attempts to address these challenges and lingering methodological issues, and concludes with recommendations for resolving these research design problems.
Firstly, there is no consensus on the pathways by which climate influences conflict. Challenges in specifying direct and indirect mechanisms for the climate-conflict relationship have led to different measurements of climate and conflict. For example, in studying the influence of climatic stress on the onset of political violence in the early stages of the Syrian 2011 crisis, Ash and Obradovich focused on long-term climate stress rather than specific climate events like flooding or drought^1. Similarly, Von Uexkull et al. examined how drought during growing seasons impacts the likelihood of ethnic civil conflict in Asia and Africa, focusing on extreme climate events such as drought^2. Despite similarities in the climate events investigated, variations in the experiences of climate exposure by the study population are rarely conceptualized. Differentiating between climate risk, stress, impacts, and disasters would support better operationalization of climate exposures and their interaction with conflict.
A similar conceptual slippage occurs in the operationalization of conflict. Scholars theorize different forms of conflict, such as political violence, political mobilization and protests, and market-related conflict, without differentiating the variations in climate exposures^3. Different forms of conflict, such as protests, riots, rebellion, and insurgency, interact differently with climate due to variations in perpetrators of violence and target populations. Delineating differences in causal mechanisms underpinning variations in climate exposures and different forms of conflict may provide better direct causal pathways for understanding climate-conflict interaction.
Conceptual slippages in operationalizing climate and conflict also translate to measurements in research design. For example, Ash and Obradovich used historical precipitation data from the Climate Research Unit to measure meteorological drought, but Raleigh contends that meteorological drought is a poor measure due to its decadal occurrence, suggesting dry spells as a more accurate measure^4. Consequently, debates on accurate measurements of climate exposures persist. Data challenges also arise, such as Ash and Obradovich's use of the Thin Plate Spline (TPS) technique to approximate climate stress across geographical gridlines^5. However, TPS could overstate climate stress effects due to difficulties in measuring confidence intervals for the geographic spread of climate stress, particularly in two- or three-dimensional plots^6.
Despite these challenges, innovations such as proxies, machine learning datasets, and event data have been employed to address research design gaps^7. For example, Ash and Obradovich used nighttime light intensity as a proxy for migration intensity to investigate climate's influence on migration patterns in Syria^8. This methodological innovation supported the measurement of population density and response strategies to climate events. Similarly, Von Uexkull et al. used geo-referenced and remote-sensing data to measure the effect of growing season drought on ethnic communities^9. However, the UCPD data used to measure conflict onset and occurrence was inappropriate due to the lack of theoretical differentiation between offensive and defensive weapons in ethnic conflicts^10.
Despite the value of methodological innovations, challenges remain, such as treating populations impacted by climate and conflict as a black box and marginal explanations of mechanisms underpinning indirect causal pathways between climate and conflict. Good practice in mixed-method research includes integrating conceptual, measurement, and theoretical insights from both methodological traditions. Richer theorization on the relationship between different climate exposures and variations in civil conflict, as well as differentiating between populations impacted by climate change and non-state perpetrators of conflict, is needed. Disaggregating the effects of these actors and variables will provide richer insights and more accurate operationalization and measurement of climate and conflict concepts. Methodological innovations, supported by robust theories, could delineate and explain mechanisms undergirding observed causal processes.
References
1. Ash, K., & Obradovitch, N. (2019). Climatic Stress, Internal Migration and Syrian Crisis Onset. Journal of Conflict Resolution.
2. Von Uexkull, N., Mihal, C., Fjelde, H., & Balvard, B. (2016). Civil Conflict Sensitivity to Growing-Season Drought. Proceedings of the National Academy of Sciences, 113(44):12391-12396.
3. Raleigh, C., Hyun, J., & Kniveton, D. (2015). Devil is in the Details: An Investigation of the Relationships between Conflict, Food Price, and Climate Across Africa. Global Environmental Change, 32:187-199.
4. Ash, K., & Obradovitch, N. (2019). Climatic Stress, Internal Migration and Syrian Crisis Onset. Journal of Conflict Resolution.
5. Ibid.
6. Columbia University. (2020, April 13). Thin Plate Spline Regression - Population Health Methods. From Columbia University Mailman School of Public Health: https://www.mailman.columbia.edu/research/population-health-methods/thin-plate-spline-regression
7. Ash, K., & Obradovitch, N. (2019). Climatic Stress, Internal Migration and Syrian Crisis Onset. Journal of Conflict Resolution; Raleigh, C., Hyun, J., & Kniveton, D. (2015). Devil is in the Details: An Investigation of the Relationships between Conflict, Food Price, and Climate Across Africa. Global Environmental Change, 32:187-199; Von Uexkull, N., Mihal, C., Fjelde, H., & Balvard, B. (2016). Civil Conflict Sensitivity to Growing-Season Drought. Proceedings of the National Academy of Sciences, 113(44):12391-12396.
8. Ash, K., & Obradovitch, N. (2019). Climatic Stress, Internal Migration and Syrian Crisis Onset. Journal of Conflict Resolution.
9. Von Uexkull, N., Mihal, C., Fjelde, H., & Balvard, B. (2016). Civil Conflict Sensitivity to Growing-Season Drought. Proceedings of the National Academy of Sciences, 113(44):12391-12396.
10. Posen, B. (1993). The Security Dilemma and Ethnic Conflict. Survival, 35(1):27-47.
Comments