Environmental Change and Conflict in Kenya: Perceived and actual Drought Effects on Sense of Personal Insecurity and Violent Beliefs

Wednesday, 17 December 2014: 9:25 AM
Andrew Martin Linke1, John O'Loughlin2, Frank Witmer1, J. Terrence McCabe3 and Jaroslav Tir4, (1)University of Colorado at Boulder, Institute of Behavioral Science, Boulder, CO, United States, (2)University of Colorado at Boulder, Geography, Boulder, CO, United States, (3)University of Colorado at Boulder, Anthropology, Boulder, CO, United States, (4)University of Colorado at Boulder, Political Science, Boulder, CO, United States
We address two main questions of interest to scholars and practitioners interested in climate change effects. First, do droughts and their associated environmental impacts affect rates of violent conflict? Second, what differences exist for individuals’ personal insecurity between the effects of perceived changes in drought patterns and the actual observed fluctuations in rainfall? Using data from a mid-2014 nationwide survey of 1400 Kenyans that asked respondents nuanced questions about the frequency, severity, and timing of rainfall, droughts, and flooding events, we significantly change the nature of existing studies in the climate change and conflict literature. This survey also asked respondents about their personal exposure to violence, as well as conflict in their general area. Our interdisciplinary team employed cutting edge field experimental questions to elicit truthful responses about the willingness of respondents to support violent activities. Our statistical analysis investigates how the self-reported exposure to violence and conflict-supporting attitudes vary with perceived changes in rainfall. We systematically control for other explanations of violence support in all of our modeling. To answer our second question, we match survey respondents’ locations to observed remote sensing and station-location records of precipitation, temperature, and vegetation health. We present differences between the perceived and actual precipitation trends with implications for the respondents’ self-reported levels of violence exposure and explain these differences using other respondent and contextual characteristics.