‘Early warning analysis is a messy business and, inevitably, the results will be imperfect and understandably subject to questioning’ (Clarke 2005, 82)[1]. Matveeva (2006, 13)[2] suggests that early warning systems can be distinguished by whether they adhere mainly to a quantitative or a qualitative methodology. A study by Barton et al (2008, 2)[3] of 30 conflict prediction models points to a wide range of conflict early warning methodologies, from qualitative assessment to complex regression and systems dynamics analysis.

Early warning methods developed around elections analyse both qualitative and quantitative data. Decisions on which methodology to adopt might be influenced by the availability of early warning tools and the capacity of implementing organisation to utilise them effectively. Operational considerations include conceptualisation of observable indicators and sources, as well as data collection, analysis and presentation methods. This is also applicable to electoral violence early warning efforts.

[1] Clarke, J.N. 2005, ‘Early Warning Analysis for Humanitarian Preparedness and Conflict Prevention’. Civil Wars 7, no. 1: 71–97.

[2] Matveeva, A. 2006. ‘Early Warning and Early Response: Conceptual and Empirical Dilemmas – The Global Partnership for the Prevention of Armed Conflict’, Issue Paper 1. European Centre for Conflict Prevention/International Secretariat of the Global Partnership for the Prevention of Armed Conflict, The Hague.

[3] Barton, F., and K. von Hippel, with S. Squeira and M. Irvine. 2008. ‘Early Warning? A Review of Conflict Prediction Models and Systems’. PCR Project Special Briefing, Centre for Strategic and International Studies, Washington, DC.