Analytic hierarchy process and sensitivity analysis implementation for social vulnerability assessment: A case study from Brazil

Abstract

One major challenge of social impact assessment concerns the implementation of multicriteria decision analysis (MCDA) to ascertain the vulnerability of households to environmental change. While MCDA has been widely used to combine vulnerability indicators into an aggregated vulnerability score, the sensitivity of vulnerability indices to uncertain appraisals and judgements of the magnitudes and weights of indicators has been largely ignored so far. In this work, based on vulnerability indicators previously selected and ranked using the analytic hierarchy process (AHP) technique, for household Brazil surveys carried out in 1998 and 2012, a sensitivity analysis (SA) was implemented to account for the variation of vulnerability indicators over time and space. In particular, two techniques were applied: the indicator removal and the threshold value tests. The indicator removal test involved setting to zero a particular indicator weight and rescaling the remaining indicator weights linearly. The threshold value test aimed to identify which indicators had the most relative influence on both indices. Finally, the critical threshold value showed the most important vulnerability indicators and allowed to summarise and contrast the standardized scores differences of the indicators between the two surveys. The results showed which indicators were the most important in increasing or decreasing the vulnerability and improved the understanding of how the overall vulnerability of rainfed farming households changed through time as a function of changes in sensitivity and adaptive capacity.