WP3 will test the second (political causal) hypothesis with the use of multivariate and multilevel regression analyses by establishing associations between welfare policy changes and grassroots political changes, controlling for structural factors.

WP3.1 will use cross-sectional regressions of social assistance on political variables based on the analysis of exiting high-quality survey datasets from seven emerging markets. These surveys will allow the project team to examine the associations between social assistance provision, political party support, race/ethnicity/religion and various demographic control variables.

In order to avoid possible compatibility problems among surveys, the project will run a multivariate regression analysis for each individual country.

WP3.1 will examine whether social assistance benefits are granted on ethnic, religious, racial or political grounds.

WP3.2 will conduct multilevel modelling and cross-sectional regressions of social assistance on protest event data, pooled over time and location.

WP3.2 will test the hypothesis that temporal and spatial changes in social assistance programs are correlated with changes in the level of protest by the poor, controlling for structural factors.

WP3.2 will use a multilevel model cross-sectional regression analysis—a technique used in comparative politics literature (Beck and Katz, 1995; Plumper et al., 2005), to establish the historically and spatially situated causal relationship between political factors and welfare provision in emerging markets.

WP3.2 will test two hypotheses that indicate that governments expand social assistance not when/where people become poor, but when/where the poor become radicalized.