Access to the Swedish welfare state’s last safety net, social assistance, is ultimately determined through discretionary decision-making by social workers. This dissertation examines intersectional patterns and discriminatory bias in social workers’ assessments about social assistance eligibility. Focusing on factors related to applicants’ gender, family and ethnicity, the project comprises four studies, all of which highlight patterns regarding which applicants assessed as being eligible for support. Altogether, the project contributes to an expanded understanding of discriminatory tendencies in how social assistance policies are given practical meaning by the professionals that bring them into force.

The first study builds on data covering all social assistance eligibility decisions implemented in 25 municipalities during one calendar month in 2012 (n=472). The remaining three studies build on data from a vignette experiment conducted in 2018, in which just over 1,000 social workers from 19 municipalities, including Sweden’s three largest cities, participated. 

Results from both sources of data confirm the impression left by previous research that social assistance assessments are gendered. They show that the likelihood of granting assistance is determined through different standards for men and women. In the view of current knowledge gaps, an important contribution lies in bringing the issue of ethnicity bias to light. The results from the vignette experiment indicate that applicants with Arabic-sounding names are responded to with more conditionality than applicants with Swedish-sounding names, and that discriminatory biases related ethnicity are highly intertwined with gender biases.

By raising much-needed questions about the assessment of couples, the project also draws attention to the dissonance between the Swedish welfare state’s gender equality regime and the conditions for accessing social assistance. The results indicate that moral judgments about applicants’ gendered family roles affect social workers’ propensity to grant support to couples, and that such judgments take form through ethnicity bias. 

In terms of theory, the dissertation draws upon feminist and postcolonial perspectives on social policy as well as a street-level bureaucracy perspective on frontline work. Social assistance is understood as part of the welfare state’s wider politics of redistribution, and the quantitative patterns formed by social workers’ individual acts are seen in the light of structural inequalities. The dissertation presents a conceptual model for thinking about social assistance eligibility, emphasising uncertainty as an inescapable dimension of means-testing. A central argument is that eligibility issues decided at the street level cannot be separated from ongoing discretionary processes of policy implementation. While the risk of discrimination in social assistance assessments is inevitable, it tends to be concealed by the administrative arrangements through which policy comes to matter.