While a considerable part of the gender & politics literature applies quantitative research approaches, statistical techniques are mostly received very critically in feminist methods debates. This can partly be traced to feminist methods being focused on diversity, a concept that is at the core of the notion of gender and of intersectionality, whereas inferential statistics are focused on drawing generalizing conclusions.
In this paper, I will unpack the concepts of gender and intersectionality and show that two dimensions of diversity that are fundamental to them, can be directly translated to the statistically vocabulary: within-group variation and cross-level interactions.
Next, I will show that particularly the development of multilevel models is crucial for feminizing statistics. This technique allows for systematic tests of how contextual characteristics, such as gender norms, institutionalized patriarchy, welfare state arrangements and more, shape male-female differences, or in slightly more complex models how these contextual characteristics shape the degree to which other relationships are gendered, such as the impact of religion on democratic support.
Finally, I will argue that these statistical developments should also challenge or stimulate gender & politics scholars to ‘statitize feminism’. Claims of context-specificity should not be a priori accepted. They can and should be developed in systematic gender theories that explain which contextual characteristic or arrangement is responsible for one situation not being comparable to the other. This way we can put our ideas and theories to the test.
The theoretical and methodological discussion will be illustrated with examples from the literature as well as brief empirical analyses on women, Islam and democratization in the Arab Middle East.
