35 Methods Conjoint experiments, also known as candidate experiments and vignette studies, have seen a considerable boom in political science research (Abramson et al., 2022; Costa, 2021; Hainmueller et al., 2014, 2015; Kirkland and Coppock, 2018; Leeper et al., 2020; Schwarz and Coppock, 2022; Sen, 2017; van Oosten et al., 2024a). Conjoint experiments are fully randomized experiments in which respondents assess profiles of political candidates. They provide an effective means of addressing research questions concerning minority politics due to several compelling reasons. Firstly, through the randomization of treatment and control groups, conjoint experiments reveal a causal relationship through eliminating the risks of spurious correlations, which is so often a pitfall in cross-sectional research. This is particularly pertinent in investigating ingroup favoritism, where untangling the role of identity versus policy is key. Furthermore, conjoint experiments embedded within surveys enable exploration of how group membership influences in-group voting. By incorporating questions about group identification before or after the experiment, researchers can gauge the impact of such affiliations on political choices, offering insights into the relationship between identity and preferences. The versatility of conjoint experiments in accommodating multiple group memberships simultaneously is another invaluable advantage, allowing to disentangle the interlocking dynamics of migration background, gender, religion, and policy. However, a recurring concern with experiments in the social sciences, particularly those in the study of minority politics, lies in the issue of sampling and the subsequent lack of diversity within the samples. This challenge threatens the external validity of findings and raises questions about the generalizability of observed trends to broader populations. A common trend in psychological and economic experiments has been the overreliance on student samples (Henrich et al., 2010), and unfortunately, political science is not exempt from this trend. Many survey experiments, including those within political science, frequently rely on convenience samples that lack significant ethnic diversity, casting doubt on the ability to draw meaningful conclusions about multi-ethnic societies. In contexts where ethnic groups are underrepresented, statistical power is compromised, and the capacity to identify significant relationships becomes restricted. The challenge of inadequate diversity within samples undermines the efficacy of conjoint experiments in studying minority politics. To overcome this limitation, I oversampled minority groups, which is challenging but indispensable (Celis and Mügge, 2018; Font and Méndez, 2013). Oversampling minority groups of interest not only allows for robust causal claims but also facilitates comparisons both within and between these groups. Addressing the sampling concern enriches the scope and relevance of conjoint experiments in understanding minority politics and contributes to the authenticity of the research findings. I chose to collaborate with Kantar Public
RkJQdWJsaXNoZXIy MjY0ODMw