Bringing the Real World Back In? Strategies to Enhance the External Validity of Experimental Research Designs on the Basis of Two Online Survey Experiments on Stereotype Activation in Flanders (Belgium)

Numerous experimental studies point to a number of stereotypical patterns in which female candidates are more likely to be perceived as competent in communal issues linked to the traditional domain of the family, such as education, health care and helping the poor, whereas men would do a better job with agentic issues, such as military spending, foreign trade, agriculture and taxes (e.g. Huddy & Terkildsen, 1993b; Lawless, 2004; Matland, 1994). Besides these issue competence stereotypes, some studies also found considerable evidence for the existence of ideological position stereotypes, i.e. the idea that female candidates are more leftist compared to male candidates (e.g. Dolan, 2014a; Koch, 2002). Previous work makes excellent use of experimental methods. Often in real-world elections, there is too much noise to be able to isolate causal effects. Therefore, most experimental studies are set up in low-information environments, in which respondents are only given a limited amount of information about the presented candidates and then asked to evaluate each candidate. However, Dolan (2014) rightly suggests that stereotypes do not act in a vacuum: they interact with other prejudices such as partisan stereotypes and/or contextual elements. It can therefore be argued that the kind of settings applied in most experimental research misses much of the nuance voters face in real-world elections as they encounter a range of information throughout the course of an election campaign. This paper will focus on strategies to maximize the potential of experimental research methods. Two online survey experiments, focusing on the impact of the information environment in which candidates come forward on the process of stereotype activation in voters’ minds, will be outlined. The experimental designs present candidates with outspoken leftist or rightist policy positions (in contrast to the mostly centrist positions used in previous experiments) and fictional news reports depicting the nature of the dominant policy issues in a given context (in contrast to previous experimental work in which contextual information is mostly held constant). Consequently, these experiments no longer take place in low-information contexts and, by bringing the real world back in, a more dynamic experimental setting is created. It will therefore be argued that these innovative designs are a useful tool to enhance the external validity of experimental studies, which comes to the heart of what this section on contemporary challenges in political methodology aims to do.