Empirical verification of weak evidence effect using crowdsourcing

This thesis is focused on cognitive bias called “weak evidence effect” by proposing an empirical experiment evaluated through crowdsourcing platform. This cognitive bias states that when presented with an argument that should strengthen one’s opinion – it does the opposite if the argument is weak. First part of this thesis builds on existing experiments from various domains and observes if the bias effects decision making while reading news articles. The proposed experiment is designed in a way that people are judging the im-provement of relations between two imaginary religions based on a news article. The article is then enhanced by a weak argument and the difference in judgment is observed. The results show that cognitive bias in fact plays a role in the decision-process while reading a news article but the determining variables remain unknown. Second part is focused on crowdsourcing tools and on a detailed description of Crowdflower platform. And finally, a design and implementation of middleware API on top of Crowdflower API is described. The middleware API is then used to pay out bonuses to respondents for high-quality an-swers.