Universität zu Köln

Pradel, Franziska ORCID: 0000-0003-4559-0772 (2022). Biased Political Information in Search Engines and Their Effects. PhD thesis, Universität zu Köln.

Preview

PDF Pradel_Biased_Political_Information_In_Search_Engines_And_Their_Effects.pdf - Accepted Version Download (4MB) | Preview

Abstract

Since the rise of digitalization, search engines and online platforms have been increasingly used to search for political information. To find out more about a political topic or political actors immediately, the first place to go for many people today is to go online to Google Search. Research showed that algorithms, including algorithms in search engines, are prone to biases that can impact users' attitudes. Still, political information in search engines and inherent biases have received low attention in empirical political science research. We do not know yet whether the information about political actors displayed in search engines may be systematically different based on specific characteristics like gender. We also do not understand how biased political information in search engines that are also widely trusted may affect individuals and their attitudes. For this reason, I systematically analyzed information for German political actors to see if there are such biases in search engines. I conducted an online experiment to investigate the potential effects of biased information in the context of a salient political topic, i.e., refugees. Another comparative political study goes beyond Germany and analyzes latent interest in European politicians in online search in multiple European countries, which has not yet been possible with other publicly available search query data. I found that the information displayed in Google search suggestions for politicians and their Wikipedia articles, which most users visit after Google searches, can be biased by their party and gender identity. However, other social identities seem to be more crucial for the latent interest in political actors in the European political online search, especially the nationality of political actors as well as having a supranational political role. Moreover, a strong political group identity plays a crucial role in how politically biased information influences political attitudes and how individuals engage with it online. Particularly, negatively biased political information about refugees in search engines has a polarizing effect on asylum attitudes among individuals at the left and right margins of the political spectrum who also show a confirmation bias in intended click behavior. Finally, I argue that media literacy in education and more research are needed to identify and address new challenges to democracy due to the increasing importance of online search.

Item Type:	Thesis (PhD thesis)
Translated title:	Title Language Biased Political Information in Search Engines and Their Effects UNSPECIFIED
Translated abstract:	Abstract Language UNSPECIFIED English
Creators:	Creators Email ORCID ORCID Put Code Pradel, Franziska UNSPECIFIED https://orcid.org/0000-0003-4559-0772 UNSPECIFIED
URN:	urn:nbn:de:hbz:38-553052
Date:	28 January 2022
Language:	English
Faculty:	Faculty of Management, Economy and Social Sciences
Divisions:	Faculty of Management, Economics and Social Sciences > Social Sciences > Political Science > Cologne Center for Comparative Politics
Subjects:	Social sciences Political science
Uncontrolled Keywords:	Keywords Language search engines English biases English political information seeking English gender bias English algorithms English digital society English hate speech English digitalization English Google English Wikipedia English search predictions English Google autocomplete predictions English social identities English European politicians English German politicians English refugees English click behaviour English selective exposure English online platforms English confirmation bias UNSPECIFIED
Date of oral exam:	28 January 2022
Referee:	Name Academic Title Proksch, Sven-Oliver Prof. Dr. Rohlfing, Ingo Prof. Ph.D
Refereed:	Yes
URI:	http://kups.ub.uni-koeln.de/id/eprint/55305