Inequities in Online Information Seeking for Making Policy Judgments

Aims and central research questions

The project investigates how individuals seek information online to acquire knowledge and make judgments about policy proposals. We evaluate information-seeking behaviour against the background of information processing models to assert how information affects political knowledge and judgment. To that end, we encouraged participants to engage in policy-related online searches while tracking their online activity. The main objectives are to demonstrate whether such an intervention affects knowledge and attitudes toward the policy topics. We will explore associations between individual differences (demographics, psychological motivations, and cognitive abilities) and information-seeking engagement, knowledge gain, and policy attitude change (Figure 1). This is theoretically and normatively important as it speaks to the question of whether the propensity to seek and process political information effectively is chronically limited to a small segment of sophisticates or whether even those who usually do not take much interest in political matters could effectively seek and process factual information if motivated to do so.

Figure 1. Inequality in the loop. Inequalities may affect the level of engagement with information seeking and information processing, which impact judgements and knowledge.

Background

From the perspective of political inequality research, the different strategies people adopt to seek information for making judgments are relevant in at least two related respects. First, knowledge of political facts is socially stratified (e.g., Converse 2000; Delli Carpini and Keeter 1996; Rasmussen 2016), and the same appears to hold for cognitive abilities and psychological motivations such as the “need for cognition” and the “need to evaluate” (e.g., David 2009; Redlawsk 2004). Second, uninformed political reasoning may entail choices against one’s interest and may thus have adverse consequences for political representation through elections (e.g., Achen and Bartels 2017; Lau and Redlawsk 2001).

Methods

Our research follows a three-pronged approach that combines (i) unobtrusive measurement of individuals’ information consumption and seeking behaviour on the Web (including web pages visited and chat logs of interactions with an LLM-powered search engine), (ii) surveys that measure participants' social characteristics, psychological attributes, political predispositions, and knowledge and opinions on policy issues, and (iii) experimental tasks that incentivise participants to search information online to arrive at judgements on specific policy proposals. More specifically, we designed and implemented an initial survey to measure individuals’ political predispositions, knowledge, opinions and preferences. Then we conducted two online experiments with intervention tasks (i.e., control, general online search, online search + financial incentives, LLM-powered search) to assess how online information-seeking affects judgments about policy proposals. Figure 2 illustrates the components of our methodology using the first experiment, exemplary using the policy topic “Kindergrundsicherung”:

Figure 2. Example of experimental design for the topic Kindergrundsicherung (wave 1, 1st Experiment). An initial survey (purple) assessed the attitudes towards the policy topics and other pertinent aspects (including demographics, internet usage, digital literacy, and political leanings). Four months after, participants were invited to the first wave of the experiment about the policy topic “Kindergrundsicherung”; it included the interventions (e.g., 20h online search, in blue), pre and post-attitudinal measures (white), an initial introductory survey (pink) and a final knowledge test (green). Participants' browsing history is collected across the entire process (grey, bottom).

Publications

• Planned

Kacperski, C., Ulloa, R., Bonnay, D., Kulshrestha, J., Selb, P., & Spitz, A. (2024). Who are the users of ChatGPT? Implications for the digital divide from web tracking data (arXiv:2309.02142). arXiv. https://doi.org/10.48550/arXiv.2309.02142

• Talks

Kacperski, C., Ulloa, R., Bonnay, D., Kulshrestha, J., Selb, P., & Spitz, A. (2024, March). Who are the users of ChatGPT? Implications for the digital divide from web tracking data. Conference of Experimental Psychologists 2024 (TEAP 24), University of Regensburg.

Ulloa, R., Kacperski, C., Kulshrestha, J., Spitz, A., Bonnay, D., & Selb, P. (2023, September). An experimental study of online information seeking on policy judgments. General Online Research Conference 2023 (GOR 23), University of Kassel.

Discipline(s)

Politics and Public Administration / Behavioural Science / Computer Science / Philosophy

Starting date

September 2022

Partners

Respondi AG