Behavioral biases in information acquisition

• Main Author: Feiler, Lauren Elizabeth
• Format: Others
• Published: 2007
• Online Access: https://thesis.library.caltech.edu/2029/1/Feiler-FullDissertation.pdf; Feiler, Lauren Elizabeth (2007) Behavioral biases in information acquisition. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/FJ1V-HR48. https://resolver.caltech.edu/CaltechETD:etd-05242007-170633 <https://resolver.caltech.edu/CaltechETD:etd-05242007-170633>

This dissertation examines two related biases in information acquisition, information avoidance and selective exposure to information. The first essay focuses on the use of information avoidance to justify self-serving decisions. Participants in an experimental dictator game are given the chance to avoid costless information about a recipient's payoffs. Many dictators choose an allocation that maximizes their own profit without learning whether this allocation will help or hurt the recipient. Even subjects who make equitable choices when non-aligned payoffs are known will avoid information, especially when it is likely that doing so will not hurt the recipient. Through assessing the role of beliefs in information avoidance, this chapter provides an evaluation of several theoretical models of avoidance. The second essay considers the effects of both information avoidance and information-seeking behavior on charitable donations. Experimental subjects are allowed to avoid or seek out a range of information about a charity before deciding how much to donate to the organization. Donation sizes are positively correlated with the amount of information subjects choose to obtain. When subjects are required to read descriptions of charities, longer descriptions lead to higher donations. This indicates that agents may avoid further information if they have already learned about a person or charity in need, because learning more could obligate them to give more. The final essay studies selective exposure, the tendency to seek information that could support or validate one's beliefs or preferences but not maximize payoffs. Subjects in a context-free environment have to guess an unknown state of nature, and we induce preferences for one particular state. When given a choice between different information sources, around half of all subjects choose a source that potentially confirms that the state is the one they prefer, but is inferior in terms of expected payoffs. This finding holds consistently across a variety of contexts. The results of these studies have implications within experimental economics, since experiments tend to impose information on subjects that they might not otherwise gain. They also demonstrate that the ability to selectively acquire or avoid information can have a large impact on economic decisions.