Advice and Beliefs in Experimental Games

New York University, New York NY

Investigators

Abstract

The decisions of economic agents are many times made on the basis of the advice they get from others. For example, people choose doctors based on the recommendations they get from friends, buy stocks on the basis of tips from strangers, choose schools for their children on the basis word of mouth advice etc. In short much of what we do is done by incorporating advice from others into our decision process. But little is know about how and why people follow advice. Do they do so because they think that others are better at doing the task demanded of them (i.e. investing their money, choosing a school for their kids) or do they do so because they think that others have access to better information? What are the characteristics of advice givers that make us more eager to accept their opinion? Is it their characteristics (i.e. gender, area of expertise, grade point average or IQ, etc.) or their track record on the task at hand (i.e. how they performed in the past, how experienced they are)? Does advice increase the efficiency of group decision-making or markets? In this experimental research we try to isolate what it is about advice that makes it so valuable to us and try to measure exactly how informative a piece of advice is. For example, in the experiments we run we present a decision problem to a subject in a computer lab where others have tried this problem before. These predecessors will furnish a pool of advice givers for the current set of subjects. The current set of subjects will be asked to make their decision under a set of different circumstances. For some subjects we will auction off the advice of subjects who has done the decision problem before them and try to infer from the winning price what the value of the advice is to them. We will also auction off the time series of decision made by the advice giver in his or her previous interaction in the problem. In other words, the advice giver may have played the game 60 times and made 60 decisions for which he or she was paid accordingly. Instead of auctioning off the advice of this advice giver, we might auction off the right to observe their history of play in the game. If advice were more valuable than information, we would expect that the advice would sell for more on the market. This would also indicate that people feel that others are better capable than they are of processing information. These types of experiments are aimed at investigating the value of advice to individual decision makers. Advice is also important to groups in the group decision problems they face. In this part of our experiment we will see if games played with advice from people who have played the game before is efficiency increasing. In other words, do societies make better and more efficient group decisions when they are able to get advice from people who have played the game before them? We think that this research is important because while advice is such a ubiquitous phenomenon, it is practically ignored in economics since economic theory is so eager to assume that all economic agents are both fully rational and perfect calculators. In such a world there obviously is no role for advice (everyone is an expert). Hence, if advice is efficiency decreasing then it would be important to know this and be able to increase our ability