Elsevier

Games and Economic Behavior

Volume 113, January 2019, Pages 651-672
Games and Economic Behavior

Communication is more than information sharing: The role of status-relevant knowledge

https://doi.org/10.1016/j.geb.2018.11.007Get rights and content

Abstract

In cheap talk games where senders' accuracy of information depend on their background knowledge, a sender with image concerns may want to signal that she is knowledgeable despite having material incentives to lie. These image benefits may, in turn, depend on the type of knowledge and its perceived social status. Theoretically, we show that when some senders care sufficiently about their image, there is both a non-informative babbling equilibrium, and a separating equilibrium, in which the average sender's message is informative and receivers always follow. In a laboratory experiment, we vary the social status of knowledge (1) by providing senders with multiple-choice questions on either (a) broadsheet topics (general knowledge) or (b) tabloid topics, and (2) by systematically modifying the degree of difficulty. We find truth-telling rates to be significantly higher when senders can signal high-status knowledge.

Introduction

In numerous situations, we need to rely on the advice by better-informed counterparts, whose interests may not be aligned with ours. Obviously, we should feel more comfortable trusting a person if we expect her to hold strong social preferences, or to be bound by social norms of truth-telling. But could people's desire for social status possibly play a similar role?

A fundamental element of communication is the fact that by transmitting information, we often also transmit our own knowledge, which in turn is a product of our own experience, interests, education, i.e. about who we are. When information depends on one's background knowledge, people who care about their image may be intrinsically motivated to demonstrate knowledge by telling the truth, even if facing material incentives to lie. A natural implication is that the scope and the (positive or negative) nature of image utility depends on the specific type of knowledge transmitted by the information. Following the concept of identity utility (Akerlof and Kranton, 2000), different areas of knowledge vary in the social status that they convey. In organizations, for instance, the value of knowledge is evaluated on the basis of its direct benefits (e.g., the usefulness to achieve a task and its uniqueness or accessibility by others), the manner in which the knowledge was obtained (formal vs. informal education) and implicit benefits from having the knowledge (feeling pride, power).2

In this paper, we study the transmission of knowledge in a sender–receiver game with misaligned preferences and compare the transmission from knowledge areas of differing social status. We propose a simple theoretical framework where the private information senders receive is not sufficient to infer the state of the world but requires a certain background knowledge. The knowledge of the sender is exogenously given by her prior education, training, interests, exposure, etc. Building on Bénabou and Tirole (2011), we assume that senders derive image utility from being perceived as high status and that they can influence their perception by showing knowledge. Different knowledge areas vary in their status-relevance (i.e. generally perceived relatedness to social status).

We show that, independent of status considerations, there always exists an equilibrium in which all senders babble and receivers ignore the messages and instead randomize evenly over all options. Yet, if there is a sufficiently large share of senders deriving sufficient image utility from being perceived as high status, there also exists a separating equilibrium in which some senders always lie and others always tell the truth. In this separating equilibrium, the average sender's message is informative and receivers always follow.

We predict (1) that when the sender's knowledge is required for extracting the true state of the world, an increase in the status-relevance of knowledge makes the average sender more likely to tell the truth and receivers more likely to follow the message, (2) that there is no effect when the sender can merely transmit information rather than knowledge, and (3) that an increase in the difficulty of the required knowledge makes senders more likely to tell the truth and receivers more likely to follow the message if difficulty is perceived to be highly status-relevant.

In the experiment, senders receive a multiple choice question with four answer options, of which one option is correct. Senders (receivers) gain a high (low) payoff when the receiver chooses a wrong answer and a low (high) payoff when the receiver chooses the correct answer to the question. In the high social status treatment senders receive questions on general knowledge topics (e.g., geography, history, literature), whereas in the low social status treatment questions cover topics such as tabloid TV, celebrities, sports and alcoholic drinks. The multiple choice questions were pre-tested to measure the difficulty of each question and the social status of the respective knowledge area. We validate the classification of questions into different status categories in a post-study with unrelated participants.

We find that correct messages in high amount to 46%, while only 32% in low. Moreover, we show that the driving channel is the ability to signal knowledge. When this opportunity is removed, the difference between high and low vanishes. In these no-signaling treatments, senders can only transmit the competence of a third person and not their own knowledge through their message. In addition, we find that the more difficult the question is, the more likely senders are to report a correct message.

Our experimental results demonstrate that knowledge-based information can enlarge the scope of truth-telling equilibria in sender–receiver games when senders care about being positively perceived. This paper contributes to the theoretical literature on communication and reputational concerns, which assumes that senders do not only want to trigger a certain action through communication but also want to be perceived as knowledgeable. In Sobel (1985) and Morris (2001) senders care instrumentally about their reputation (in order to be perceived as credible in the repeated game). Levy (2007) and Swank and Visser (2007) study the interaction of career concerns and transparency of the decision-making process in a committee. This paper differs in two regards. First, there is no instrumental benefit of the sender to build a reputation of being knowledgeable. We therefore propose a theoretical framework where the sender derives image utility from signaling her status. Second, we experimentally show that the psychological benefit depends on the social status of the knowledge area, which has not been explored yet.

This paper also relates to a number of studies which show that exaggerated self-assessments are related to social signaling. Burks et al. (2013) provide evidence that people who care about their social image make overly optimistic self-assessments. Ewers and Zimmermann (2015) provide causal evidence for this relationship. In their experiment people are willing to give up money to signal high ability. Individuals report higher ability in front of an audience compared to a situation where statements are private. Charness et al. (2018) show this effect in a strategic interaction where individuals use exaggerated self-assessments to deter entry of others.

The experimental literature on communication in cheap talk games has previously studied the role of behavioral motives to explain the frequently observed overcommunication (e.g., Gneezy, 2005; Cai and Wang, 2006; Sánchez-Pagés and Vorsatz, 2007, Sánchez-Pagés and Vorsatz, 2009). While standard theory predicts people to misreport information (Crawford and Sobel, 1982), a recent meta study on preferences for truth-telling (Abeler et al., 2018) identifies three types of motives: (1) a preference for being honest, i.e., individuals face a lying cost when deviating from the truth (Kartik et al., 2007; Kartik, 2009) or gain extra utility from being honest (Sánchez-Pagés and Vorsatz, 2007; Ellingsen and Östling, 2010), (2) a preference for being perceived as honest, i.e., individuals care about their reputation (Mazar et al., 2008; Fischbacher and Föllmi-Heusi, 2013; Utikal and Fischbacher, 2013; Gneezy et al., 2018), and (3) social norms, i.e., individuals care about descriptive social norms or social comparisons. For instance, individuals feel less bad about lying when they believe others are lying as well (Rauhut, 2013; Diekmann et al., 2015).3 This paper adds a further motive for truth-telling: Image concerns related to social status drive people to report honestly their high-status knowledge. This becomes crucial once information (what is being communicated) depends on senders' knowledge.

The remainder of the paper is organized as follows. The next section introduces the model. Section 3 presents the experimental design, Section 4 the results from the experiment, and Section 5 concludes.

Section snippets

Theoretical framework

We provide a simple framework that captures the experimental sender–receiver game of knowledge transmission (Ottaviani and Sørensen, 2006) and illustrates how information, depending on background knowledge and image concerns (Bénabou and Tirole, 2011) about one's social status perception, can influence truth-telling. The game deviates from a standard sender–receiver game (Crawford and Sobel, 1982) in two ways. First, the information's precision depends on the sender's characteristics. Second,

Experimental design

The following experiment provides a test of the hypotheses derived in Section 2. In particular, we vary (1) whether or not the sender has signaling ability, (2) the social status of the knowledge area, and (3) the level of difficulty. The experimental design comprises a pre-study, the main experiment, and a post-study.

In the main experiment (Table 1), we have a binary variation of social status. Subjects play a sender–receiver game, in which senders receive several multiple-choice questions

Results

This section first reports results concerning the senders' behavior and receiver's behavior across the two treatment dimensions, the social status of the knowledge area and the signaling ability. Second, we show results on the effect of difficulty.

A crucial prerequisite for the subsequent analysis is that the distribution of participant knowledge does not differ between the pre-study and the main experiment. For that purpose, we compare the share of correct answers in the pre-study to the share

Conclusion

This paper has investigated whether knowledge is used as a vehicle to signal social status and whether this affects people's propensity to tell the truth. In particular, we report evidence on the effect of knowledge-based information in a sender–receiver game with misaligned interests. In the game, the information's precision depends on sender's characteristics, i.e., her ability to extract the true state out of the given information. We demonstrate that senders communicating high-status

References (38)

  • R. Bénabou et al.

    Identity, morals, and taboos: beliefs as assets

    Q. J. Econ.

    (2011)
  • S.V. Burks et al.

    Overconfidence and social signalling

    Rev. Econ. Stud.

    (2013)
  • G. Charness et al.

    Self-confidence and strategic behavior

    Exp. Econ.

    (2018)
  • V.P. Crawford

    Lying for strategic advantage: rational and boundedly rational misrepresentation of intentions

    Am. Econ. Rev.

    (2003)
  • V.P. Crawford et al.

    Strategic information transmission

    Econometrica: J. Econometr. Soc.

    (1982)
  • A. Diekmann et al.

    Lifting the veil of ignorance: an experiment on the contagiousness of norm violations

    Ration. Soc.

    (2015)
  • T. Ellingsen et al.

    When does communication improve coordination?

    Am. Econ. Rev.

    (2010)
  • M. Ewers et al.

    Image and misreporting

    J. Eur. Econ. Assoc.

    (2015)
  • U. Fischbacher

    z-Tree: Zurich toolbox for ready-made economic experiments

    Exp. Econ.

    (2007)
  • 1

    We would like to thank Claudia Cerrone, Gary Charness, Peter Dürsch, Christoph Engel, Manuel Förster, Uri Gneezy, Jonas Hedlund, Oliver Kirchkamp, Sebastian Kube, Mark Le Quement, Moti Michaeli, Pedro Robalo, Joel Sobel, Franziska Tausch, and Christoph Vanberg for helpful comments. Research support from the Max Planck Institute for Research on Collective Goods and the Technical University of Munich is gratefully acknowledged.

    View full text