The authors have declared that no competing interests exist.
Conceived and designed the experiments: EOK JPR. Performed the experiments: EOK JPR. Analyzed the data: EOK JPR. Wrote the paper: EOK JPR.
Spiteful, antisocial behavior may undermine the moral and institutional fabric of society, producing disorder, fear, and mistrust. Previous research demonstrates the willingness of individuals to harm others, but little is understood about how far people are willing to go in being spiteful (relative to how far they could have gone) or their consistency in spitefulness across repeated trials. Our experiment is the first to provide individuals with repeated opportunities to spitefully harm anonymous others when the decision entails zero cost to the spiter and cannot be observed as such by the object of spite. This method reveals that the majority of individuals exhibit consistent (non-)spitefulness over time and that the distribution of spitefulness is bipolar: when choosing whether to be spiteful, most individuals either avoid spite altogether or impose the maximum possible harm on their unwitting victims.
The Stanford Prison Experiment revealed the startling facility with which individuals lapse into antisocial and sadistic behavior when given the means and opportunity
Explanations for these (mis-)behaviors variously emphasize the importance of situational factors (e.g. dehumanization, differences in relative power, anonymity, action unobservability)
When we call an action “spiteful”, we mean that it directly imposes harm on another and provides no immediate benefit to the spiteful actor. Our notion of spite differs from that typically employed by evolutionary biologists in that the latter require the spiter to undertake an (expected) cost when reducing the relative fitness of the other
However, many empirical studies of spiteful behavior with human subjects suffer from identification problems. In the case of the classic social-psychology research
One advantage of studying spite in auctions, particularly second price auctions, is that spite can be measured in the intentional increase of the price that another bidder must pay. This element was recognized in previous studies of spite in auctions. For example,
Evidence from mosquitos indicates that when the costs are borne only by the target, spiteful behaviors can and will persist
Although (non-)spitefulness is a prominent behavioral pattern, little is known about how
The remainder of the paper is organized as follows: In the next section, we introduce the experimental design. The following sections report our results and a discussion of their implications. Then, we provide a detailed description of our experimental procedures, and
To generate opportunities to observe spiteful behavior and measure the extent of spitefulness, subjects participate in a sequence of 16 market periods in each of which they attempt to buy a single unit of a fictitious item. In each of the 16 periods, one unit of supply is available for purchase in a two-stage auction, and unlimited supply of an identical item is available at a fixed price after the auction ends. In the first stage of the auction, bidders are informed about their value and the fixed price at which they can buy after the auction, and they submit an initial bid for the auctioned item. In the second stage, everyone is informed about the highest initial bid, and subjects submit their final bids. The item is allocated to the highest bidder at a price equal to the
We call a bidder
We informed subjects only about the highest initial bid and
After reviewing the highest initial bid, a subject who is not the high bidder in the first stage can choose whether to engage in spite (by driving up the price) or not (by keeping his bid constant), but he may also choose to increase his bid above the highest initial bid in an attempt to win the auction. To measure the extent of spitefulness, it is essential that subjects frequently encounter a decision where the latter action is undesirable, or else we would rarely observe a choice between spiteful and non-spiteful bidding. To increase the likelihood that a subject may decide whether to be spiteful, we provide each bidder not winning the auction with the opportunity to buy an item - identical to the auctioned item - at a fixed price after the auction. Thus if the highest initial bid meets or exceeds the (expected) fixed price, a bidder should never choose to buy at auction and will instead wait to buy the identical item later on, but nevertheless, each bidder has to submit a final bid.
We collected all initial bids before providing subjects with any feedback and before beginning the second stage of any auction. This removes the possibility that spitefulness can be justified as a way to teach other subjects that submitting unreasonably high initial bids can be a costly mistake. We provided feedback on the auction outcome after all bidders submitted final bids. This provided subjects with many opportunities to observe how spiteful bids affected the winning bidders' earnings so that repeated submission of spiteful bids by the same subject cannot be dismissed by inadvertency.
We designed this experiment as part of a research program on price formation in auctions, unrelated to spiteful behavior. For this reason, our design incorporates an individual choice “real effort” task between the auction stages and an extra decision, prior to submitting the final bid, in which subjects may choose whether to recall the fixed price at which they may purchase in the aftermarket or to earn money in the “real effort” task.
Each experimental session consists of three stages: an “Opening” stage in which they submit initial bids in the auction; an “Effort” stage in which they earn money by completing a real effort task; and a “Closing” stage in which subjects submit a final bid in the auction and then subjects who are unable to purchase an item at auction.
In the “
Following the initial bids, subjects enter the “
At the end of the Effort stage, subjects enter the third and final stage of the session in which each auction ends in the same sequence in which the subject submitted bids in the
After each subject submits a new bid (or resubmits the original bid), those subjects who did not choose to observe the fixed price participate in one minute of the slider task in which each correctly placed slider yields a return of
Each auction consists of
Subjects are rematched across auctions and receive no additional information about the other bidders in their auction. The bidder's role is framed as that of a seller facing the opportunity of buying several commodities in the auctions for resale to the experimenters, one commodity per auction. One advantage of this framing is that it better motivates participation in multiple auctions. In the instructions, we inform subjects that they will learn the fixed price in the
We define
The solid line indicates the observed relative frequency of subjects making a spiteful bid in each trial, and the dashed line indicates the observed relative frequency of a maximally spiteful bid. Capped spikes display the standard errors of the mean in each trial. The subgraph displays the number of opportunities to observe a spiteful bid in each trial, and suggests that the relative frequency of observing spite is not related to the number of opportunities to be spiteful.
To facilitate inter- and intra-subject comparisons of the
The height of each bar represents the relative frequency of observing spitefulness in each of 50 intervals of length 0.02. Nearly 70% of the weight of the distribution is in the extreme values zero and one, indicating that, given the opportunity, subjects are either maximally spiteful or not spiteful at all.
Our measure of spitefulness
The panels present the data separately for males and females. Each data point represents one observation of spitefulness relative to the magnitude of maximum harm possible.
Panel regression analysis (
Independent Variable | Coefficient | Rob. Std. Err. |
|
95% conf. interval | |
Maximum Possible Harm | 0.00015 | 0.00011 | 1.32 | 0.187 | [−0.00007, 0.00036] |
Aftermarket Price | −0.00020 | 0.00022 | 0.92 | 0.357 | [−0.00063, 0.00023] |
Female | −0.05767 | 0.11480 | −0.50 | 0.615 | [−0.28267, 0.16733] |
Female (Max. Harm.) | −0.00005 | 0.00014 | −0.37 | 0.708 | [−0.00033, 0.00022] |
Trial | 0.02105*** | 0.00652 | 3.23 | 0.001 | [0.00827, 0.03383] |
Constant | 0.38353*** | 0.11886 | 3.23 | 0.001 | [0.15058, 0.61649] |
Significance levels are denoted by:
Finally, we examine the heterogeneity of spitefulness between and within individuals. While 5 subjects are maximally spiteful at every opportunity (i.e.
More generally, many subjects display striking consistency in their level of (non-)spitefulness (
One potential concern with our method of measuring spite is that some bids which we label spiteful may result from alternative bidding strategies. For example, bidders who discover that the initial high bid is greater than their value have a weak best response to bid their value, and if their initial bid is less than their value, this strategy will produce a positive measure of spitefulness. Similarly, bidders who know the posted aftermarket price have a dominant strategy to bid the minimum of their value and the posted price; here too, for a sufficiently low initial bid, a final bid that follows this strategy will be measured as spiteful.
Panel (a) shows the relationship between potentially spiteful final bids and values, and panel (b) shows the relationship between such bids and the aftermarket price. Each point represents a single bid, and the lines provide a reference showing where the bid equals each value or aftermarket price.
Our experiment places subjects in repeated situations in which they choose whether to spitefully harm an anonymous other at zero cost and in which spiteful acts are not revealed as such during or after the interaction. Because we have access to information not only about how much harm was imposed, but also how much harm
All experiments were conducted with the informed consent of 48 healthy adult subjects who were free to withdraw from participation at any time. Only individuals who voluntarily entered the experiment recruiting database were invited, and informed consent was indicated by electronic acceptance of an invitation to attend an experimental session. The experiments were conducted following the peer-approved procedures established by Maastricht University's Behavioral and Experimental Economics Laboratory (BEElab). Our study was approved by the BEElab at a public ethics review and project proposal meeting that is mandatory for all scholars wishing to use the BEElab facilities.
The experiments were conducted at the BEElab of Maastricht University with 48 students. Subjects' decisions were fully incentivized using Experimental Currency Units (ECUs). Their ECU-profits were converted to EUR at a rate of 100 ECU = 1 EUR and these from two randomly selected auctions were paid to them in cash at the end of the experiment. In order to avoid the influence of wealth effects, at the end of the experiment, we randomly select two auctions for each subject for payment. Then subjects receive private cash payments including a 4 EUR payment for arriving to the experiment on time, their earnings from the Effort stage, and their earnings from the two randomly selected auctions (including what they earned in the slider task, if they participated).
In total we ran 4 sessions of 12 subjects drawn from the undergraduate population of Maastricht University (Average age = 22.6, 46% Female). On average, subjects earned 15.32 EUR for a 60-minute session ranging from a low of 6.93 EUR to 26.65 EUR, including show-up payment. All sessions were conducted in May 2011.
Our experiment was programmed using the z-Tree software package
Instructions that were provided to experiment subjects.
(PDF)
We thank Björn Bartling, Dan Friedman, workshop participants at Simon Fraser University, two anonymous referees, and an associate editor for helpful comments. This research was funded by grants from Maastricht University's METEOR research program in Economic Behavior, Theory and Computing.