Alcohol influence on economics-related decision-making

Размещено на http://www.allbest.ru/

Introduction

Throughout my research, I have discovered that no behavioral experiments related to alcohol and economic decision-making have been conducted in Russia. This paper observes the first try in this field, an experiment attempting to test three sets of embedded hypotheses regarding how alcohol influences our choices.

It describes a number of problems faced on the way and suggestions to avoid them, as well as several successes which have been attained with this first try in this field of research.

Even though statistical tests proved that not all of the results of this study fulfil the hopes put in them, it definitely may serve as a foundation for anyone who wishes to develop this topic, which is quite significantly important for business-related and interpersonal relationship in other country.

The experiment was an unusual and pleasant experience for everyone involved, and I hope that this paper will be an interesting and educational read too.

Literature review

A vast amount of work has been done to determine the way in which alcohol affects our thinking, behavior and decisions. In the early stages prior to writing this paper, many scientific disciplines, e.g. medicine, economics, sociology, psychology investigated into this topic from a 360 degrees perspective.

The neuro-imaging study by Gilman et. al. covered the topic of risk-taking by alcohol-induced individuals. They found that brain areas responsible for rewarding process become more active under alcohol influence, while areas responsible for feedback and responsiveness, such as reaction to loss, were blunted. The overall conclusion was an increase of risk-taking, especially when the stimulating effect of alcohol on an individual was prevailing over the sedative one. (Gilman et al., 2012) A study following this one broadened the range of research, proving the existence of influence on “brain's dynamic response to task demands and regions thought to govern motor control, motivation, and executive control, such as working memory and attention”. Both of these studies used the intravenous method of alcohol delivery - a technique decided to be left aside for the current research due to possible difficulties with finding participants.(Bjork and Gilman, 2014). I decided not to aim this research at checking the difference in risk-taking, as a vast multitude of works has already observed and tested it. The risk-attitude is assumed to be increasing under alcohol in this work. The notion of two effects of alcohol - sedative and stimulatory - with opposite direction of influence on certain aspects of behavior have been paid additional attention to. During the experiment, effort was made to mitigate the spread of sedative effect - most often associated with relaxation, melancholy and boredom. The tables with participants, who finished their tasks before others and have to wait without any activity were specially attended to by the moderator or his assistant to engage the participants in conversation. Also, an additional game was added in the break period (necessary for alcohol to have its effect), to help the time pass in a more lively way. Overall, the researches above became a foundation, telling that the effect of alcohol on certain brain areas which determine humans cognitive abilities exists, and thus, the effect on behavior is to be expected.

The accomplishments of works in this field made by economists include testing how General Axiom of Revealed Preferences holds, as well as the independence axiom. They have found no ground for claiming a significant change for subjects under alcohol. No irrational behavior or unresponsiveness to options provided was found even at high doses of alcohol. The findings, on the other hand, included the observation of slowed-down reaction time and worsened coordination, as well as higher risk taking. However, this experiment used volunteers in the bar, with no control over the prior dose taken. (Burghart et al,2014) Nevertheless, their conclusion that the extent to which GARP and independence axiom hold depends on the individual, and not on the degree of intoxication (even for high doses) was considered to be credible enough to not include testing the two axioms into this work.

Several studies were concerned with exactly which dose the participants should be targeted at. A research by a group of scientists concerned with drunk-driving problem proved that the effect on both dispersed and focused attention and visual and motor control appears significantly on 0,05 - 0,07% Blood Alcohol Concentration (BAC) among the subjects. Another research tested the level of 0,03% BAC, finding that it doesn't affect functional or rhetorical abilities, but the mechanisms of obtaining and processing of visual information are already impaired. (Allen et al,2008). These results led to believe prior to the experiment that the BAC of 0,05% can already be enough for obtaining significant results

Moreover, no evidence of a behavioral experiment observing alcohol-induced condition was found among studies conducted in Russia. The problems usually tackled in the works observed mostly address problems caused by excessive alcohol use, connecting alcoholism to various micro- and macro-level phenomenon, as well as observing the nature and effects of continuous alcohol abuse. Several studies connected alcohol consumption with religious confession, while others described how people perceive themselves under alcohol.

This fact proved to be encouraging, as it confirmed the novelty of the then planned experiment in this field in Russia.

Perhaps the vastest study on this matter has been conducted by a group of Italian researchers in their work “Economic Behavior under Alcohol Influence: An Experiment on Time, Risk, and Social Preferences”. They used placebo to test pharmacological and psychological effects of alcohol on optimism, risk preferences, time preferences and altruism. Their findings regarding risk-taking support absence of change in risk-aversion for male subjects and increased risk-aversion for female ones. Impatience is found to be positively correlated with intoxication, and the effect is “essentially pharmacological and goes beyond the expectation-mediated effects”. This was one of the most inspiring results found during the research on the topic. The work itself proved to be a rich resource of knowledge of the field and its specificities. The original plan of experimental procedures of my work included a placebo design as well. However, the placebo procedures used by the Italian researchers (peach juice with ethanol for treatment and peach juice with a little grappa on the surface and borders for control group) did not appear to have enough credibility in causing subject's misperception. A firm producing non-alcohol analogue of vodka, claimed to be non-discernible in taste or smell was found. But delivery and costs problems made me stick to having no placebo group in the study.

As for generosity, they checked the stated potential donations to humanitarian projects, with variety regarding sympathy towards project's cause (“cold” and “hot” projects). They found that the subjects gave less under alcohol influence, which was attributed to “lower conformity to social norms”. This explanation was used in this work for one of the hypothesis stated.( Luca Corazzini, Antonio Filippin, Paolo Vanin,2015). In addition, this is supported by several medical studies, both those conducted on humans and on animals. The kinds of behavior, realization of which is slowed down by certain brain areas, appears under the alcohol influence (Eysenck, 1973). An experiment on cats proved that the negative stimuli normally stopping he subjects from realizing their desired type of behavior tends to lose effect after alcohol consumption.

The overall outcome of the analysis led to choosing the following set of consequences of alcohol consumption as positively existing to a varied degree:

1. Change in risk attitude. Depending on which effect was dominating - sedative or stimulating - a person becomes more risk-loving or risk-averse. Studies connect one or the other effect prevailing with individual's expectations, rather than the alcohol effect itself. Some specific features of the experiment tend to lead us to expecting the stimulating effect to be prevailing. Consumption of vodka in shots with biting lemon after, together with the whole unusualness of the setup led to participants showing more signs of excitement rather than drowsiness. (Fanny Kreuscha, Aurйlie Vilennea, and Etienne Quertemont,2013)

2. Decrease in alertness towards external stimuli, less attention towards the task and more difficulties in information processing. Basically, people under alcohol influence feel more relaxed when working on a task. They do not take in all the details, as the details and the task itself provide a bigger challenge now. But at the same time, this challenge grasps subject's attention, so he/she notices the surroundings less. (Dirk Breitmeier , Irina Seeland-Schulze, Hartmut Hecker & Udo Schneider 2007)

3. Rationality does not disappear. As much as we can call an individual rational, the level of rationality stays for moderate-to-high doses. Responsiveness to incentives, reaction to difference in options and consistence in preferences depend on individual's specific characteristics more than on blood alcohol level. (Burghart et al.,2014)

In addition, a number of studies regarding the determinants of blood alcohol concentration were assessed. Although the list of factors found to be significant for the determination is vast, including drinking experience, years in school, personal level of alcohol inhibition and many others. Most of these factors are inaccurately stated when self-reported or impossible to determine without medical testing. Thus, the dose-determining factors were reduced to the objective minimum: height, weight, gender and stomach fullness. (John T P Hustad, Kate B Carey,2005)

Methodological framework

Three sets of `embedded' hypotheses were considered:

1. Alcohol affects perception at individual level by making individuals less sensitive to stimuli. Thresholds of perception of changes in stimuli become wider, resulting in greater tolerance to potential losses, and neglect of profitable opportunities. In terms of prospect theory, for instance, this may result in mitigating possible losses and boosting up possible gains, altering the shape of the value function: instead of

it may become

As an experimental treatment, we used several lotteries to be given within-subject to a person before and after alcohol consumption to 1) make pairwise choice, and 2) state the certainty equivalent (minimum WTA) to the twelve lotteries in the list

#	Outcome1,p1,Outcome2,p2	EV	Outcome1,p1,Outcome2,p2	EV
1	$4.00, 35/36; -$1.00, 1/36	3.86	$16.00, 11/36; -$1.50, 25/36	3.85
2	$2.00, 29/36; -$1.00, 7/36	1.42	$9.00, 7/36; -$0.50, 29/36	1.35
3	$3.00, 34/36; -$2.00, 2/36	2.72	$6.50, 18/36; -$1.00, 18/36	2.75
4	$4.00, 32/36; -$0.50, 4/36	3.50	$40.00, 4/36; -$1.00, 32/36	3.56
5	$2.50, 34/36; -$0.50, 2/36	2.33	$8.50, 14/36; -$1.50, 22/36	2.39
6	$2.00, 33/36; -$2.00, 3/36	1.67	$5.00, 18/36; -$1.50, 18/36	1.75

These tables are first given before taking alcohol, and then, the same tables after alcohol, but with gains expressed in rubles and probabilities in decimals, balancing the order of questions among the subjects. The expected values are not given, and the use of calculator is not allowed. (standard for preference reversals test) (Slovic and Lichtenstein, 1971) When an individual first chooses first option in a pair, and then it turns out that he/she assigned a higher certainty equivalent to the second one - that is a preference reversal.

The data from this game will allow to compare for consistency of preferences through pairwise choices and higher certainty equivalents as well as for the frequency of the reversals themselves.

Also, a test was decided to be conducted against the control group to find whether expectations of alcohol intake affect the number of reversals in any way.

So, the first hypothesis is that the number of preference reversals increases under alcohol effect.

2. Alcohol distorts the utilities in general, resulting in not only parametric changes in utility shapes, but in utility specification itself. Specifically, consider a person affected by alcohol whose preferences before have been characterized by a utility function v(x). After this person got some shots, his or her utility becomes u(x) - not necessarily affine or monotonic transformation of v(x). Testable implications may be stated in terms of inequity aversion theory: instead of

where x is player's gains, y is player's partner's gains

The utility may become

 (insensitive to losses of the other player),

(insensitive to own losses)

or any other specification. Natural test for that would be a direct attempt to decompose preferences in terms of ultimatum game (taken in form of impunity, generosity and envy games).

Conventional ultimatum game: one player proposes a division of the pie (continuous variable), another accepts or rejects.

As a significant number of test subjects had economic background and thus were familiar with ultimatum game, the conventional ultimatum game was left out. Players with experience in a certain game tend to form a certain strategy which they stick too, so actions in such games are unlikely to be affected by mild intoxication. Thus, the following versions were used:

· Impunity game (also known as dictator game): same as ultimatum, but in which the proposer receives whatever he or she offers no matter what the decision of the responder is.(Bolton et al, 1998)

· Envy game: same as ultimatum, but proposer is a residual claimant who chooses a pie size from which the responder receives a fixed share. The responder can either accept or reject the choice of the proposer. (Sandro Casal & Werner Gьth & Mofei Jia & Matteo Ploner, 2011)

· Generosity game: the proposer determines the size of the pie, knowing that his/her share is fixed, and the share of the responder is decided by the proposer, and can vary in a systematic way. The responder may then accept or reject the offer. (W.Guth,2010)

Upon, knowing this information, behavior of drunken and sober proposers can be compared in these three games, with the hypothesis that drunken proposers will be less sensitive to the possible gains of the other player.

Also, the pattern of offers and acceptance were expected to differ between endowment options, because subjects, with a high degree of probability, consider the choice of any option in terms of value relative to the endowment. Of course, exact calculation is unlikely to be expected, so a rule of thumb is probable: “the offer is this number of steps away from the endowment”. As the scales are relatively different for three game instances (endowment is located to the right of the scale center for the endowment of 70 instance, to the left of it for 50, and exactly at the center for 30), the decision involved should differ too. The offers of same relative value (for example, sender offers 75, when his/her endowment is 70, or 55, when his/her endowment is 50, which are both one step higher than the endowment value), are likely to face differing rejection patterns due to receiver seeing the possible range of not chosen options (75 is almost the maximal offer possible, while 55 is far from top offers).

Design used in studies on the games (Gueth et al., 2011) was used to work on the specific versions of the games, including a change in measurement units to make the tasks look different before and after alcohol.

One half of the participants provided data regarding the offers in these games, while the other half provided data on the acceptance rates and patterns. Players knew they were playing for real money rewards to guarantee more serious approach to the task.

Based on the results discussed in the literature review, the second set of hypotheses is the following:

a) Offers in Envy game and shares claimed in Impunity game by senders increase. The social norms preventing individuals from choosing the most self-benefitting options will break down with alcohol.

b) Rejection rate for receivers in Envy and Generosity Game become higher. Individuals lose sensitivity to their own losses, valuing punishment of unfair actions higher.

3. Alcohol inhibits abilities for strategic reasoning, which go beyond utility specification, but apply to strategies' profiles (typically, in equilibrium). For instance, players in the four games (Falk Fehr Fichbacher 2003) will typically fail to recognize what games are more and less inoffensive for the receiver. For example, the game where proposer has two same options of offers is expected to receive much less rejects than the one where proposers can give 80 percent either to himself or the partner and chooses to give less to the partner. In Western countries, proportions of rejections decrease exactly in the order of: b>a>c>d, whereas it is known that among the animals, there is no significant effect. The hypothesis would be that drunk responders will behave in a way similar to animals, i.e. will lose sensitivity to the generosity of the proposers. Thus, the tested hypothesis is that after alcohol, the proportion of rejection for all 4 options becomes roughly the same.

For all the games above, data before alcohol intake is compared with data obtained after and tested for difference using chi-squared or Wilcoxon-Mann-Whitney test. Also, a comparison with a control group is made, to control for effect of difference occurring due to the task repetition after certain time period. Stata 9.2 software was used for conducting the statistical tests on the data obtained.

All participants were asked to fill in an online survey after the experiment. The questions concerned the existence of economic background of the participant, experience with alcohol and its self-perceived effect on the individual. The survey data may show the difference between decisions of economists and non-economists. People with economic background were expected to put more weight on their own gains and neglect the gains of others, perceiving maximization of their own utility. Also, they were expected to choose the marginal options and stick to their strategy after alcohol intake more than others. For non-economists, more emotional behavior was expected, shown by choosing a certain value corresponding to “fair” distribution of gains between them and the partners.

Experimental procedures

Participants for the experiment were found using posts in social media and word-of-mouth. The total number of participants was 61 for treatment group, 30 for control group for game 1 and 11 for control group for games 2 and 3. More than 60% of the participants had not been acquainted with the moderator prior to the experiment. The online advertisement for the experiment suggested an opportunity to take part in an experiment in behavioral economics for a monetary reward, while warning about the necessity to consume a certain amount of alcohol during the procedure.

All experiments with the treatment group took place in a cafй located at Shabolovka 24.

The cafй administrator allowed to use one of the cafй halls as the experiment room, free of charge. The cafй does not serve its own heavy liquor, so it was allowed just for the purposes of conducting the experiment to bring bottles and lemons bought elsewhere. Shot glasses, saucers for lemon slices, a knife and a cutting board were provided by the cafй staff.

Consumption of food or drinks was not allowed during the experiment for the participants. However, after all the tasks were completed by all participants, there was a period of time during which the moderator had to calculate the rewards of all players. Participants were encouraged to order something during that period.

There were 9 such sessions. Experiment sessions took place at varying time, depending on when it was most convenient for participants of the particular group to come. The earliest starting time was 12am, while the latest being 10pm. The number of participants in one session varied from 4 to 11. In case when an odd number of players took part, filled task sheets with data from previous experiments were used to calculate the rewards for games requiring two-player interaction. The cafй contained normal visitors too at the time of the experiment. Some of them showed various degrees of interest towards it, but none were allowed to take part due to having had already consumed alcohol recently before. No incidents regarding visitors bothering participants occurred. The moderator answered any questions addressed to himself or the participants individually to each asking visitor in order not to disturb the participants during tasks completion. But during most of the sessions no visitors approached any people involved in the experiment during sessions.

There were two more sessions with control groups, where no alcohol was involved, taking place at Shabolovka 26, inside the building of HSE University. This site could not have been used for treatment group experimental procedures due to ban of drinking on the university grounds.

The experiment was moderated either with the main moderator alone or with help from a volunteering assistant. The assistant was sometimes chosen from among previous participants, but he/she was never allowed to take part in the experiment after assisting, even if he/she hadn't participated before. The assistant's role was to help with faster distribution of task sheets and collecting them after, as well as answering the participants' questions. Usually, one team was assigned to be controlled by the main moderator, while the other was taken care of by the assistant. The main rules were still announced by the main moderator. Assistant also helped with alcohol procedures details: he/she took the shot glasses, saucers, knife and cutting plate from the bar and cut lemons while the main moderator was explaining the rules, and then distributed shots and saucers with lemons. The main moderator was the only one pouring alcohol to participants to ensure the correctness of each dose.

Upon arrival to the experiment site, participants were sat at the tables in such a manner to follow this set of rules:

- no more than 3 people at the same table

- there are two teams of equal sizes: receivers and senders (player A and B). People from different teams sit at different tables.

- participants who know each other well belong to the same team. This condition is important to prevent collaboration between possible partners in games.

After the whole group of participants got together, instructions and agreement forms were given.

Time is given for everyone to read the instruction and fill in the form, then the most important rules are highlighted by the experiment moderator for everyone to hear:

1) Participants can't communicate with players from the other team. This will be punished with a decrease in the final reward. Due to only one room being available for experiment purposes, players had incentive to communicate for higher outcome. This rule successfully helped to avoid the communication.

2) It is not forbidden to talk with your teammates, but it is necessary to make decisions on your own. Participants were still allowed to discuss the rules with their teammates to help them understand.

3) Every participant's instruction contains a Greek letter written on its back side. Participants should not let the other team members get to know their letter. Every sheet of paper received from the moderator during the experiment should be signed with this letter, as well as with the word “before” or “after”, depending on whether the particular sheet was received before or after consuming alcohol. The letter method was decided to be used instead of signing with names to avoid bringing in personal emotions about a particular player into decision making in games. The “before” and “after” part helped the experiment moderator to analyze the results quicker.

4) Participants may not leave the experiment room or consume any food or beverages except for those given by the moderator until the experiment is over. This rule is important to ensure equal treatment for all test subjects.

Right after, the first game sheet was given. The order of the game sets varied between experimental groups, but was the same for all the participants inside.

Each next task sheet is given to all the players at the table simultaneously, when all of them complete the previous task. When the next task belongs to a new game set, the task sheet is given to all players from both teams after everyone has completed the previous one.

For games from set 2 and 3 (those requiring interaction between players from opposing teams), players are paired randomly for each task sheet for outcome calculation. So each player interacts with different players from the opposing teams even among the tasks in the same game set. All of this information is known to participants.

It was allowed for the participants to ask the moderator any questions they want. If these questions were regarding the understanding of tasks, the moderator answered immediately and then repeated the answer to all the members of the inquirer's team or to all the participants, depending on whether the question concerned a task specific to one team or to both of them. For all other questions, the inquirer would be asked to leave the question until the end of the experiment.

Games description

Game set 1:

Lottery Choice and Certainty Equivalent

Participants receive a list of 6 pairs of lotteries. For each lottery, the two possible outcomes are known, as well as the probabilities of these outcomes. The participants had to choose one preferred lottery in each pair.

Upon completing the task, the second sheet was given, containing 12 lotteries, also with outcomes and probabilities known. The task was to state the certainty equivalent - the minimal sum of money the participant would agree to receive instead of the possibility to play the lottery - for each option.

Game set 2:

Envy, Generosity and Impunity Games

The participant's role in these games depended on the team he/she had been assigned to.

Players from team A were making the offer about rewards among the options in the tables, while players from team B were choosing whether to accept or reject the offer for each possible offer.

Envy game: Player B receives an endowment. Player A chooses what would be his/her own reward in this game and makes an offer. If that offer is accepted by Player B, Player A receives the proposed amount and Player B holds the endowment. If the offer is rejected, both players get zero.

Generosity game: Player A receives an endowment. He/she then chooses what would be the reward of Player B in this game and makes an offer. If that offer is accepted by Player B, he/she receives the proposed amount and Player A holds the endowment. If the offer is rejected, both players get zero.

Three instances of the two games above differing in endowment sizes were provided. The endowment sizes in rubles were 70, 50 and 30. The ranges of offers were from 40 to 80 for endowments 70 and 50, and from 15 to 45 for endowments of 30, all with the step of 5 between the offer options. This implies that players, after observing all three game instances, see the difference in scales and endowments and likely consider the offers in sense of relative units, comparing the offer and the endowment.

Impunity game: only players from team A take part in this game. They decide how to split a money prize of 100 rubles between themselves and a random player from team B. Player B cannot reject the offer, which is known to players from team A.

Game set 3:

Strategy profiles

All players receive a set of 4 pairs of offers regarding splitting money prizes. For each pair, the players should decide one preferred offer they make to a player from the opposing team. After the 4 choices are made by each of the participants, the task sheets are taken by the moderator and interchanged for task sheets filled in by a member of the opposing team for each player. The players then observe the offers received and decide whether to accept or reject it for each of the 4 pairs. If the receiver agrees, the money prize is divided according to the chosen offer. If the receiver rejects, he/she and the sender then get zero.

All 3 game sets are given two times: before and after players consume alcohol. The tasks received after alcohol are altered in a sense that endowments, outcomes and offers are given in different units, while the way probabilities are given switches from simple to decimal fractions, or vice versa.

In game set 1, players get tasks with either decimal probabilities and outcomes in rubles or with simple fractioned probabilities and outcomes in dollars (one dollar considered to be equal to 50 rubles). They solve both options during the experiment, with random distribution regarding which is before alcohol consumption. Players at the same table were given these options in the same order.

In game set 2, players first get the tasks with offers and endowments in rubles and then, after consuming alcohol, the tasks with special money units (equal to 5 rubles).

In game set 3, players randomly got either the task with offers in rubles or the task with offers in percent points of the money prize before alcohol consumption and then the other one. All players are given these options in the same order.

Upon completing 3 game sets, each table received the shot glasses and saucers with sliced lemon. Alcohol was then distributed and consumed, with nothing to drink or eat after, except for the lemon slices provided. In 7 minutes after the consumption, the participants are asked to switch tables and are given the new task sheets. During the waiting period, a rock-paper-scissors tournament was held to pass the time. The winners among each table's players went to the final rounds to compete for the prize of 50 rubles. This helped to pass time quicker and to support the stimulatory effect of alcohol over the sedative effect.

The control group did not consume alcohol at this time, but the pause between tasks still remained for them. The overall pause between end of the last task of the first part and handout of the first task of the second part was the same for treatment and control groups.

For game sets 2 and 3 interaction between two players is required. In game set 3, players first filled in the offers and then received offers from the other team and filled in acceptance/rejection decisions. The partner's identity was not known due to use of Greek letters instead of names, and interacting pairs varied before and after alcohol randomly, so that players had different partners and did not have motives for retaliation. For some sessions, the experiment moderator secretly filled in the offers himself secretly, putting the most offensive offers, to get more data on rejection rates for these offers. Players still were told that the offers come from the other table's players.

For game set 2, players did not observe their partner's decisions. However, after the end of the experiment, task sheets were `partnered' randomly to calculate the money rewards. The task sheets were paired in a manner providing that each player could have up to three different partners for game set 2 (one for each game), but having only one partner was possible too. This implies that the players knew they don't play with the same player in games inside the set, or in games before and after alcohol. The rewards calculation was made right after the end of the experiment by the experiment moderator. The reward was based on games of set 2 only. Only one set was used for calculation - either before or after alcohol consumption. The choice was made using a flip of a coin. Maximum possible reward was 455, plus 50 from the first prize in rock-paper-scissors tournament

Alcohol procedures

All participants consumed “Green Mark” vodka (“Зелёная Марка”).

The dose was calculated individually for each player before the experiment, using a professional alcohol calculator developed by and for forensic medics.

The determining parameters are gender, weight, height and stomach fullness. The calculator also allows for input of values for minimal, average and maximal alcohol elimination rates of the organism, but these were set to default due to impossibility of determining without special medical tests.

All participants' doses were targeted to achieve the blood ethanol concentration level of 0,5%.

This level of BAC is considered the border line for legal driving in several European countries, such as France. The proposal by Ministry of Healthcare of Russia of 1967suggested 0,5% BAC as the starting point for mild intoxication. Clinical studies (Zoethout et al., 2011) tested the acute effects of alcohol on central nervous system and determined the 0,5-0,7% BAC to be the level at which not only speaking abilities and information processing are affected, but visuomotor controls and focused and divided attention are also impaired. Thus this level is expected to be enough ground for appearance of evidence for changes in task-solving, but not high enough to cause negative consequences among the test subjects. The doses given to participants ranged from 65 ml to 150 ml, with mean being around 120 for males and 95 for females.

A measurement glass was used for exact pouring of each dose. Shots provided at the cafй were exactly 50 milliliters, and the measurement glass was 30 ml, with markings for each 5 ml. Only lemon slices were available for all participants and were allowed to be used to follow the vodka. The switching the table procedure is conducted upon players for the reason that switching the position from sitting to standing helps alcohol to spread through the organism (and the effects of alcohol to be felt) more effectively.



No accidents connected to alcohol consumption occurred, the lemon provided proved to be enough to help through the possible troubles with vodka drinking.

Data analysis

Data from 61 participants from treatment group, 11 participants from control group for game sets 2 and 3, and 30 participants from control group for game set 1 was collected as a result of experiment sessions conducted. Observations from one participant from treatment group was eliminated from the analysis, as it was found later that he had already consumed alcohol by the start of the experiment.

In the following description, significant result means significance at 5% level.

Game set 1:

Data on preference reversals for each pair of lotteries has been collected and analyzed.

The proportion of reversals for treatment group before alcohol was calculated and then compared using binomial tests with the proportion after alcohol and the proportion for the control group. No significant difference has been found.

After calculating the proportion of players having each a certain number of reversals (maximum being 6), a chi-square goodness of fit test was used. It did not show any significant difference either.

Therefore, it cannot be concluded that preferences become less consistent under mild intoxication.

Game set 2:

Offers from Players A and acception/rejection rates from Players B were analyzed separately.

Player A

A difference in offers was calculated separately for each endowment instance and each instance. Then, two one-sided and a two-side t-test was used with H0:difference=0.

No difference between Generosity game instances proved to be significant. Standard deviation decreased after alcohol intake for all three endowment instances, but changes in mean value do not allow us to make a significant conclusion.

As for the Envy game, for the endowment of 70, one-sided t-test successfully rejected H0 in favor of Ha: difference<0.

A reminder: in this game, Players B received an endowment, while players A could offer the amount of money in their own reward in this game. Players B could then accept or reject this offer.

However, for the endowment of 70, the scale of offer choices was from 40 to 80, so there were only two options higher than 70, in contrast with endowment of 50, where 6 such options were available, or of 30, with 3 such options. Less players chose 80 (the maximum value) after alcohol intake. It can be observed that standard deviation increases after alcohol intake both for endowments of 50 and 70, so we can say that many individuals changed their decision, but the changes were in both increasing and decreasing directions for the endowment of 50, with more people choosing the maximum of possible offers and more people choosing the minimum of possible offers at the same time. Still, only for the endowment of 70 a significant conclusion can be made.

Also, no players chose the minimum offer for endowments of 70 or 30, supporting the assumption that players would deal in values relative to the endowment when making the choice. The interpretation can be that after alcohol player became more humble, feeling uneasy with choosing a reward higher than their partner's, or they feel more compassionate to another person, who also drank with them, and find it uncomfortable to take a higher reward for yourself.

The differences in offers in Impunity game were calculated and tested through t-test as well first, and then using Wilcoxon signed-rank test. The results showed a significant change after alcohol: the players became more generous towards their partners. The explanation for this is consistent with a commentary given by one of the participants: players felt more compassionate and sorry for their choiceless partners.

Thus, the hypothesis of increase in offers is not supported. The opposite is true: the Envy game and Impunity game instances show a decrease in offers. It implies that the alcohol consumed did not lead to the subjects maximizing only their own rewards - it caused them to be more compassionate and share more.

Player B

For each value of offer, for each endowment instance, the proportion of acceptance by the receiver was calculated for data after alcohol consumption. Binomial tests were then run through each value of offer of each endowment instance from data before alcohol intake to compare with the proportion in the matched pair data. The significant results for Envy game are only for the following treatments:

1) For endowment of 70: players accept proposal of Player A's reward being equal to Player B's reward more after alcohol intake. A number of players accepted only offers where Player A was trying to claim a reward less than Player B's endowment. A significant percentage of them became less strict and agreed to accept equal distribution of rewards.

2) For endowment of 50: players accept proposals of Player A's reward being 55 and 60 less after alcohol consumption. Player A's decision to earn a bit more than Player B's endowment apparently becomes more annoying to participants after alcohol. At the same time, these players are still as fine as before with their partner claiming the highest prize possible, as they find it rational.

3) For endowment of 30: players accept proposal of Player A's reward being 45 (the maximum for this scale) more after alcohol consumption. The possible cause of change is that a large number of participants rejected only the offer of highest value before alcohol. After alcohol, a significant part of them relaxed and decided to accept everything, for either generosity reasons or desire to save the effort with filling the form.

For generosity game, only 1 result was significant: for Player A's endowment of 50, players accepted the offer of getting the same reward from the partner less after alcohol. The equal offer of 50 is located at the very beginning of the scale for this endowment, with only 40 and 45 being worse options. So an explanation can be that after alcohol participants that even this offer is too small, as it literally costs nothing for the partner to give more. Also, according to one of our assumptions, they become less responsive to possible losses, so it is easier for them to get zero if that means punishment of an unfair partner.

For another test, the proportions of acceptances of each player were counted for each game instance before and after alcohol. A Wilcoxon signed-rank test was used to compare these proportions pairwise, and the results were significant for Envy game instances with Player B's endowments of 50 and 30.

The personal acceptance rates decreased for endowment of 50 and increased for endowment of 30. This is consistent with results of binomial tests and can be explained by the difference in scales.

Therefore, we cannot say that our hypothesis of a decreased acceptance rate has been rejected. The pattern certainly changes, but the implications turned out to be more complex than just insensitivity to losses.

No significant difference has been found between decisions in the control group's participants.

Game set 3

Offers in 3 pairs of options have been observed:

1. Player gets 20% - partner gets 80% against Player gets 80% - partner gets 20%

2. Almost 47% of players changed their decision in this pair after alcohol.

3. Out of them, only 38% changed their decision for the option more beneficial for their partner.

4. However, even more players in control group (about 55%) changed their decision in this pair as well, with only 33% changing it for their partner's benefit. Thus, we can conclude that this effect is more likely to be attributes to retaliation, despite the effort of convincing the players that the partner after alcohol will be different.

5. Player gets 80% - partner gets 20% against Each of the two gets 50%

6. More than 30% of players changed their decision in this pair after alcohol. Only 35% of them changed it to make an equal division, most changed to get more themselves. This is a significant result, as no player in control group changed his/her decision in this game.

7. This can mean that although the general desire for retaliation existed, it did not outweigh the inequality aversion for the control group. However, alcohol helped to magnify this effect, showing significant signs of retaliation in this pair too among the treatment group participants.

8. Player gets 80% - partner gets 20% against Player gets zero - partner gets 100%

9. Almost no player changed his/her decision here in neither treatment nor control group.

Acceptance/rejection: for each unfair offer in each pair (for example, player taking 80 to himself when an equal division was the other option) the proportion of rejections has been calculated. The resulting pattern turned out to be different from the one obtained by the European researchers:

instead of

The difference is that participants of the experiment described here found the offer in the pair with two equal options the least offensive while the most offensive being the one where player can claim more to himself/herself instead of an equal division. This sounds more logical to me too, so maybe this difference can be attributed to cultural differences.

However, the pattern remained the same after alcohol, so our hypothesis of insensitivity to strategic choice for the drunk people cannot be supported.

Survey

54 out of 60 participants, whose data was used in the experiment, filled in the survey, providing data about the sample used.

The age ranged from 17 to 27, with 15 females and 45 males. Only 12.7 percent of those having filled the survey drink heavy liquor at least once a week, while more than 49 percent drink heavy liquor less often than once a month. The drink of choice is wine, vodka taking the second place. 50.9% stated that 1-3 shots of heavy liquor is enough for them to feel moderate intoxication. A reminder: the range of doses was from 65 to 150 ml per person. The most popular reason for drinking being to support a company of friends (more than 92% stated this among the reasons), while festivals and stress relief also being reasons relevant for more than a half of participants. The mean number of shots a person drinks during a drinking event turned out to be 7.6.

Only 20% of participants stated that neither their work no study experience had ever been connected to economics. This may explain a large number of marginal decisions and sticking to the chosen strategy, as economists are more prone to considering their utility to be coming from only their own reward, with low responsiveness to circumstances, which are often considered a source of deviation from rational behavior.

Conclusion

The significant effect of alcohol can be observed in Envy and Impunity games, as well as in in-pair choice of strategic offers. Individuals feel more compassionate towards their partners, if the final outcome depends on their own choice only. Envy decreases after alcohol when own endowment is high. Partner's choice of having the maximum possible reward becomes more tolerated after drinking when endowment is low, but choosing to have just a bit more than the partner faces more reject than before alcohol when there is still a range of superior options. Players choose to be more humble in deciding their own reward when the partner's endowment is high, but not when his/her endowment is medium or low. In Generosity game players indeed become less sensitive to their own losses, but this conclusion does not have enough support across all endowment instances and Envy game results.

In strategic interaction, repetition serves as a source of retaliation motives, which become enhanced for those having consumed alcohol. However, the participants did not lose sensitivity to the left-out option presence in strategic interaction, neither have they become less consistent in their preferences.

The hypotheses stated before the experiment could not be confirmed in the given form, which reflects the pioneering nature of this work. We could not observe increased inconsistency in preferences, decreased influence of social norms, or indifference between degrees of inoffensiveness of offers. However, the results showed interesting nuances in behavior which were not accounted for before (f.e. individuals starting to accept equal division of reward more after alcohol, but only for when their endowment is high).

This work incorporated procedures of alcohol treatment never used before in an attempt to test the hypotheses never tested before. The results obtained helped to make clear the directions further research in the field can move into, and the experience it allowed to get can prove useful for ensuring finer outcomes of the subsequent studies. All papers which had been given to participants, as well as the survey questions, are included in the attachment for possible use and reference.

The following alterations can be made in future experiments to obtain more certain, significant and life-consistent results:

1) Increase in dosage - 0,5% BAC is the lower border for mild intoxication. More sever intoxication levels will provide more changes in decisions. According to the survey, the average dose a participant obtained during a drinking evening is about 3,5 times higher than the one consumed at the experiment.

2) Less people with economic background in the sample - due to economic games being a frequent part of the studies, economic students make their choices with a similar pattern, completing such games with results not actually representing their preferences and beliefs. Alternatively, a more real-life form of game can be used instead of tasks on paper.

3) Two days for each group - leaving at least a day-long gap between completing the first three sets of tasks and drinking alcohol and completing the rest of the tasks. This way, participants are less likely to rely on their memory when filling tasks after alcohol consumption.

4) Placebo effect incorporation - existence of non-alcoholic vodka leaves immense possibilities for placebo studies. Its incorporation will allow to develop the research of alcohol influence into numerous new branches.

alcohol economics decision making

References

1. Allen, A. J., Meda, S. A., Skudlarski, P., Calhoun, V. D., Astur, R., Ruopp, K. C. and Pearlson, G. D. (2009), Effects of Alcohol on Performance on a Distraction Task During Simulated Driving. Alcoholism: Clinical and Experimental Research, 33: 617-625.

2. James M. Bjork, Jodi M. Gilman, The effects of acute alcohol administration on the human brain: Insights from neuroimaging, Neuropharmacology, 2014, vol 84, pp. 101-110

3. GE Bolton, E Katok, R Zwick, Dictator game giving: Rules of fairness versus acts of kindness International journal of game theory, 1998, vol. 27, pp. 269-299

4. Dirk Breitmeier , Irina Seeland-Schulze, Hartmut Hecker & Udo Schneider,Clinical study:The influence of blood alcohol concentrations of around 0.03% on neuropsychological functions--a double-blind, placebo-controlled investigation, 2007 The Authors. Journal compilation, Society for the Study of Addiction

5. Daniel R. Burghart, Paul W. Glimcher, Stephanie C. Lazzaro, An expected utility maximizer walks into a bar…, Journal of Risk and Uncertainty, 2013, vol. 46 (3), pp. 215-246

6. Casal S., Ploner M. Guth W. Would you mind if I get more? An experimental study of the envy game Journal of Economic Behavior & Organization 2012 vol 84 (3) pp 857-865

7. Corazzini L, Filippin A, Vanin P Economic Behavior under the Influence of Alcohol: An Experiment on Time Preferences, Risk-Taking, and Altruism, 2015

8. Jodi M. Gilman, Ashley R. Smith, Vijay A. Ramchandani, Reza Momenan andDaniel W. Hommer, The effect of intravenous alcohol on the neural correlates of risky decision making in healthy social drinkers, Addiction Biology, 2012 vol 17 (2), pp. 465-478

9. Guth W., The Generosity Game and calibration of inequity aversion The Journal of Socio-Economics, 2010, vol. 39 (2), pp. 155-157

10. Guth W., Vittoria Levati M., Ploner M. An experimental study of the generosity game. Theory and decision, 2012, vol. 72, pp 51-63