6. Investigating Honesty Flashcards
Why is honesty economically relevant?
Because many situations are characterised by asymmetric info
Describe the set up of die in a cup task. (Fischbacher & Follmi- Heusi 2013)
Treatments
Baseline
Control treatments
- stakes: 3x payouts in baseline
- externality: another subject receives 5- claimed amount
- double anonymity: subjects take claimed amount out of an envelope and return the envelope with the rest of the money to an urn
Repetition
Beliefs
Results of Fischbacher & Follmi-Heusi 2013
Baseline
0= 6.4%
1= 7.2%
2= 11.6%
3= 12.6%
4= 27.2%
5= 35.0%
-People are marginally more honest in higher stakes treatment.
-People are more honest in externality treatment
-No effects of double anonymity
Lying aversion
People don’t like the feeling of lying. They might balance the feelings of guilt with the benefit of lying
Lies in disguise
If people care about the credibility of the lie, they might not choose 5. People might also refrain from lying to maintain a positive self concept
Describe set up of Cohn, Marechal, Fehr 2014
-Swiss bankers asked to toss a coin ten times and to report number of heads which pays $20
-treatment makes banker identity salient, control doesn’t
Results of Cohn, Marechal, Fehr 2014
-Bank employees in the professional identity condition reported 58.2% successful coin flips which is significantly above chance and significantly higher than the reported success in the control group.
-Bankers are predicted as less honest than general public.
-Results suggest the prevailing business culture in the banking industry favours dishonest behaviour
Describe set up of Mazar Et al 2008
-Matrix task: find pair of numbers that sum up to 10
-Control: do task for 5 mins then show answers to experimenter
-Recycle: do task for 5 mins but answer sheets are shredded so experimenter can’t see the results
-Recycle and honour code: same as recycle but sheet says “I understand that this short survey falls under MIT’s honour system”
Results of Mazar Et al 2008
-There is only limited cheating since people want to maintain self image of being an honest person.
-No cheating in recycle and honour code
-Slightly more lying when higher stakes are used. ($5 vs $2.50)
Describe the set up of Gneezy 2005
-P1 has private info about payoffs and P2 chooses between actions A or B. Message from P1 procedes action but only actions determine payoffs.
-two actions, A and B, and two messages:
Message A “Option A will earn you more money than option B”
Message B “ Option B will earn you more money than option A”
Results of Gneezy 2005
-P1 lies 52% of the time when there is a high profit to be gained from lying
-P1 lies 17% of the time when there is a low profit to be gained from lying and P2 loses a lot
-P1 lies 36% of the time when low profit gain but low loss for P2
What do Abeler, Nosenzo, Raymond find in meta analysis on using die roll paradigm?
90 papers, 44,000+ subjects, 47 countries
Level of incentives don’t matter
Describe set up of Gerlach 2019
Meta analysis comparing all methods of honesty (sender-receiver, coin flip, die roll, matrix)
565 experiments, 44,000+ participants
Results of Gerlach 2019
-roughly 50% rate of liars across all methods
-Highest rate in sender receiver, lowest in coin flip
-men appear to lie 4% more often than women
-younger people lie slightly more
-more lying happens in lab when compared to field
Describe the set up of Pruckner & Sausbruger 2013
Newspapers sold in Austria by honour system
3 treatments
- paper costs €0.60
- paper costs €0.60. Stealing a paper is illegal
- paper costs €0.60. Thank you for being honest
Results of Pruckner & Sausbruger 2013
2/3 of people don’t pay and most pay less than €0.60.
Legal appeal has no effect, moral appeal increases payments
Describe set up of Cohn & Marechal 2018
Students often misbehave. Can a lab measure predict it?
-162 Swiss middle and high school students (12-20, 43% female)
-students receive envelope with 10 50p coins. Asked to toss each coin in private and allowed to keep head tosses, put tails back into envelope.
-teachers evaluate the students on following questions
1. Is the student often absent?
2. Does the student rarely complete his/her homework?
3. Is the student frequently disruptive in class?
Results of Cohn & Marechal 2018
Students took 62.8% of the coins- 25% cheating rate
Cheating is positively related to teachers evaluation of misconduct and statistically significantly across all 3 evaluations
Describe the set up of Dai Et al 2018
-244 participants recruited in Lyon train station and asked to participate in research study on transport behaviour if they had no monthly pass. 35 subjects recruited from fine collection office. Payment in gift vouchers.
-at the end, participants were asked to exchange their ticket (if they had one) for a valid one
-study details
1. Elicitation of attitudes towards risks and losses
2. Measure of honesty using a version of die roll experiment
3. Measurement of dishonesty in a “public transportation game”
4. Final questionnaire
In the Dai Et al 2018 study describe the public transport game
-decision to buy a ticket or not for an imaginary bus
-for every possible final stop (stops 1-7) subjects have to say whether they buy ticket or not. One decision randomly selected from payment.
-price of ticket €1.80
-benefits either increased by €2 per stop, expected benefit €8. Or were constant €8.
In Dai Et al 2018, how many people didn’t possess a valid ticket and how many self reported fare dodging at least 1 out of last 10 trips?
41.8% don’t have ticket
54.9% fare dodging at least 1 out of last 10 trips
Results from die in a cup in Dai Et al 2018?
-no ticket people lie significantly more than those with tickets
-subjects who have just paid a fine at collection office are more honest
Results from public transport game in Dai Et al 2018?
-self reported fraudsters commit ticket fraud more than self reported nonfraudsters regardless of whether either have a ticket.
-36% is the mean rate of ticket fraud
-those from fine collection office commit more fraud than average
What does the experiment with 192 students in Lyon lab show?
Subjects are overpaid. Those who don’t return the money are also more likely to lie in die in a cup.
Results support robustness and generalisability of previous findings
What is an important foundational question for all applied research?
What can we learn from stylised lab experiments which are often guided by game theoretical models? Are field studies more “realistic”?
What is the hallmark of all experimentation?
Controlled variation. The only important issue is how to achieve the best causal identification. Distinction between lab/field is less important
Why is the die rolled twice in the die in a cup experiment?
So that the subject has credible deniability. If we ask them to roll once and we then pick up the cups we can see what they rolled
What does partial dishonesty predict in die in a cup?
Higher reports are more likely than lower reports
What can the number of reports of 6 tell us in the die in the cup experiment?
The degree of honesty
Which papers are relevant to a question on honesty?
-Fischbacher & Follmi-Heusi (2013): die in a cup
-Cohn Et Al (2014): Swiss bankers
-Mazar Et Al (2008): matrix
-Gneezy (2005): sender receiver
-Abeler et Al- meta analysis on die roll
-Gerlach et Al (2019): meta on honesty methods
-Pruckner & Sausbruger (2013): newspapers
-Cohn Et Al (2018): coin toss with students
-Dai Et Al (2018): public transport France