Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
Consent forms
General
Moscow
Oxford
Santiago
Erratum_Once_A_Liar.pdf
Once_a_liar_always_a_liar_.pdf
Protocol for Tax Experiments.pdf
README.md

README.md

Once A Liar 2018

Data, and Stata/R Replication Code for Duch Laroze Zakharov

Lying is a puzzling phenomenon. It prevalent on both a grand scale and in mundane, day-to-day interactions, but for many people there are intrinsic costs that prevent them from distorting information to one's favor. The goal of this study is to investigate how these costs depend on the magnitude to which the truth is distorted. We observe over 1000 individuals from the U.K., Russia and Chile making over 10000 lying decisions in a public goods game, while varying the benefit of lying. We find that the incidence and magnitude of lying do not depend on the benefits, which is not consistent with the marginal cost of lying being increasing in the size of the lie. Instead, we find that some subjects tend to be maximal liars with very low intrinsic lying costs, while some others lie up to a threshold that is not very sensitive to the extrinsic benefits of lying. We argue that maximal and partial lying are distinct phenomena. First, in two countries out of three lying is not highly conditional on the behavior of other individuals. Second, both ability at a real effort task and selfish behavior in the Dictator Game are strong and consistent predictors of maximal, but not partial, lying. Finally, the reaction time for a partial lying decision was much longer than for either a maximal lie or an honest declaration.

This replication package includes:

Data:

  • UK Sessions: [1,18]; [28, 33]; [40, 42]
  • Chile Sessions: [19, 27]; [34, 39]; [43, 46]
  • Russia Sessions: [47, 64]
  • For each of these sessions: z-tree code, doc instructions, xls data sets
  • Comprehensive STATA dataset

Code:

  • STATA code for data analysis
  • z-tree code for individual sessions

Instructions:

The instructions for the different treatments are included in each of the folders. Within each treatment there were different sessions that varied with regards to the level of taxation (from 10-40%) and audit rate (0-100%). The instructions for each of these variations only change with regards to the level of taxation and audit that was, correctly, informed. All the rest stayed the same. Note that for the Chilean sessions, instructors who ran the Shock treatments in Chile read out the wrong instructions, reading the Baseline instructions instead.

Manuscript:

  • once_a_liar_July_2018.pdf
  • TeX files
You can’t perform that action at this time.