Back to History Current Version

Overcommunication: The role of past experience

Last registered on October 13, 2020

Pre-Trial

Trial Information

General Information

Title
Overcommunication: The role of past experience
RCT ID
AEARCTR-0006387
Initial registration date
October 10, 2020

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
October 13, 2020, 9:20 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Region

Primary Investigator

Affiliation
University of Amsterdam & University of Birmingham

Other Primary Investigator(s)

Additional Trial Information

Status
In development
Start date
2020-09-30
End date
2020-11-30
Secondary IDs
Abstract
A typical finding in experimental cheap talk games is overcommunication. Senders are more truthful and receivers are more credulous that game theory predicts. I test whether overcommunication is observed because people are more familiar with common interest environments where such behavior is optimal. To this end, I use an experiment to establish causality between past experience and behavior in a new environment. Subjects play repeated sender-receiver games with random rematching in two stages of 30 rounds each. My design varies two dimensions, Past and Future. Past varies whether the first stage is with aligned preferences (b=0) or with conflicting (b=2). Future varies whether the environment in the second stage changes permanently (30 rounds with b=1 on stage two) or not (20 rounds with the same bias as first stage and 10 rounds with b=1 in random order on stage two). I compare behavior in the second stage across treatments.

Registration Citation

Citation
Ioannidis, Konstantinos. 2020. "Overcommunication: The role of past experience." AEA RCT Registry. October 13. https://doi.org/10.1257/rct.6387-1.0
Experimental Details

Interventions

Intervention(s)
My design varies two dimensions, Past and Future. Past varies whether the first stage is with aligned preferences (b=0) or with conflicting (b=2). Future varies whether the environment in the second stage changes permanently (30 rounds with b=1 on stage two) or not (20 rounds with the same bias as first stage and 10 rounds with b=1 in random order on stage two).
Intervention Start Date
2020-09-30
Intervention End Date
2020-11-30

Primary Outcomes

Primary Outcomes (end points)
Correlations between states and messages, messages and actions, and actions and states. The leading hypothesis is correlations during the second stage of the experiment between treatments.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Response times
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Subjects play a standard cheap talk game directly based on Crawford & Sobel (1982). The bias parameter takes three values (b can be either 0, 1 or 2) and is varied both between and within treatments. The experiment consists of two stages. I use two treatments in the first stage. In the Aligned-Mixed treatment, subjects play 30 rounds of a game with aligned preferences (b=0). In the Conflict-Mixed treatment, subjects play 30 rounds of a game with misaligned preferences (b=2). In the second stage of the experiment, all treatments play the same game with partial preference alignment (b=1).

After the main experiment, I elicit a few additional incentivized measures. I elicit risk aversion using the lottery choice task of Eckel & Grossman (2008) and cognitive ability using a modified version of the CRT by Frederick (2005). I also survey some information without incentives: demographics (age, gender, field of study) and attitudes towards trusting strangers using two questions adapted from World Value Survey.
Experimental Design Details
If an effect is established, I plan to run a control treatment where subjects only play the game with partial preference alignment (b=1). This Mixed-Mixed treatment will serve as a benchmark to compare with main treatments. The comparison will allow me to see whether the effect is attributed mostly to one of the experimental past experiences (Aligned or Conflict).

If the effect is persistent also in late rounds, the control treatment will allow to discuss equilibrium selection. If the two main treatments result in different end-round behavior, it is important to document what would the end-round behavior look like without the first stage of the experiment.

The states are randomly drawn before the experiment (8 random sequences, one per matching group) and I will use the same sequences for all treatments to avoid order effects.
Randomization Method
Randomization done by a computer.
Randomization Unit
Matching groups of 8
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
256 participants from 32 matching groups (16 sessions with 16 participants in each split in two matching groups)
Sample size: planned number of observations
128 observations (since each pair of two participants gives one observation) times 60 rounds = 7,680
Sample size (or number of clusters) by treatment arms
64 participants (8 matching groups) with Past = Aligned-Mixed and Future = Temporary
64 participants (8 matching groups) with Past = Conflict-Mixed and Future = Temporary
64 participants (8 matching groups) with Past = Aligned-Mixed and Future = Temporary
64 participants (8 matching groups) with Past = Conflict-Mixed and Future = Temporary
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Based on a pilot, the minimum detectable effect size is 0.13 (measured in differences in correlations between treatments).
IRB

Institutional Review Boards (IRBs)

IRB Name
Ethics Committee Economics and Business (EBEC), University of Amsterdam
IRB Approval Date
2020-09-07
IRB Approval Number
EC 20200907110931
Analysis Plan

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials