Influencing Youth Offenders: How Social Status Affects Cheating in Youth Corrections

How to cite

Zuo, Sharon Xuejing; Leong, Kaiwen; Li, Huailu. 2020. "Influencing Youth Offenders: How Social Status Affects Cheating in Youth Corrections." AEA RCT Registry. AEARCTR-0006658. https://doi.org/10.1257/rct.6658-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Sharon Xuejing Zuo* , Kaiwen Leong , Huailu Li

* Primary investigator

Registry ID
AEARCTR-0006658
DOI
https://doi.org/10.1257/rct.6658-1.0
First registered
Last updated
Status
Completed
Trial dates
2019-06-01 – 2019-11-30
Countries
Singapore ()
JEL codes
C93, K00, K14, K42, Z10
Registry record
http://www.socialscienceregistry.org/trials/6658

Abstract

We conducted an experiment with 204 youth inmates to study how the moral cost of cheating that was shaped by peers changed inmates’ cheating behavior. We find that innately dishonest inmates who naively revealed their higher willingness to cheat indeed cheated more in the actual game. When given the chance to observe an imperfect signal of whether a peer cheated, only innately dishonest inmates followed this signal and cheated more. This positive treatment effect increases with the saliency of the signal, and becomes more pronounced when the cheating signal is from an influential peer. No treatment effect was found for all inmates who observed cheating signals from non–influential peers.

Intervention

Our study was conducted at the Charlie Housing Unit (CHU) and Prison School where male youth offenders in Singapore are incarcerated. To measure each participant’s dishonesty, we followed Cohn et al. (2015) in asking them to play a coin toss game. Every participant was asked to toss a coin 10 times and to report the outcome after every toss. He will receive 5 reward points for every head he reports and 1 reward point for every tail he reports.We randomize (a) whether a participant sees a peer’s decision, and (b) the actual
number of heads his paired peer reported. This strategy allows us to identify the causal impact of peer effects through the cost of cheating.

Experimental design

Our randomization procedure follows standard experiment design. Within each community group, participants were pre–randomized into two groups: the treatment group (T) and the control group (C). In the treatment group, half of the participants were randomly chosen to play the game first (T1) and the remaining half were randomly chosen to play later (T2). Each participant in the T2 group was randomly paired with a participant in the T1 group. For the control group, participants were also randomly split into C1 and C2, in which C1 played first and C2 played later. But, both C1 and C2 did not have the chance to view anyone’s game results.

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
The randomization was performed at the community group level. Participants’ peer networks are mostly confined within the community groups. Half of the peers listed by participants are from their current community group.
Clusters
About 20 communities
Observations
About 200
Arms
About 100 subjects will be assigned into the treatment group and another 100 will be assigned into the control group.
Primary outcomes
Every participant was asked to toss a coin 10 times and to report the outcome after every toss. He will receive 5 reward points for every head he reports and 1 reward point for every tail he reports. So the number of heads reported by each participant will be our primary outcome.

Cited by

No citation data available for this registration yet.

Keywords