Are public misperceptions of crime trends a cause of criminality? Evidence from a randomized experiment

ABSTRACT

Economists and psychologists have found that people tend to follow descriptive norms (what others do). Accordingly, the view that an increasing number of people engage in crime may increase the acceptability of criminal choices to individuals. Utilizing a randomized experiment embedded in an original online survey, this study explores people’s willingness to offend as a potential consequence of perceptions of rising crime trends, and if providing people with accurate crime information reduces criminal intentions. Results suggest that though perceptions of rising crime trends and willingness to offend are not directly related, exposure to accurate crime information reduces willingness to offend among respondents who initially perceive rising crime trend in the short term. The effect is not significant among respondents who already perceive downward crime trends in both short and long terms.

KEYWORDS:

• Perceptions of crime
• willingness to offend
• descriptive norms
• pluralistic ignorance

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. When the initial score in period 1 is set as 10, the lowest score one can get in the last period (period 6) is 0. As such, all crime trend perception scores are not negative and the lowest crime trend perception score is 0, making it easy to compare and understand crime trend perception scores in different time periods visually. If a different number other than 10 is chosen as the initial score (e.g., 0, 1, or 12), results derived the analyses will not be affected.

2. Crime trend perceptions estimates are censored by a minimum scale of 0 and a maximum scale of 20.

3. The dependent variables were not Normally distributed. For remedy, I used robust standard errors in the analyses. I also ran separate probit models (not shown), in which we dropped the middle groups of respondents who reported ‘neither likely nor unlikely’ to engage in delinquent behaviors and recoded the dependent variables into dummies. The results from OLS models and probit models were similar.

Additional information

Notes on contributors

Luzi Shi

Luzi Shi is a Ph.D. student in the School of Criminal Justice at the State University of New York at Albany. Her research interests center on public opinion and criminal justice. She is also interested in victimology, gender, survey research methods, and cross-national theory testing.