Status Gains versus Status Losses: Loss Aversion and Deviance

Abstract

The relationship between associating with (non)deviant peers and one’s own delinquent tendencies is often attributed to the motivation for positive reinforcement and status attainment. Guided by prospect theory and loss aversion, we assert that there is an alternative mechanism through which individuals conform to peer influence – to prevent loss of status for not conforming to the peer group. We surveyed over 1,200 college students at multiple universities across the United States and randomly provided them with hypothetical scenarios related to fighting, driving drunk, and using marijuana where the social consequences were framed as either gains or losses in status. Respondents reported a greater willingness to engage in both deviance and non-deviance when the social consequences were framed as status losses compared to status gains. Our findings are supportive of loss aversion and we advocate for further research that merges individual decision making and peer influence.

Keywords:

• Prospect theory
• peer influence
• loss aversion
• status

Acknowledgements

We would like to thank Derek Kreager, Tom Loughran, the anonymous reviewers, and Marvin Krohn for their helpful feedback on earlier versions of this manuscript. We would also like to thank Alex Nur and Brandy Parker (FG) for their assistance in data collection and entering.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Our point is that although we present loss aversion as a singular phenomenon, it manifests in human behavior in multiple ways.

2 There is an established literature detailing positive peer influence on a variety pro-social behavior (Berndt, 1979), such as better educational outcomes (Rambaran et al., 2017), and safer sex practices (Kapadia et al., 2012).

3 It was noted above that reference dependence is a decision making characteristic relevant for the current study. Our use of the terms “gains” and “losses” incorporates reference dependence by framing the consequence as changes from one’s current state, and it is thus embedded in our scenarios.

4 Providing respondents two behavioral options for each scenario is consistent with prior research in behavioral economics assessing loss aversion. We recognize, however, that this is not common in much of the vignette research in criminology. To mitigate this concern, we constructed our surveys so that the provided responses were realistically exhaustive in the scenarios. Further, during piloting process we solicited reactions to the response options and they reported the responses reflected realistic and exhaustive reactions to the scenarios.

5 We used different vignettes in the “deviant” peer and “conforming” peer conditions simply because we could not think of a scenario where friends would realistically both encourage deviance and pro-social behavior. We stress that our use of different scenarios does not affect the central research question of the current study, as our analyses examine willingness to offend within the deviant and conforming scenarios. That is, we assess the impact of framing social consequences as losses versus gains among those who receive the scenario where friends promote driving intoxicated and a separate analysis among those who receive the scenario where friends discourage driving under the influence. See the appendix for the full vignettes.

6 Recreational marijuana use is legal for those 21 years or older in one of the states in which we conducted the survey. The overwhelming majority of respondents at this university were under the age of 21. We conducted supplemental analyses remove those 21 years of age or older and the results are substantively the same as those presented in text.

7 Impulsivity was measured through the mean of three items: “I often act spur of the moment without stopping to think;” “I don’t devote much thought and effort for preparing for the future;” and “I often do what brings me pleasure here and now, even at the cost of some distant goal.” Importance of friends was captured through the single item: “I care what my friends think about me.” All items were on a 4-point Likert scale.

8 The randomization process should, in theory, balance characteristics that are unobservable, as well. For example, research subjects may have different experiences in the past with regards to the extent at which peers have encouraged deviance. Such differences should not bias our results, as we can be relatively confident that prior experiences are equally distributed across groups.

9 None of the individual predictors were statistically different at p < .05; however, some approached statistical significance. In the conforming-fighting conditions those with loss frames were more likely to report a prior tendency to fight (.20 versus .12, p = .16); in the conforming-drunk driving conditions individuals in the loss frame were more likely to be male (.48 versus .39, p = .20); and in the deviant-marijuana conditions those receiving a loss frame were slightly more likely to use marijuana in the past (.67 versus .57, p = .17). In each case we estimated logit models controlling for these covariates and the results are substantively and statistically the same as those presented in text.