Experimental Design
Participants:
The panel provider Cint will help us collect a representative sample of 1’000 adults living in the German-speaking part of Switzerland to our online experiment. At the start of the survey, we will ask participants about their age, their home canton and their citizenship status, in case there was an error in the recruiting process conducted by the panel provider. We include subjects who are at least 18 years old, who live in a primarily German-speaking canton and who possess Swiss citizenship. Then, we conduct an attention check – following Oppenheimer et al. (2009) – to screen out subjects that are not reading or following our instructions diligently. In a next step, we will ask subjects whether they were ever incarcerated in their life. As our treatment (for details, see below) might trigger psychological distress for subjects who were previously incarcerated, we do not want to expose such subjects to the statements of real inmates for ethical reasons. In contrast, we will consider these subjects as an observational group that will not be randomized into the treatment arm of the information experiment and consequently not be included in the estimation of treatment effects. For explorative purposes, we will compare the answers of these subjects with those that were never incarcerated and who are randomized into the control group (for details, see below).
Treatments:
For subjects who were never incarcerated, we will randomly assign them into a treatment and a control group. Subjects in the treatment group will be presented 5 randomly selected and anonymized statements from real prison inmates. The statements were collected by surveying inmates in a prison in the canton of Zurich, Switzerland. More specifically, the statements represent the prisoners’ answers to the question: “What situations are causing you the most stress right now?”. To expose our study participants to a more representative impression of prison experience, we tried to include as many inmates’ statements into the treatment as possible. We excluded any statements with less than five words, since these were not informative enough to provide any insight into prison conditions. We translated statements from six languages (English, French, Italian, Portuguese, Albanian and Serbian) into German, minimally adjusted these with regards to spelling and capitalization, and screened out statements where inmates hinted at or mentioned their own nationality, since this could cause resentments from survey respondents against those nationalities. From the curated pool of statements each subject in the treatment group will be shown statements from 5 randomly selected inmates and for each statement they will rate the wellbeing of the inmates to ensure the subjects are engaging with the statements. Subjects in the control group will not be presented any statements.
Following the experimental manipulation, all participants will then be answering survey questions about their trust in institutions and their perception of procedural fairness.
Afterwards, we will elicit subjects’ preferences for law and order (first primary outcome). For this, we will first collect a behavioral measure of attitudes towards law and order by presenting participants with an incentivized donation task with political organizations as beneficiaries. In this task, subjects have the chance to donate to one of two organization: one that is advocating for harsher punishment of crime and one that is advocating against harsh punishment. Such a donation task allows to cleanly elicit subjects’ attitudes towards detention conditions in an incentivized way. Apart from the donation task, we will elicit subjects’ preferences for law and order with a survey question on how strongly they support strict law enforcement and harsh sentencing.
In a next step, we will study the weight subjects’ attach to different motives of incarceration (secondary outcome). For this, we will instruct them to think about their answer to the previous question on how strongly they support strict law enforcement and harsh sentencing. Then, we will ask them how much weight they assigned to each of the following five motives: (i) punishment/restoring justice, (ii) general deterrence, (iii) specific deterrence, (iv) incapacitation, and (v) rehabilitation. Moreover, we will elicit subjects’ support for criminal justice reforms to i) improve incarceration conditions, ii) use alternative sanctioning forms, and (iii) provide more rehabilitation programs. on (secondary outcome). This will allow to validate their preferences for law and order (first primary outcome) by running correlational analysis and comparing this to subjects’ support for these policy proposals.
The survey proceeds with the elicitation of subjects’ beliefs about subjective wellbeing during incarceration. We will implement an incentivized guessing task to get a behavioral measure of their beliefs. Subjects will be told that we asked actual inmates in a Swiss Prison about their wellbeing on a 11-point Likert-scale from 0 (extremely bad) to 10 (extremely good). Subjects will then be instructed to estimate the inmates’ average response. The three most accurate answers will receive a CHF 50 voucher for an online store. Apart from the guessing task, we will elicit subjects’ beliefs with a survey question on their expected hypothetical wellbeing if they were to be incarcerated on the next day for 6 months.
Finally, we administer a 4-item version of the interpersonal reactivity index to obtain a measure of trait empathy. With this empathy index we explore whether the treatment has a differential effect on participants who score above the median in terms of their trait empathy.
The survey concludes with questions on subjects’ risk preferences, political tendencies, criminal identity and some demographic information. To control for possible demand effects, we include the social desirability-gamma short scale by Niessen et al. (2019) and finish the survey with an open question regarding what subjects think the purpose of this study was.