Experimental Design Details
We run laboratory experiments. Subjects in the role of “Choosers” have to make different types of decisions (in random order): i) a decision between options which differ with respect to the point of time when money payments are received: either immediately or in six months, where impatience is designed to be costly (intertemporal decision). ii) a decision between options that differ in the distribution of money payments to a charity aiming to reduce Co2 emissions and to oneself (social decision). Subjects in the role of "Choice Architects" are informed about the decision types and decision options of Choosers and decide which choice options are available to Choosers in future experiments. Choice Architects are provided three possibilities for their paternalism decisions: i) no intervention and leaving the full choice set to Choosers (no paternalism), ii) recommend against options that are available to Choosers (soft paternalism), iii) forbid single options and restricting the choice set of Choosers (hard paternalism).
The experimental part that involves Choice Architects has four stages and all Choice Architects go through all four stages. The second stage differs across (between-subjects) treatment arms.
Stage 1: Choice Architects make individual paternalism decisions.
Stage 2: Choice Architects make paternalism decisions (same ones as in stage 1)). The way the paternalism decision is made in stage 2 varies across four treatment arms. See below for more details on the treatment arms.
Stage 3: Belief elicitation: Participants are asked to guess how people make their intertemporal and social decision in situations with unrestricted choice set. The belief elicitation is incentivized.
Stage 4: Delegation: In a final stage, participants are asked whether they want their own decision from Stage 1 to be replaced by the decision of a random other participant.
Subsequently and before finishing the experiment, subjects fill out a questionnaire which includes questions on: standard demographics, views on real-world paternalistic problems (e.g., sugar tax), 'paternalistic' behavior of subjects in the real world (i.e., if they are engaged in political parties, student body groups, etc.), political attitudes, and engagement in the protection of the environment and climate.
The between-subject treatment variation in stage 2 is as follows:
Exogeneous Individual: Paternalism decision is made individually
Exogenous Group: Two subjects are randomly assigned to form a group. The two group members communicate via chat and have to find a group decision. The consented paternalism decision of the group is then implemented.
Endogenous Individual: Subjects are asked whether they want to make an individual paternalism decision in stage 2. If yes: they make another individual paternalism decision; If no: they move to stage 3.
Endogenous Group: Subjects are asked whether they want to make the paternalism decision in stage 2 as part of a group. If yes: they make a paternalism decision as part of a group; If no: they move to stage 3. The motivation behind this treatment is twofold: i) study how many and which types of subjects select themselves into group decisions, ii) study if endogenous groups differ in paternalistic behavior relative to exogeneous groups and individual decision-making.
We thus have a fully crossed 2x2 design with dimensions exogenous vs endogenous and group vs individual. Variation is both between and across experimental sessions. We vary dimension individual vs group within experimental session. Dimension exogenous vs endogenous is varied across experimental sessions.
Payment: Choice Architects receive a flat fee of 10 EUR for participating in the experiment and, in addition, receive a bonus in the context of the belief elicitation in stage 3.
Analysis: We compare the paternalism decisions in stage 2 across the four treatment arms. In this context, we examine if subjects intervene at all (dummy indicating either soft or hard paternalism vs no intervention) and we explore how exactly they intervene (3-point-scale variable capturing no intervention, soft paternalism, hard paternalism). We also compare stage 2 decisions to stage 1 decisions. In addition, we consider the treatments Endogenous and study how many and which types of subjects select to make paternalism decisions and how many and which types of subjects select into groups to make paternalism decision. The Delegation decision in stage 4 is also of interest for us (e.g., how many and which types of subjects make use of delegation). In additional analyses, we study the role of beliefs about Choosers behavior for paternalism and we relate decisions in the experiment to the answers in the questionnaire at the end of the experiment (for example to study heterogeneity and to link the paternalism decisions to real-world behavior).
In addition to the above described experimental setting, we will collect observations about intertemporal and social decisions of Choosers with unrestricted Choice Sets. These observations are used for incentivizing Choice Architects in stage 3 and to collect information about behavior of Choosers in a situation with unrestricted choice set.
In a reasonable amount of time after the experiments involving Choice Architects we run sessions where Choosers make intertemporal and social decisions where, in the available choice sets of Choosers, we implement the decisions of Choice Architects.