Experimental Design
Our experiment will be conducted via phone surveys through a local survey company. The survey was programmed using the software SurveyCTO.
The protocol for the survey is as follows. Enumerators will start by explaining who they are and that they are conducting a short survey in line with the municipality’s recycling programme. As a first check, enumerators will ask whether the household is already enrolled in the programme, and if so, thank the respondent and end the survey, as we are interested in the households that are not yet part of the programme. If a household is not enrolled yet, enumerators will inform respondents about how the data will be used and that all their answers will be treated confidentially and analysed anonymously. Respondents will then be asked to give their verbal consent to participate in the study. Moreover, respondents will be informed that the survey will take less than 10 minutes and that they can win a prize (one of 15 Falabella gift cards of 50 Soles each) when participating in the study. If the respondent agrees to take part in the study, enumerators will start with the actual survey.
In the first step of the survey, we will elicit people’s pre-treatment beliefs about dynamic and injunctive norms regarding recycling in Miraflores as well as their personal first-order beliefs about the importance of recycling. The beliefs questions are incentivized in the sense that people will be able to win another Falabella gift card of 10 Soles each for each question where their belief will be accurate, in addition to the 50 Soles Falabella gift card they can win for their general participation.
Afterwards, we will collect demographic information about the respondent and the household in general as well as control questions. These general questions also aim to serve as buffer questions between the beliefs questions and the treatment messages, which both focus on the same information (for example, we aim to reduce the potential effect that respondents might still be thinking about the injunctive norm - as they were asked to make a guess about it - when receiving the dynamic norm information in the treatment message, or vice versa).
In the next step, the treatment messages will be conveyed, for which participants will be randomly assigned to one of the four treatment groups based on a 2x2 design, with injunctive norm information on the one side (no/yes) and dynamic norm information on the other side (no/yes). Depending on the treatment group, people will receive
- A (control group): no additional information
- B (injunctive norm treatment): information about how many people think it is important that households participate in the recycling programme
- C (dynamic norm treatment): information about how the number of households participating in the recycling programme has developed over time
- D (dynamic + injunctive norm treatment): information about how the number of households participating in the recycling programme has developed over time + how many people think it is important that households participate in the recycling programme.
The treatment messages will be directly followed by the question whether the household would like to sign up to the recycling programme. The sign-up decision during the survey will be measured as a binary variable (yes/no) and will be our first outcome variable.
After that, we will elicit people’s post-treatment beliefs about future norms in recycling behaviour as well as personal and collective response efficacy. Moreover, we will collect additional control variables that are related to recycling and will therefore be asked after the sign-up decision to not influence the respondents’ decisions and reduce potential experimenter demand effects.
Finally, we will collect the contact details for the household (in case the household wants to sign up to the programme) and inform respondents that they will receive an official registration link from the municipality within the following days. All participants will further be informed that they will be notified whether they won the Falabella gift cards after completion of the data collection.
Over the subsequent weeks following our initial survey, we will then observe whether a household actually signs up to the recycling programme through the official link of the municipality, which will be our second outcome variable (again measured as a binary variable). The official registration link will be sent via email or WhatsApp (depending on the participant’s preference) and will contain the treatment message again to reinforce its strength.
Our analysis will focus on average treatment effects as well as heterogeneous treatment effects as a result of individual level belief updating about dynamic and/or injunctive norms, depending on the treatment.