Trust in institutions and consequentiality perceptions in a stated preference survey

In total, we have three experimental conditions/groups: high trust, low trust, and control.

We use an open question to prime high trust in institutions. The question text is as follows:
"We ask you to recall a situation in which you dealt with authorities (such as the local government, the police, or the government) and in which you felt that they meant well, did their best to help you, and took your concerns seriously. In short, a situation where the authorities were trustworthy. In the box below, write down what you experienced and how you felt about this situation."

We use an open question to prime low trust in institutions. The question text is as follows:
"We ask you to recall a situation in which you dealt with authorities (such as the local government, the police, or the government) and in which you did not feel that these authorities meant well with you, did not do their best to help you, and did not take your concerns seriously. In short, a situation where the authorities were not trustworthy. In the box below, write down what you experienced and how you felt about this situation."

The control group does not answer an additional question.

2021-05-28

2021-06-21

Discrete choice experiment among two alternatives with four attributes: Type of registration (qualitative, 2 levels: automatic/non-automatic), Asset management (qualitative, 2 levels: state/private), Degree of compliance with EU taxonomy (quantitative, interval between 0% and 100%), Additional tax for your household (quantitative, 0.05% to 0.75% of annual household income).
1: The corresponding variable "choice" takes the value 1 for the chosen alternative and 0 for the alternative that was not chosen. We are mainly interested in the willingness to pay for the degree of compliance with EU taxonomy that is estimated using discrete choice models (especially mixed logit models) in willingness to pay space (cost attribute: additional tax for your household).

2: Policy consequentiality measured on an ordinal scale with the question "My answers in this survey will have an impact on whether or not the new voluntary pension scheme is adopted." with five ordered answer categories "do not agree at all", "rather do not agree", "undecided", "rather agree", and "strongly agree". The corresponding variable "policy consequentiality" thus takes the values 1, 2, ..., 5.

3: Payment consequentiality measured on an ordinal scale with the question "If it is decided to introduce the new voluntary pension scheme, I believe I will have to pay the corresponding tax for it." with five ordered answer categories "do not agree at all", "rather do not agree", "undecided", "rather agree", and "strongly agree". The corresponding variable "payment consequentiality" thus takes the values 1, 2, ..., 5.

4: Perceived policy consequentiality of other respondents measured with the question: "Please indicate the percentage of respondents to this survey who rather agreed or strongly agreed with the following statement: My answers in this survey will have an impact on whether or not the new voluntary pension scheme is adopted", with answers ranging from 0% to 100%. The corresponding variable "perceived policy consequentiality of others" thus ranges between 0 and 1.

5: Perceived payment consequentiality of other respondents measured with the question: "Please indicate the percentage of respondents to this survey who rather agreed or strongly agreed with the following statement: If it is decided to introduce the new voluntary pension scheme, I believe I will have to pay the corresponding tax for it", with answers ranging from 0% to 100%. The corresponding variable "perceived payment consequentiality of others" thus ranges between 0 and 1.

Manipulation checks to assess whether the trust priming treatments did work:
Word stem completion test for all three groups (control, low trust, high trust). Participants complete the three words (German/Polish) Ger____ / Spr____ ("Gerechtigkeit"/"Sprawiedliwość", Justice), Ehr___/Ucz___ ("Ehrlichkeit"/"Uczciwość", Honesty), and Miss___/Nie___ ("Misstrauen"/"Nieufność", Mistrust). For each group, we count the number of "correct" word completions, i.e. either the word as expected above or other forms (inflections/grammatical forms/combinations with other words/synonyms/…) of the word according to an official dictionary. We also correct for spelling errors.

We measure trust in scientists, trust in politicians, and trust in financial professionals on an ordinal scale with five answer categories ("Do not agree at all", "Do rather not agree", "I am undecided", "Do rather agree", "Fully agree") for all three groups after the treatments. The corresponding variables "trust in scientists", "trust in politicians", and "trust in financial professionals" thus take the values 1, 2, ..., 5, respectively.

We examine a discrete choice experiment that is embedded into an online survey. Prior to the discrete choice experiment, respondents are randomly assigned to three conditions (high trust priming treatment group, low trust priming treatment group, and control group) in each country. After the discrete choice experiment, respondents are asked for their own perception of how consequential they think their answers in the discrete choice experiment will be, and how they think other respondents in the survey answer to the consequentiality questions.

We base our analysis on a large framed field experiment, which will be implemented in broad representative online survey among the electorate in Germany and Poland during May and June 2021. The survey will be carried out by a professional market research institute. The surveyed population is defined as individuals that are at least 18 years old and have the right to vote in the general election in their respective country. Participants that do not fulfill any of these criteria will be screened out.

The experiment is embedded into the online survey described above. Prior to this experiment, respondents are randomly assigned to three conditions (see above) in each country, namely high trust priming treatment group, low trust priming treatment group, and control group. We employ a word stem completion test as a manipulation check after the treatments. Then, respondents face a discrete choice experiment where they have to choose between two private retirement schemes (one is always the current scheme in place, the other one is an alternative, modified scheme) twelve times. After the choice experiment, respondents are asked for their own perception of how consequential they think their answers in the discrete choice experiment will be, and how they think other respondents in the survey answer to the consequentiality questions (see above), whereby both questions are incentivized via the choice matching approach by Cvitanić et al. (2019, AER: Insights, 1(2), 179-192). Lastly, respondents are asked how much they trust different institutions (scientists, politicians, and financial professionals) in designing retirement schemes.

Respondents are randomly assigned to the three groups with equal probability and without their knowledge by the survey questionnaire software (simple randomization).

Unit of randomization for the trust priming treatments: individuals

No

2 countries (Germany and Poland)

2000 individuals overall, 1000 in Germany and Poland each

800 individuals in control (400 in each country), 600 individuals in high trust treatment (300 in each country), 600 individuals in low trust treatment (300 in each country)