Trust, institutions, and the welfare state

Last registered on July 07, 2022

Pre-Trial

Trial Information

General Information

Title
Trust, institutions, and the welfare state
RCT ID
AEARCTR-0009008
Initial registration date
March 29, 2022

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
March 31, 2022, 3:18 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated
July 07, 2022, 7:15 PM EDT

Last updated is the most recent time when changes to the trial's registration were published.

Locations

Region

Primary Investigator

Affiliation
UNSW

Other Primary Investigator(s)

PI Affiliation
UNSW
PI Affiliation
Magdalen College, University of Oxford

Additional Trial Information

Status
On going
Start date
2022-04-01
End date
2022-12-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
One of the most important functions of democracy is to aggregate preferences and information towards the choice of policies that deliver sound economic security in the long run. However, democracies that share similar institutions exhibit very different results in terms of the perceived ability of their governments to effectively offer economic security. Why, for example, do two similarly designed social democracies like Sweden and Italy differ so much in the implementation of their welfare plans? Why is the role of the federal state in guaranteeing economic security in the United States much smaller than in Germany? More broadly, what is the role played by inherited values and social norms—what we may collectively refer to as "culture"—in determining the functioning of institutions? And if culture affects the working of democratic institutions, should different cultures adopt different “versions” of democracy?

This project aims to explore the relationship between voters’ perception of the quality of democratic policy making processes and their support for long-term ambitious policies. In particular, the project will provide insights into how voters' perception of the efficacy of electoral accountability and their ability to select honest politicians affect their support for long-term and hard-to-verify policies. In particular, we are interested in studying whether voters with lower trust in politicians are more likely to vote for politicians that promise short-term and directly verifiable policies and whether voters with lower opinions of the quality of the media and institutions of electoral accountability are more likely to vote for politicians that promise short-term and directly verifiable policies.
External Link(s)

Registration Citation

Citation
Gratton, Gabriele, Barton Lee and Hasin Yousaf. 2022. "Trust, institutions, and the welfare state." AEA RCT Registry. July 07. https://doi.org/10.1257/rct.9008-1.1
Experimental Details

Interventions

Intervention(s)
Please contact Principal Investigators for more information.
Intervention (Hidden)
This project uses a quantitative research methodology that includes surveys and questionnaires for data collection. To test our hypothesis, we will perform a large online survey experiment across the whole United States. United States is an ideal country to test our hypothesis because it has a large and diverse voting age population. In addition, the results of our findings will be easy to compare to numerous studies that predominantly use United States population.

The survey questions have been developed based on leading surveys in political economy such as Cooperative Congressional Election Study, American National Election Studies, and National Annenberg Election Survey.

We will use survey questions to elicit subjects' beliefs (both quantitative and qualitative) about the level of honesty among politicians in general and their preferences for different policies. It is an important concern that correlations between beliefs and preferences may reflect other beliefs about politics that may affect both trust in politicians and political preferences. To overcome this potential concern, we will experimentally vary the trust in politicians among voters using methodology similar to Alesina, Stantcheva and Teso (2018) and Alesina, Miano and Stantcheva (2020).

In particular, we will use the Corruption Perception Index to provide participants with the level of corruption in a representative list of countries around the world (e.g., Tunisia, Spain, Uruguay, and Denmark). We will ask participants to place on the scale what they believe is the level of corruption in the United States. After answering this question, we will provide the “treatment” group with the actual level of corruption in the United States as calculated by the Corruption Perception Index. To the “control group” we will not provide any information about the level of corruption in the United States.

We will test whether our survey questions and methods are sound in the pilot. The main aims of the pilot are two-folds. First, we will use the pilot to determine the magnitude of the treatment effect induced by our information treatment to find the appropriate sample size to detect the effects on political preferences. Second, we will test different versions of the question asking voters whether they trust politicians to see (i) whether these questions are appropriately measuring the trust level in politicians, and (ii) which question is better at measuring the trust levels.
Intervention Start Date
2022-04-01
Intervention End Date
2022-12-31

Primary Outcomes

Primary Outcomes (end points)
Political Preference for Policy making with various degree of ambitiousness and verifiability
Primary Outcomes (explanation)
Respondents will be asked several policy questions to respond whether they would like to see numerous policy carried out (asked one by one). The respondents will reply to these questions on a scale of 1 (Definitely Not) to 5 (Definitely Yes), with Probably Not, Maybe/Not sure and Probably Yes (2 to 4) being the other options. The policies are (weakly) ordered from less to more ambitious policies:

$100 million of Federal funds to upgrade local transportation in your state
$20 million more Federal funds to support local industries in your area
Renew Congress' efforts for a durable peace in the Middle East
Renew Congress’ effort to advance women’s rights in the developing world
Simplify the tax code and reduce loopholes
Increase funding to the U.S. Fire Administration
No new useless federal agencies

For each policy, we would ask five set of questions that would help us gauge (i) support for the policy, (ii) policy is executable, (iii) support for candidate proposing the policy, (iv) likelihood that the policy could be feasibly implemented, and (v) likelihood that voters will find out whether policy is implemented. In particular, we will ask how much they agree with the following five statements after each policy:

I would LIKE this policy to be carried out.
A politician who SAYS they want to carry out this policy would actually TRY to.
A politician who TRIES to carry out this policy would be ABLE to.
It would be easy for me to obtain timely information (including from media and other sources) on the progress of this policy.
I would support a politician who says they want to carry out this policy.

We will also elicit support for policy before treatment.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Please contact Principal Investigators for more information.
Experimental Design Details
Data will be collected using an online survey administered through the Qualtrics platform by a reputable data collection organization, Respondi. Randomization is generated by using ``Randomizer'' in Qualtrics software. The unit of randomization is individual. Since in our treatment individuals can under or over-estimate the level of corruption in the United States (discussed in detail below), there will be practically three treatment arms: respondents who over-estimate corruption, respondents who under-estimate corruption, and control group (who are not informed about true corruption levels). Assuming individuals are equally likely to over-estimate and under-estimate corruption, we plan to have equal number of respondents in the three categories. That is, we plan to have treatment to control ratio of 2:1. That is, we intend to have 2,000 respondents in treated group and 1,000 respondents in control group. Our pilot will reveal whether the assumption of respondents equally over or under-estimating corruption is accurate or not. Depending on the findings, we may alter the treatment to control ratio accordingly.

Both treatment and control groups will be asked identical questions. The two groups will differ only whether they find out the actual answer to the level of corruption in United States.
In particular, we will use the Corruption Perception Index to provide all participants with the level of corruption in a representative list of countries around the world (e.g., Tunisia, Spain, Uruguay, and Denmark). We will ask all participants to place on the scale what they believe is the level of corruption in the United States. After answering this question, we will provide only the “treatment” group with the actual level of corruption in the United States as calculated by the Corruption Perception Index. To the “control group” we will not provide any information about the level of corruption in the United States. We will also perform a post-treatment "check" to see if people thought their guesses were over/underestimates of true level of corruption.
Randomization Method
Randomization is generated by using "Randomizer" in Qualtrics software.
Randomization Unit
The unit of randomization is individual.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
We plan to survey 3,000 respondents across United States. Our sample does not intend to limit to particular counties, districts or states. Instead, the sample will be representative of the U.S. voting age population, and will be randomly assigned into control and treatment.
Sample size: planned number of observations
We plan to survey 3,000 respondents across United States.
Sample size (or number of clusters) by treatment arms
We plan to have 2,000 respondents in treated group and 1,000 respondents in control group.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The main outcome variables in our sample are measured on a Likert scale (1 to 5). The variables are expected to have a mean of 3 with a standard deviation of 1.5. Based on our sample size of 3,000, power of 0.9, and treatment to control ratio of 2:1, the minimum detectable effect size we are able to detect with our sample is 0.1758.
Supporting Documents and Materials

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
IRB

Institutional Review Boards (IRBs)

IRB Name
University of New South Wales Human Research Ethics & Clinical Trials Governance
IRB Approval Date
2021-10-15
IRB Approval Number
HC210761
Analysis Plan

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials