Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
Partisanship, Information, and Policy Preferences
Last registered on October 30, 2020


Trial Information
General Information
Partisanship, Information, and Policy Preferences
Initial registration date
October 29, 2020
Last updated
October 30, 2020 9:27 AM EDT
Primary Investigator
Harvard Business School
Other Primary Investigator(s)
PI Affiliation
Harvard Business School
Additional Trial Information
In development
Start date
End date
Secondary IDs
Since the early 2000s, political polarization has skyrocketed among US voters, who are now more divided than ever on a diverse array of social, cultural, and economic issues (Abramowitz, 2018; Gentzkow, 2016; Klein, 2020). Political identity is powerful enough to govern the behavior and the preferences of Americans on both sides of the political spectrum. Recent studies have shown that, due to partisan identification, voters may support policies that oppose their own economic interests (Bonomi et al., 2020; Dokshin, 2020; Grossman and Helpman, 2019). This paper will test whether political identity distorts voters’ responses to new information in the context of healthcare policies in the US. We will conduct an online survey experiment in which we cross-randomize exposure to i) an information treatment on life expectancy in the US (compared to other countries, over time) and ii) a partisan prime and document how these two forces influence voter preferences, beliefs, and values.
External Link(s)
Registration Citation
Tabellini, Marco and Jaya Wen. 2020. "Partisanship, Information, and Policy Preferences." AEA RCT Registry. October 30. https://doi.org/10.1257/rct.6665-1.0.
Experimental Details
We will implement two cross-cutting interventions:
i) An information treatment revealing the performance of U.S. life expectancy over time, compared to other countries
ii) A series of questions designed to prime partisan identity
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
The outcomes of interest include policy preferences, values and beliefs, and political attitudes.

For policy preferences, we will elicit whether individuals change their support for various public policies that may influence life expectancy. These include health care, redistribution, law enforcement, firearm, and additional policies. We will also elicit directly individuals’ support for a “public option”, which would compete with private insurance plans, as well as a “national health plan”, which would eliminate the private health insurance market.

For values and beliefs, we will elicit a battery of World Values Survey questions about interpersonal trust, locus of control, altruism, and more. We will also elicit how much individuals believe life expectancy is due to luck versus personal choices. Finally, we will assess moral universalism (Enke et al, 2019; 2020) by asking respondents to split a hypothetical $100 between a randomly selected American vs. a randomly selected American Democrat. Then, we ask them to perform the same exercise for a randomly selected American Republican.

For political attitudes, we will ask where individuals place themselves on a liberal/conservative scale, as well as how much they approve of Donald Trump’s performance as president. We will also ask individuals which presidential candidate they most supported in 2016 and 2020.

In addition to the comparisons we outlined in the abstract, we will explore heterogeneity in the treatment effect along the following dimensions:
The difference between respondent priors and reality
Party affiliation and intensity of affiliation
Level of education
Media consumption
Political participation
Insurance status
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
Our experimental design involves two cross-cutting treatment groups, one with two arms, and one with three. In total, we will therefore have six unique treatment cells. Individuals will be randomized into one of these six cells, with randomization stratified by gender and race.
All individuals will take a 15-minute online survey. No identifying information will be collected.
The first group of treatment arms, which we call “PA”, varies whether or not we prime individuals with their party affiliation. In treatment arm PA = 0, we do not prime individuals, and in PA = 1, we do. The prime takes the form of two survey questions, which asks respondents about their affiliation and remind them of the efficacy of partisan affiliation as a lens of political decisions.
The second group of treatment arms, which we call “INFO”, varies whether or not respondents receive information about the United States’ true performance in the dimension of life expectancy compared to the other G7 countries (Canada, France, Germany, Italy, Japan, and the United Kingdom). For INFO = 0, no information is given. For INFO = 1, national averages are given. And for INFO = 2, respondents see both the US average and an own-demographic average (based on the respondent’s gender and race).
The order of the survey will proceed as follows. It will begin with the elicitation of consent and priors about where the United States falls in life expectancy rank among the G7. Then, for PA = 1, we will administer the partisan prime. Afterward, we will implement the information treatments. Finally, we will ask four survey modules of all participants: policy preferences, beliefs, behaviors, and demographics. We will randomly alternate the order in which we elicit beliefs over the determinants of life expectancy and policy preferences. This randomization allows us to measure the importance of mechanism priming in determining policy preferences.
Experimental Design Details
Not available
Randomization Method
We will use the randomization feature of the Qualtrics survey platform to assign treatment groups.
Randomization Unit
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
3,000 indvidiuals
Sample size: planned number of observations
3,000 indvidiuals
Sample size (or number of clusters) by treatment arms
500 individuals in each of the six treatment groups.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB Name
IRB Approval Date
IRB Approval Number