Affective Polarization and Long-Term Partnership

How to cite

Woo, Jaemin; Adams, Camilla. 2026. "Affective Polarization and Long-Term Partnership." AEA RCT Registry. AEARCTR-0018541. https://doi.org/10.1257/rct.18541-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Jaemin Woo* , Camilla Adams

* Primary investigator

Registry ID
AEARCTR-0018541
DOI
https://doi.org/10.1257/rct.18541-1.0
First registered
Last updated
Status
In development
Trial dates
2026-05-04 – 2026-12-31
Countries
United States of America ()
JEL codes
D10, J12, J16, D72
Secondary IDs
AsPredicted #288515; Brown IRB STUDY00000970
Registry record
http://www.socialscienceregistry.org/trials/18541

Abstract

We study how partisan aversion shapes long-term partnership formation in the United States. We field a discrete choice experiment on approximately 1,200 partisans aged 25-45 in which party identity, education, income, attractiveness, religiosity, and family plans vary independently across hypothetical partner profiles. The experiment measures willingness to pay for each attribute, with primary focus on cross-party avoidance. We embed the experimentally identified preferences in a transferable-utility matching model following Choo and Siow (2006), calibrated to observed population type masses and couple shares. The experimental design has two waves. Wave 1 fields profiles drawn from uniform marginals over each attribute. Wave 2 fields profiles selected per task by Bayesian adaptive design, using the Wave 1 hierarchical Bayes posterior as the Wave 2 prior. We use the calibrated model to compute counterfactual singlehood rates and cross-party couple shares under (i) removing partisan aversion and (ii) rebalancing partisan composition within gender × education cells.

Intervention

See experimental design.

Experimental design

We use a discrete choice experiment to estimate willingness-to-pay (WTP) for partner attributes among US adults aged 25-45 recruited through Prolific. We seek to estimate WTP for cross-party partnership separately by gender, party, and education at the BA threshold. We embed the estimated preferences in a transferable-utility matching model and compute counterfactual singlehood and cross-party couple shares.

Design details

Randomization method
Randomization is performed by computer. In Wave 1, both forced-choice and accept-or-reject task designs are drawn at random over the design space. In Wave 2, both blocks use a Bayesian adaptive algorithm (BACE) that selects each task to maximize expected mutual information about respondent-specific preference parameters given the running posterior. The display order of the six attributes within a profile is randomized at the respondent level. Block order is fixed (forced choice first, accept-or-reject second).
Randomization unit
Individual.
Clusters
NA
Observations
~1,200 respondents
Arms
~300 respondents in each of four cells defined by gender × education at the BA threshold.
Primary outcomes
1. Willingness to pay for cross-party avoidance, expressed in USD of partner annual income, estimated separately for the eight cells defined by gender × party × education at the BA threshold. 2. Counterfactual change in the singlehood rate when partisan aversion is removed, relative to baseline equilibrium. 3. Counterfactual change in the cross-party couple share when partisan aversion is removed, relative to baseline equilibrium.
Secondary outcomes
1. Willingness to pay for the remaining attributes (education, attractiveness, religiosity, family plans). 2. Counterfactual change in singlehood and cross-party couple share when partisan composition is equalized within each gender × education cell. 3. Heterogeneity in willingness to pay across self-reported relationship status, religion, and family plans.

Cited by

No citation data available for this registration yet.

Keywords