Article Summary
Introduction
How do politics spillover into non-political realms of behavior? A large literature has documented the rise of affective polarization in U.S. politics, but few studies have directly studied the consequences of such polarization beyond political attitudinal responses given in a survey context. In their paper, McConnell, Margalit, Malhotra, and Levendusky analyze whether partisanship affects economic behavior, something central in the everyday lives of typical Americans.
The authors hypothesize partisanship should condition economic behavior insofar as partisanship has increasingly become a social, group-based identity. Evidence from research on other social identities suggests individuals with strong social identities act in ways to further the best interests of their group, regardless of context. These partisan biases, then, are likely to manifest in apolitical realms such as economic decision-making. The authors go beyond attitudinal correlates by directly attempting to measure behavior in a series of field experiments, greatly expanding how we understand the effects of affective polarization in everyday life.
Analytical Approach
The authors deploy three experiments: two field experiments in an online marketplace and one incentivized population-based survey experiment (details in the table below). In each, they attempt to determine how partisanship conditions economic behavior.
| Study | Description | Treatments | Outcomes |
|---|---|---|---|
| Study 1: Field Experiment in an Online Labor Market | Workers of an online labor market are asked to copyedit a website with randomized partisan cues | Control: Text suggested founders met working for a non-profit Republican: founders met working for Republican party Democrat: founders met working for the Democratic party | Reservation wage (wage required for worker to be contracted again), number of errors caught, number of corrections made |
| Study 2: Partisan Congruity and Consumer Choice | Respondents asked whether they would like to purchase a gift card at a discounted price, with randomized partisan cues | Baseline: email suggests cards were leftover from work with a non-profit Republican: leftover from work with Republican campaigns Democrat: leftover from work with Democratic campaigns | Whether respondent responded to offer, whether responded continued to purchase card |
| Study 3: An Incentivized, Population-Based Survey Experiment | Respondents had to choose between two cash payments for their survey completion | Baseline: Choice between receiving $3 or $6 and $4 donated to opposing party national committeeHigher Payment: Choice between $3 or $9 and $4 donated to opposing partyReligion: $3 or $6 and $4 to religious outgroup (American atheists, Christian legal society) Geography: $3 or $6 and $4 to Association of Western (Eastern) States (based on where they lived) | Choice response |
In the first study (n = 1,232 over two waves), participants were recruited from MTurk to copyedit the text of a website. Within the editing task, the authors embedded randomized treatments suggesting the founders of the company owning the website met working for a non-profit, the Republican party, or the Democratic party. After gathering pre-treatment demographic information, the authors measured the participants’ willingness to do another editing task for the company in the form of a retention wage, the number of errors they caught (out of a possible 11), and the number of corrections made.
The second study utilized a group of primarily Democratic respondents (n = 1,657) recontacted from a previous survey. The respondents were sent an email explaining the organization was giving away $50 Amazon gift cards at a discounted price of $25. The text of the email explaining where the gift cards came from varied by condition. In the baseline category, the email explained the cards were leftover from work with a non-profit, but in the partisan treatments they were leftover from work with Democratic/Republican campaigns. The authors measured the participants’ willingness to respond to the email inquiry and ultimately go through in purchasing the cards.
Finally, the third study recruited a nationally representative sample from GfK (n = 3,266) to conduct an incentivized survey. At the end of the survey (after demographic information was collected and distractor tasks administered), respondents were thanked for their participation and asked to choose between two payment options, which varied by treatment condition (shown in the table above). Across all conditions, respondents had the opportunity to be paid a small amount of money or a slightly larger amount that would accompany a donation to an outgroup (partisan, religious, geographic). The researchers recorded their choice. All experimental results were analyzed with ordinary least squares.
Main Findings
Together, the studies offer evidence of partisanship influencing economic behavior. While results vary in size and significance, the overarching trend is participants behaving differently based on partisanship.
In study 1, partisans editing for their own party demanded a lower reservation wage, but there was no difference in the wage demanded by those editing out-party websites, suggesting greater effect of in-group favoritism than out-party hostility. There were no consistent effects on errors caught and total number of edits.
In study 2, while respondents contacted by their favored political group seemed more likely to respond, this result was not significant at conventional levels. There was no effect for the counter-partisan request. However, among strong partisans, the effects were strong and significant. A follow-up study using an ecological design seems to confirm these results.
Lastly, in study 3, the acceptance rate for the lower monetary amount in the non-partisan condition was about 75%, meaning ¾ of partisans were willing to give up a larger payout to prevent the opposing party from receiving money. This persisted even in the higher payment condition, reducing the acceptance of the non-partisan offer by only 4.8%. These effects are greater than the effects for the placebo geography condition (33% acceptance rate) and were similar to those of the religion condition (77%).
Implications
Most fundamentally, the results underscore the prevalence of partisan reasoning and biases in apolitical behavior. Partisanship is seemingly a deeply held social identity that influences more than just political attitudes, reaching into parts of our everyday lives in ways previously undocumented by researchers.
From a measurement perspective, these results provide deeper behavioral insights to patterns previously only documented at the attitudinal level. By showing how partisanship conditions behavior, the authors move beyond the possibility of mere partisan “cheap talk” or cheerleading that might influence survey responses.
Additionally, the results give mixed insight into the effects of in-group love versus out-group hate. Study 1 seemingly showed the dominant influence of in-group love, but study 3 showed how partisans were willing to forego higher payouts just to without money from a rival party. The authors suggest this variation may be explained by the salience of partisanship in any given condition; the greater the salience, the more willing participants are to demonstrate out-party animosity. In more banal day-to-day interactions, however, in-group love may more strongly condition behavior.
Questions left unanswered
The authors give a comprehensive and detailed answer to their question of what the consequences of partisan attachments are in economic situations. As they concede, they are unable to determine if these consequences are primarily driven by in-group warmth or out-group antipathy. More broadly, there are some unanswered questions on the mechanisms behind the observed effects. The authors mention in the appendix, for example, that they are ambivalent on whether the propensity to decline the partisan offer in study 3 is programmatic- or animus-driven. How exactly affective polarization comes to bear on these economic interactions, then, is a question requiring further research.
Methods and Analysis
Was the study and its analyses pre-registered?: Yes
Did the study rely on proxy variables to measure polarization?: No
Were standard p-value thresholds used (p<.05 or 95% Confidence Intervals that don’t overlap zero)?: Yes
- Largest p-value presented as significant: 0.1
Were correlational results interpreted with causal language?: No
Limitations / Weaknesses
One limitation the authors admit is that their experiments cannot causally identify the effect of individual-level partisanship on behavior. This is because partisanship itself is always endogenous, potentially correlated with heterogeneity between parties that could be related to the resulting behavior. Of course, this is a limitation shared by much of the affective polarization literature, as directly manipulating partisanship is a seemingly impossible task. Another, perhaps more fundamental limitation about the paper is in its scope. While the authors do exceptionally well to provide behavioral measures instead of the usual attitudinal results, those behavioral measures are still limited to the online environment. All participants are either reacting to or completing an online task, and there may be something unique about the online environment that allows partisanship to act within seemingly apolitical areas. Perhaps results may differ in an offline scenario where consequences of actions are more acutely felt socially.
Open Data & Analyses
Does the article make the replication data publicly available?: Yes
Does the article make the replication analysis scripts publicly available?: Yes
Article Citation
McConnell, C., Margalit, Y., Malhotra, N., & Levendusky, M. (2018). The economic consequences of partisanship in a polarized era. American Journal of Political Science, 62(1), 5–18. https://doi.org/10.1111/ajps.12330
Bibtex
1
2
3
4
5
6
7
8
9
10
11
12
@article{https://doi.org/10.1111/ajps.12330,
author = {McConnell, Christopher and Margalit, Yotam and Malhotra, Neil and Levendusky, Matthew},
title = {The Economic Consequences of Partisanship in a Polarized Era},
journal = {American Journal of Political Science},
volume = {62},
number = {1},
pages = {5-18},
doi = {https://doi.org/10.1111/ajps.12330},
url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/ajps.12330},
eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1111/ajps.12330},
year = {2018}
}