Article Summary
Introduction
Iyengar et al (2018) examine how rising political polarization has affected intra-family political agreement. The authors theorize that increasing polarization will lead to greater political agreement between spouses which will in turn increase the frequency of intergenerational transmission (i.e., children identifying with the same political party as their parents). They test this question by comparing partisan agreement between “prepolarization” (1965-1973) and “postpolarization” eras (2014-2016). After demonstrating that there is an increase in political agreement within families pre- vs. post- polarization, the authors address several reasons for why this may be the case.
Iyengar et al (2018) identify and test four potential explanations for why spouses have higher levels of political agreement in this polarized era:
- Choice homophily - Partisans may be intentionally selecting their spouses based on shared partisanship (or avoiding some potential spouses who are from opposing political parties).
- Induced homophily - Partisans may be more likely to marry those from the same political party because they are selecting for other characteristics that tend to vary with partisanship (e.g., religion).
- Constrained partner markets - Cross-party marriages may be rarer because of politically homogenous communities and limited interaction with people of different political parties.
- Convergence/persuasion - Political views of spouses may grow more similar after their relationship begins.
The authors then turn to explanations for intergenerational transmission. In line with classic studies on the topic (e.g., Jennings & Niemi, 1974; Jennings et al. 2009) they suggest that increased spousal political agreement is responsible for higher rates of partisan correspondence between parents and their children. When spouses identify with the same political party their children are more likely to pick up on partisan cues from their parents, and frequent political conversations within the home where there is agreement is more likely to shape children’s political views to be like their parents.
Analytical Approach
The authors first compare the levels of spousal and intergenerational political correspondence in 1965 (data from Jennings & Niemi, 1974) to 2015 (two original YouGov surveys). They compare these differences by calculating polychoric correlations (a specific type of correlation commonly used for ordinal variables). They also compare intra-family political agreement from their 2015 surveys to 2014 and 2016 voter file data (compiled by TargetSmart) and conclude that the estimates from their surveys are similar to those seen in voter files. They find that both spousal and intergenerational partisan agreement have increased since the 1960s and 70s.
Next, the authors compare potential mechanisms explaining spousal partisan agreement. They compare the choice homophily explanation to the other three potential mechanisms (induced homophily, constrained partner markets, and convergence/persuasion) and conclude that choice homophily is the best explanation of the four. Table 1 contains the analyses used as evidence against three of the potential mechanisms (in the order presented in the paper).
Table 1: Analyses comparing choice homophily to other potential mechanisms for spousal partisan agreement
| Mechanism | Analyses |
|---|---|
| Constrained Partner Market | New marriages are identified using 2014 and 2016 voter files. Spousal homogeneity rates are calculated for each zip code and compared to the percentage of Democrats and Republicans in that zip code. They find that there is a “modest” increase in spousal partisan homogeneity in zip codes that are more politically homogenous. However, they compare this to the relationship between racial homogeneity of zip codes and spousal racial homogeneity and find that the effects are larger for race than party. |
| Convergence/persuasion | Using 2015 survey data they estimate three separate multiple regression models predicting disagreement of partisan attitudes, religion, and lifestyle interests. They find that length of marriage is a non-significant predictor for partisan disagreement (but predicts significantly less religious agreement). Additionally, they compare recently married couples (who theoretically have had little time to converge) in 1973 to recently married couples in the 2014 voter file and find that spousal agreement is much higher in 2014. And finally, they compare the partisanship of new couples in 2016 to their partisanship in 2014 (prior to living together) and find only a 3 percentage point increase in correspondence after moving in together. |
| Induced Homophily | In separate regression models predicting disagreement on partisanship and religion, they show that initial disagreement (when controlling for covariates) is higher for religion than partisanship (although not significantly so). Next, they show that there is no interaction between agreement on nonpolitical traits (religiosity or authoritarianism) and length of marriage when predicting partisan agreement, suggesting that agreeing on nonpolitical topics does not lead to greater political convergence over time. Additionally, they note that partisan agreement in new spousal pairs is higher than educational agreement. |
After evaluating the potential mechanisms for spousal political agreement, the authors examine intergenerational transmission rates and potential explanations for why they have changed. They compare intergenerational transmission rates between consistent parents (same political party) to inconsistent parents (different political party) and find that transmission rates are much higher for consistent parents. Importantly, this pattern was not observed at all in the 1965 data. They also compare the rates by which parents misperceive the partisanship of their parents in 1965 vs. 2015 and find that the frequency of misperceptions has dropped dramatically. They then examine the relationship between political discussion in the household and transmission rates in both time periods, and find that in both eras when parents discuss politics at home the transmission rates are higher.
Main Findings
Iyengar et al (2018) document a marked increase in spousal and intergenerational partisan agreement among Americans from the 1960s and 1970s to the 2010s, in the same time period that affective polarization has risen in the United States.
They examine potential explanations for increasing spousal agreement and conclude that many Americans are actively considering political partisanship when deciding who – and who not – to marry. The authors argue that other explanations for increasing spousal partisan agreement are insufficient to explain this increase. They demonstrate that intentional selection based on partisanship is a stronger mechanism for spousal partisan agreement than explanations such as lack of out-partisan options, convergence of political views over time once married, and selection of spouses based on traits that tend to vary with partisanship.
They find that spousal political agreement is a major predictor for intergenerational transmission. Additionally, individuals are much less likely to misperceive their parents’ partisanship in the 2010s compared to in the 1960s, which likely plays a role in the rates of transmission. They also find, consistent with previous research (Jennings & Niemi, 1974), intergenerational transmission rates are higher when parents discuss politics with each other.
Implications and Questions left unanswered
The authors demonstrate that spousal political agreement is widespread. They demonstrate that spouses are selecting intentionally based on political views. For those who find their argument persuasive, this raises several questions about why people are selecting on political views. Do people perceive a potential partner’s partisanship indicative of the type of spouse or parent they will be? Are there social pressures against marrying someone from a different political party? Or do political arguments cause dating couples to become frustrated and break up?
Additionally, as partisanship becomes more homogenous within familial relationships, is there greater relational harm when family members disagree? For example, if a child joins a political party that his/her parents do not belong to, will this cause relational harm? If one spouse changes their partisanship does this increase the likelihood of divorce or marital issues?
From a methodological perspective, this paper uses several creative analyses that may be useful for future research. Iyengar et al note that there is no longitudinal data tracking spouses before and after meeting – this would be very helpful for identifying whether couples select based on partisanship or grow more similar over time. However, the authors found a clever work around to utilize a before and after research design: they identified new spousal pairs by finding people who lived separately in the 2014 voter files but lived together in the 2016 voter files. While this method isn’t perfect (e.g., it is unclear if the couple was already dating in 2014) it allowed for the authors to test for change across time and across relationship stage (living together vs. living apart) to look for convergence of political views. This method may be useful in future research where pre-post data is desired but not readily available.
Methods and Analysis
Was the study and its analyses pre-registered?: No
Did the study rely on proxy variables to measure polarization?: Yes
In describing the relationship between spousal agreement and polarization, the authors use feeling thermometer ratings of presidential candidates (Hillary Clinton and Donald Trump) rather than political parties.
Were standard p-value thresholds used (p<.05 or 95% Confidence Intervals that don’t overlap zero)?: Yes
- Largest p-value presented as significant: 0.05
Were correlational results interpreted with causal language?: Yes
Discussion and Conclusion Section, Paragraph 5: “Partisan agreement within the family strengthens polarization.”
Limitations / Weaknesses
The authors make a persuasive case and they include extensive analyses to answer a complicated question. Nevertheless, there are a few limitations of this paper. First, as the authors point out, there is no longitudinal survey data of (potential) spouses before and after meeting and marrying. The authors use a clever analysis to effectively create pre-post data, but a longitudinal survey approach (if feasible) may provide additional information. Second, the authors note that there is a “modest” effect of constrained partner markets (political homogeneity of communities) on spousal political agreement. Although they demonstrate that even in heterogeneous communities there is nearly 80% spousal homogeneity, they could have included a discussion of how communities have become increasingly homogeneous since the 1960s (e.g., Bishop & Cushing, 2009) to provide additional context. Third, the authors suggest induced homophily is unlikely to explain spousal homophily because spousal agreement on partisanship tends to be higher than for other individual characteristics (religion and education; although they note that religion is not significantly different from partisanship). But they do not address the cumulative effect of agreement on multiple characteristics which correlate with partisanship. Their analyses show a reasonably high level of spousal agreement on religion and education. Additionally, race tends to have high agreement with only a 17% intermarriage rate for new marriages in 2015 (Pew, 2017). Their argument would be strengthened by being able to demonstrate that partisan agreement is still an important factor in spousal selection even over and above the cumulative force of these other factors. Research on social sorting (e.g., Mason, 2016, 2018) could be considered in this section. Finally, the difference in feeling thermometer ratings of in-party vs. out-party presidential candidates is found to be much higher for spouses who are politically aligned compared to spouses where one is a Democrat and the other is a Republican. Based on this evidence, they conclude “Partisan agreement within the family strengthens polarization.” This causal language is not sufficiently supported by their correlational analysis. It is a distinct possibility that polarized people are much more likely to marry a co-partisan to begin with, which could also explain this difference.
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
Iyengar, S., Konitzer, T., & Tedin, K. (2018). The home as a political fortress: Family agreement in an era of polarization. The Journal of Politics, 80(4), 1326-1338. https://doi.org/10.1086/698929
Bibtex
1
2
3
4
5
6
7
8
9
10
11
@article{iyengar2018home,
title={The home as a political fortress: Family agreement in an era of polarization},
author={Iyengar, Shanto and Konitzer, Tobias and Tedin, Kent},
journal={The Journal of Politics},
volume={80},
number={4},
pages={1326--1338},
year={2018},
publisher={University of Chicago Press Chicago, IL}
}