Introduction and Motivation
The rise of polarization over the past 25 years has many Americans worried about the state of politics. This worry is understandable: up to a point, polarization can help democracies, but when it becomes too vast, such that entire swaths of the population refuse to consider each other’s views, this thwarts democratic methods for solving societal problems~\cite{heltzel2020polarization}.
Political polarization occurs when subsets of a population adopt increasingly dissimilar attitudes toward parties and party members (i.e., affective polarization; [5]), as well as ideologies and policies(ideological polarization;[6]).1 With little-to-no polarization, most people support a mixture of liberal and conservative stances acrossissues, and they can support one party without disliking others. With very high polarization, large, separate clusters of the population endorse ideologically consistent stances across all issues, and love their own party while loathing the other(s).
https://royalsocietypublishing.org/doi/full/10.1098/rstb.2020.0143
Politicalscientists continue todebate the costs andbenefits of polarization. At its best, polarization can be benign, and produce more effective, stable democracies. It encourages civic engagement: Polarized citizens more often vote, protest,andjoin politicalmovements,all ofwhicharenecessary for functioning democracy [15] and help disrupt undesirable status quos [16]. Polarization also entails pluralistic policy alternatives [17]; this is crucial for democracies, which rely on citizens being able to consider multiple policies and have thorough, constructive debates between them [18]. Ideally, this kind of engagement and pluralism ultimately produce effective, stable government: It helps societies identify policies that are both optimal for solving their biggest problems [89], and unlikely to be overturned when a new party takes power since they are mutually agreed-upon [85].
Polarization is a topic of intense interest among social scientists, with analysis of congressional affiliation and voting patterns, sociological studies on popular attitudes, and laboratory studies on media influence and attitude change all in search of a better understanding of central mechanisms~\cite{fiorina2008political,iyengar2012affect,ura2012partisan,druckman2013elite,grosser2014candidate,lauderdale2013does,levendusky2013partisan,prior2013challenge,leeper2014informational,thomsen2014ideological,weinschenk2014polarization,mason2015disrespectfully}. Claims regarding polarization, however, often remain frustratingly vague. The problem is not restricted to popular presentations but appears in the technical literature of sociology, economics, and political science as well. Entire articles appear on polarization with little attempt to make it clear what precisely is meant by the term~\cite{bramson2017understanding,bramson2016disambiguation}.
A multitude of polarization measures have been investigated, along with a number of mechanistic models to explain the emergence of polarized social systems. However, models are often limited in how much, if any, of real observations may be used in informing the assumed rules of belief shift. Indeed, imitation and the influence of social contacts are assumed as ubiquitous drivers of opinion dynamics, despite the fact that in many of these models opinions tend ultimately tend towards global convergence~\cite{abelson1964mathematical}. Alternatively, models have bifurcations, tentatively driving all simulated actors to extreme opinions. Inevitable movements either toward a central consensus or to fully polarized extremes are not reflective of actual effects we observe in social systems~\cite{bramson2017understanding,bramson2016disambiguation}.
Notwithstanding the ambiguity of definition, it is common wisdom has it that American society is becoming increasingly polarized (Fiorina, Abrams, and Pope 2005; Brownstein 2007; McCarty, Poole, and Rosenthal 2008; Hetherington and Weiler 2009). There are measurable aspects of political reality that support that common wisdom. In 1980, only 43% of Americans polled said that they thought there were important differences between the parties. The figure is now 74%. In 1976, almost a third thought it did not make a difference who was president. That figure is now cut in half. Between 1969 and 1976, the Nixon and Ford years, the rate at which Republicans voted along party lines was about 65% in both the House and the Senate. The same was true of Democrats. Between 2001 and 2004, under George W. Bush, Republicans voted with their party 90% of the time. Democrats voted with their party 85% of the time (McCarty et al. 2008). It has been argued, however, that a focus on political polarization within the political elite obscures a stable or declining cultural polarization within the broader population. On most issues, public polarization has not increased between groups, regardless of what groups are being compared: the young and the old, men and women, the more and the less educated, different regions of the country, or different religious affiliations. On a number of points, polarization has clearly decreased. Racial integration was once fought vociferously by major portions of the population, but that is certainly not true now. Views on women’s roles in public life were once extremely contentious in ways that are now quite generally recognized as archaic.
Social media has also been cited as contributing to polarization, via the creatation of echo-chambers and filter bubbles that reinforce their existing opinions (Bakshy et al. 2015; Bessi et al. 2016). In such cases, instead of smoothing the differences, online social networks reinforce them, thus leading to increased polarization.
A precise definition of polrization that satisfies the key qualitative or intuitive requirement sof teh concept is still missing, as are effective measures of computing such measures.
@article{heltzel2020polarization,
title={Polarization in America: two possible futures},
author={Heltzel, Gordon and Laurin, Kristin},
journal={Current opinion in behavioral sciences},
volume={34},
pages={179--184},
year={2020},
publisher={Elsevier}
}
@misc{fiorinasamuel,
title={with Samuel J. Abrams and Jeremy C. Pope. 2005. Culture war? The myth of a polarized America},
author={Fiorina, Morris P},
publisher={New York: Pearson Longman}
}
@book{brownstein2008second,
title={The second civil war: How extreme partisanship has paralyzed Washington and polarized America},
author={Brownstein, Ronald},
year={2008},
publisher={Penguin}
}
@book{mccarty2016polarized,
title={Polarized America: The dance of ideology and unequal riches},
author={McCarty, Nolan and Poole, Keith T and Rosenthal, Howard},
year={2016},
publisher={mit Press}
}
@book{hetherington2009authoritarianism,
title={Authoritarianism and polarization in American politics},
author={Hetherington, Marc J and Weiler, Jonathan D},
year={2009},
publisher={Cambridge University Press}
}
@article{fiorina2008political,
title={Political polarization in the American public},
author={Fiorina, Morris P and Abrams, Samuel J},
journal={Annu. Rev. Polit. Sci.},
volume={11},
pages={563--588},
year={2008},
publisher={Annual Reviews}
}
@article{iyengar2012affect,
title={" Affect, not ideology: A social identity perspective on polarization": Erratum.},
author={Iyengar, Shanto and Sood, Gaurav and Lelkes, Yphtach},
year={2012},
publisher={Oxford University Press}
}
@article{ura2012partisan,
title={Partisan moods: Polarization and the dynamics of mass party preferences},
author={Ura, Joseph Daniel and Ellis, Christopher R},
journal={The Journal of Politics},
volume={74},
number={1},
pages={277--291},
year={2012},
publisher={Cambridge University Press New York, USA}
}
@article{druckman2013elite,
title={How elite partisan polarization affects public opinion formation},
author={Druckman, James N and Peterson, Erik and Slothuus, Rune},
journal={American Political Science Review},
volume={107},
number={1},
pages={57--79},
year={2013},
publisher={Cambridge University Press}
}
@article{grosser2014candidate,
title={Candidate entry and political polarization: An antimedian voter theorem},
author={Grosser, Jens and Palfrey, Thomas R},
journal={American Journal of Political Science},
volume={58},
number={1},
pages={127--143},
year={2014},
publisher={Wiley Online Library}
}
@article{lauderdale2013does,
title={Does inattention to political debate explain the polarization gap between the US congress and public?},
author={Lauderdale, Benjamin E},
journal={Public Opinion Quarterly},
volume={77},
number={S1},
pages={2--23},
year={2013},
publisher={Oxford University Press UK}
}
@article{levendusky2013partisan,
title={Why do partisan media polarize viewers?},
author={Levendusky, Matthew S},
journal={American Journal of Political Science},
volume={57},
number={3},
pages={611--623},
year={2013},
publisher={Wiley Online Library}
}
@article{prior2013challenge,
title={The challenge of measuring media exposure: Reply to Dilliplane, Goldman, and Mutz},
author={Prior, Markus},
journal={Political Communication},
volume={30},
number={4},
pages={620--634},
year={2013},
publisher={Taylor \& Francis}
}
@article{leeper2014informational,
title={The informational basis for mass polarization},
author={Leeper, Thomas J},
journal={Public Opinion Quarterly},
volume={78},
number={1},
pages={27--46},
year={2014},
publisher={Oxford University Press UK}
}
@article{thomsen2014ideological,
title={Ideological moderates won’t run: How party fit matters for partisan polarization in Congress},
author={Thomsen, Danielle M},
journal={The Journal of Politics},
volume={76},
number={3},
pages={786--797},
year={2014},
publisher={Cambridge University Press New York, USA}
}
@article{weinschenk2014polarization,
title={Polarization, Ideology, and Vote Choice in US Congressional Elections},
author={Weinschenk, Aaron C},
journal={Journal of Elections, Public Opinion \& Parties},
volume={24},
number={1},
pages={73--89},
year={2014},
publisher={Taylor \& Francis}
}
@article{mason2015disrespectfully,
title={“I disrespectfully agree”: The differential effects of partisan sorting on social and issue polarization},
author={Mason, Lilliana},
journal={American Journal of Political Science},
volume={59},
number={1},
pages={128--145},
year={2015},
publisher={Wiley Online Library}
}
Our definition
Thus, definition of societal polarization, despite the diverse attempts at quantification, needs work. In this study we develop a fundamentall novel approach to quantifying divergence of opinion, without mapping multi-dimensional opinions which tend to be either categorical or ordinal variables to cardinal variables or normalized scalars between 0 and 1. We have the following advantages:
- data driven inferrence, we infer cross-dependencies from survey data
- no need to identify groups in the data, or stratify by socio-economic variables
- metric adapts to changing social norms over time
- applies to multidimensional opinion space
- No need to map opinions to cardinal values
- measures polarization as a canonical distance between extreme opinion vectors, which changes to reflect changes in the cannonoical metric over time
Filter bubbles
It has been suggested that users tend to create connections with like-minded individuals, and create echo-chambers and filter bubbles that reinforce their existing opinions (Bakshy et al. 2015; Bessi et al. 2016). In such cases, instead of smoothing the differences, online social networks reinforce them, thus leading to increased polarization.
This might not be true, mixing might also increase polrization counterintuitively
Major families of models for social phenomena that have been appealed to as offering clues to the central mechanisms of polarization: 1) Axelrod’s Cultural Diffusion and Polarization models represent one modeling tradition (Axelrod 1997; Klemm et al. 2005; Flache and Macy 2006a; Centola et al. 2007). 2) The Hegselmann-Krause Bounded Confidence model and Deffuant’s Relative Agreement model define another approach (Deffuant et al. 2002; Hegselmann and Krause 2002; Deffuant 2006). 3) Models in a Structural Balance tradition constitute a third family (Heider 1946; Cartwright and Harary 1956; Harary 1959; Macy et al. 2003; Klemm et al. 2005; Kitts 2006). We extend the analysis to mechanisms for ‘group polarization’ suggested within social psychological theories of selfcategorization (Lord, Ross, and Lepper 1979; Hogg, Turner, and Davidson 1990). Each of the models analyzed purports to capture polarization, but it is clear that both the kinds and the patterns of phenomena they generate vary widely. In the Axelrod models, polarization only occurs strongly when the characteristics in which cultures can vary are few relative to the number of ways each characteristic can be expressed. It is not clear how we could systematically or objectively enumerate the number of either features or traits for real societies in order to check the plausibility of this claim. As a formal requirement for at least some cases, however, it may not be implausible. For example, one interpretation is that there must be more potential positions to take on each political issue than there are distinct issues on the table. The common reality of familiar forms of social polarization is that of roughly balanced oppositional groups. Polarization of the type that appears in the Axelrod models at equilibrium, in contrast, almost always either a is radically one-sided or b appears as a myriad of tiny groups.
%In an influential series of articles, Hegselmann and Krause develop a ‘bounded confidence’ model of opinion polarization that functions in terms of mutual influence among those within a specific threshold of similarity (Hegselmann and Krause 2002, 2005, 2006).
The family of "bounded confidence" models draw on a long tradition of belief averaging as a simple representation of the important psychological phenomenon of peer influence (French 1956; DeGroot 1974). But we are aware of virtually no evidence that real polarization occurs ‘from the edges’ as it does in these models or that it crystallizes in virtue of a specific distance from the edges as it does here. Indeed, there is a great body of evidence that the dynamics of developing polarization are quite different. The classic study by Lord et al. (1979) shows groups in laboratory conditions that progressively polarize, increasing the distance between them over time despite balanced bodies of evidence (see also Miller et al. 1993; Kuhn and Lao 1996). Cooper, Kelly, and Weaver claim that “one of the most robust findings in social psychology is that of attitude polarization following discussion with like-minded others” (2001, 267).
In this study we develop a fundamentall novel approach to quantifying divergence of opinion, without mapping multi-dimensional opinions to ordinal values.
Polrization in US Society
Common wisdom has it that American society is becoming increasingly polarized (Fiorina, Abrams, and Pope 2005; Brownstein 2007; McCarty, Poole, and Rosenthal 2008; Hetherington and Weiler 2009). There are measurable aspects of political reality that support that common wisdom. In 1980, only 43% of Americans polled said that they thought there were important differences between the parties. The figure is now 74%. In 1976, almost a third thought it did not make a difference who was president. That figure is now cut in half. Between 1969 and 1976, the Nixon and Ford years, the rate at which Republicans voted along party lines was about 65% in both the House and the Senate. The same was true of Democrats. Between 2001 and 2004, under George W. Bush, Republicans voted with their party 90% of the time. Democrats voted with their party 85% of the time (McCarty et al. 2008). It has been argued, however, that a focus on political polarization within the political elite obscures a stable or declining cultural polarization within the broader population. On most issues, public polarization has not increased between groups, regardless of what groups are being compared: the young and the old, men and women, the more and the less educated, different regions of the country, or different religious affiliations. On a number of points, polarization has clearly decreased. Racial integration was once fought vociferously by major portions of the population, but that is certainly not true now. Views on women’s roles in public life were once extremely contentious in ways that are now quite generally recognized as archaic.
Polarization is currently a topic of intense interest among social scientists, with analysis of congressional affiliation and voting patterns, sociological studies on popular attitudes, and laboratory studies on media influence and attitude change all in search of a better understanding of central mechanisms (Fiorina and Abrams 2008; Iyengar, Sood, and Lelkes 2012; Ura and Ellis 2012; Druckman, Peterson, and Slothuus 2013; Großer and Palfrey 2013; Lauderdale 2013; Levendusky 2013; Prior 2013; Leeper 2014; Thomsen 2014; Weinschenk 2014; Mason 2015). Claims regarding polarization, however, often remain frustratingly vague. The problem is not restricted to popular presentations but appears in the technical literature of sociology, economics, and political science as well. Entire articles appear on polarization with little attempt to make it clear what precisely is meant by the term~\cite{bramson2017understanding,bramson2016disambiguation}.
%but as early as 1964 Robert Abelson~\cite{abelson1964mathematical} noted that models in which agents imitate the opinions of others seem to tend inevitably toward central convergence.
Even simple emergence of bimodility, an often observed charcateristic of polarized soicial discourse, is difficult to achieve without strong and often unstantiated assumptions. Another extreme effect observed in model simulations is bifurcation, that iteratively drives all actors to extreme , and inevitable movements either toward a central consensus or to full polarized extremes is not reflective of actual effects we observe in society~\cite{bramson2017understanding,bramson2016disambiguation}. [REF]
%Polarization is a topic of intense interest among social scientists, but there is significant disagreement regarding the character of the phenomenon and little understanding of underlying mechanics.