The other day I got into an argument. Nothing new there except maybe that it happened over Zoom instead of face-to-face. My opponent, a usually mild-mannered member of my church book club, was insisting that only a minor fringe of society believed in disbanding the police. I disagreed, and I was confident that I had enough evidence1 at my disposal to change his mind. After all, I had just seen three pieces of confirming evidence in the weeks running up to the book club meeting. The first was a New York Times Article (Yes, We Mean Literally Abolish the Police). Second, a Gallup poll which revealed that 15% of Americans, 27% of Democrats, and 33% of Americans under age 34 selected “abolishing police departments” in a nationally-representative survey. Third, the graduate student union I currently belong to emailed its members with an update saying they would work to “disband the UIPD” (email to GEO members on September 15, 2020).
Clearly, this was more than just a fringe view. I presented this evidence sequentially to my opponent. He dismissed the Times article as coming from the Opinion section, and therefore not necessarily representative of a mainstream view. All well and good, I said, but what about the survey results? Surveys are often biased, he replied; he didn’t trust this one. And why didn’t he trust it? Because it presented evidence contrary to his prior belief. If the survey results had favoured his argument, he assuredly wouldn’t have been so squeamish about survey methodology.
Now, I don’t want to suggest that there is anything mentally deficient in my opponent, although he was wrong, dang it! This process of evaluating new information based on our prior beliefs (what scholars call motivated reasoning) comes with the territory when you are a human being. And that includes human beings with the unenviable task of making political decisions in a highly partisan and polarized democratic environment.
It is these human beings that I talk about in this essay, these citizen political actors. I first describe motivated reasoning and how it works, then move on to how this relates to partisanship, and finally speculate on what this all means for our understanding of democracy.
What is motivated reasoning? I presented an example of motivated reasoning in the opening paragraph. More formally stated, motivated reasoning is our tendency as human beings to find arguments in favour of conclusions we want to believe more compelling than arguments that have conclusions we do not want to believe (Kunda, 1990). In other words, we prefer new information that confirms our existing attitudes and beliefs. This is fairly similar to Zaller’s resistance axiom from his RAS (“Receive-Accept-Sample”) model of citizen learning, where people reject messaging that is inconsistent with their political values (1992). The principal difference that I detect between them is that Zaller’s model suggests mostly information avoidance or passive declining of contrary information, whereas motivated reasoning involves constructing mental arguments against contrary information. It is a more combative process than Zaller describes.
In their excellent article on motivated reasoning (which I will use heavily in this paper), Taber and Lodge provide several reasons why human beings use this motivated reasoning (2006, p.767). The first is that, presumably, we have developed our initial attitudes through either careful deliberation or personal experience spanning years. If we have made it more or less unscathed in life up to this point with our current beliefs, it wouldn’t make any sense to change those beliefs or attitudes lightly. Secondly, because we exerted cognitive effort in gaining those beliefs in the first place, we are unwilling to see those cognitive resources wasted. We could think of this as an example of the sunk cost fallacy. I face this every time I am writing a paper and someone makes a suggestion that will significantly alter (and probably improve) it. I often get defensive because I have put in hours doing a literature review, coming up with a research design, etc. It is hard to let all those hours “go to waste.” Thirdly, we may be suspicious that the new contrary information reflects the biases and nefarious aims of other people who are not seeking our welfare. In that case, it is within our interest to ignore that information. And finally, although not mentioned by Taber and Lodge, I believe that we might fit motivated reasoning in with the underlying assumption that all human beings seek to maximize their feelings of self-worth (e.g. Hogg and Abrams, 1990). It is not pleasant to find out that you are wrong, especially if you find out you have been wrong about a long-held belief. So you resist that change as much as possible.
How does it relate to partisanship? The answer to this question depends upon which model of citizen learning is informing our understanding of partisanship.2 If we believe that partisanship is an expression of social identity, i.e. what Huddy, Mason, and Aaroe (2015) call the expressive approach, that only increases the role of motivated reasoning. That is, new information that challenges one’s group identity (which is a key part of one’s personal identity and self-esteem) is likely to invoke even stronger negative reactions. We might, however, follow the rational or economic model proposed by scholars like Downs (1957), where party identification is the result of keeping a simple running tally of each party’s performance and choosing the one that brings the most benefit to oneself.3 Motivated reasoning in this case would mean that citizens are less likely to keep an objectively accurate running tally. They will dismiss good evaluations of the opposing party even when they objectively perform well, and they will reject information that suggests that their preferred party is underperforming. So their running tally may end up horribly inaccurate from a factual standpoint, and be more like a running list of confirmatory instances where any new event backs up pre-existing beliefs.
Achen presents a slightly more nuanced view of this running tally in his Bayesian updating model (1992). With his modifications, people do still update accordingly as they receive new information, but as your store of knowledge increases, the new information matters less. Gerber and Green point out that the logical fallout of this model is that eventually new information ceases to matter, which they argue is not the case in reality (1998, p.796). Of course there are situations where people change their minds, but for the most part I think Achen’s model mirrors my personal observations well. How many staunch old partisans, who have developed their attachments over a life-time of experience, would be willing to change their partisan identity? When President Trump rules our country as a warlord from his bunker in Mar-a-Lago, my eldery relatives will still cheerfully be voting Republican.4
Motivated reasoning, however, would suggest that new information should consistently matter. Take Taber and Lodge’s sequential study design (2006). When presented with new information (including both for and against arguments), the process of motivated reasoning made respondents more polarized when answering the second set of questions than compared to the first set of questions. The new information mattered, and it mattered especially for political sophisticates. According to a Bayesian updating process, one should be swayed little by new information when one already has a great store of information in one’s priors. But Taber and Lodge find that political sophisticates are the most reactive to new information (in the confirmatory direction).
Although Green and Gerber rightly criticize aspects of Achen’s model, their own model of partisan learning is fairly similar (see pages 796 to 809). In their model, citizens still begin with some beliefs about their expected payoffs for voting Democrats or Republican, and they combine their prior belief and what they estimate to be the party differential benefit. When they encounter new information, that has variable weights in their model. The amount each new piece of information is weighted depends on how much information it contains, and a bigger weight means more attention and greater updating of belief. So, unlike Achen’s model, people do still pay attention to current information, which has the possibility to change belief. While I think the updates to Achen’s model are sensible, I still dislike the underlying assumption that citizens are evaluating what they could get out of each party. To me, that is not how partisanship forms. Indeed, I am much more sympathetic to the argument I detailed about ingroup attachment and partisanship. It is more akin to tribalism than a rational choice about which party offers more to the voters. As Achen points out in the opening sentence of his 2002 article (p.151) : “All I know is we’re not Republicans. My father isn’t. – Judith, age 10.
Achen’s point (representing a slightly updated view from his 1989 article) is that partisan identity precedes any knowledge of what each party stands for. How can you have any beliefs about a payout or a “party differential” when you haven’t the slightest notion of what Democratic or Republican means? This is similar to what Bartels calls “the Michigan model”–partisan loyalties are formed very early and remain difficult to move later in life (2002, p.17). Including motivated reasoning into Green and Gerber’s model fits nicely with their findings about current information still mattering. If we believe their notion of citizens working out expected payoffs between the parties, motivated reasoning would suggest that citizens will be quite poor at accurately working out those payoffs. As I alluded to with the running tally model, if citizens seek out only confirmatory information and reject dissenting information, how can they know what the other party actually offers in any sort of accurate way?
So far I have discussed models of partisanship that have largely ignored the role of motivated reasoning, and I have explained ways each model suffers from this omission. I now turn to Taber and Lodge’s discussion of partisanship which is primarily concerned with motivated reasoning (2006). There are several key takeaways from their work in regards to partisanship. First, that motivated reasoning affects citizens heterogeneously. Those who feel more strongly about an issue are affected by motivated reasoning more strongly. And of course, those who feel more strongly about an issue are usually those who know more about that issue–the political sophisticates. Greater sophistication in this case is not protection from motivated reasoning, in fact, their experiments reveal that the more politically knowledgeable use motivated reasoning more (p.761). Secondly, that more cognitive resources are spent dismantling the counter argument from the other party than on supporting arguments. Again, this is more true of political sophisticates than knowledge. Thirdly, that people systematically choose to expose themselves to supportive rather than contrary information. Fourthly, that even exposing people to both sides of the argument leads to more extreme partisan polarization. Indeed, that is expected from the previous finding about more cognitive resources being spent on the opposing article. “Hearing the other side” as Mutz would put it in this case leads only to strengthened convictions of one’s prior.
One puzzling aspect from Taber and Lodge’s discussion of partisanship is temporal. If this process is endemic to every human being, and the impact of new information only strengthens one’s partisan identity and increases polarization, then why only recently have we begun to see increased polarization? Why hasn’t the United States (and other countries for that matter) drifted steadily towards partisan polarization from the founding until now? Instead, we have had long periods of very little political polarization. Was that simply because the parties were more ideologically muddled? Taber and Lodge only present a 2-step temporal experiment, but it would be interesting to see their experiments over a longer time frame. That may help us identify how this process works over time.
One slightly different interpretation of Taber and Lodge’s story comes from Gaines et al. (2007). Although they reference Taber and Lodge’s work on motivated reasoning, they do not engage with it directly. Instead, they propose a model of partisan interpretations of common facts. That is, even though Republican and Democratic partisans may hold similar factual beliefs , the difference lies in their interpretation of those facts.5 And like Taber and Lodge found with political sophisticates, those who are better informed are able to more effectively interpret facts to buttress their existing partisan views.
What does this mean for democracy? Despite the ubiquity of partisanship, Americans have always found the term somewhat distasteful. People often talk in scornful terms of those who enter the ballot box and select Republican or Democrat all down the ballot without any knowledge of any of the races. After all, that is antithetical to the popular conception of republican democracy described by Baker (2002). Citizens are supposed to be public-spirited and dedicated to finding the common good. Policy and opinion are supposed to originate from the bottom up, through reasoned debate, deliberation, and communication with other citizens. And the best argument is supposed to prevail as a result of this deliberation.
Of course, as I’ve examined in this paper, the best argument does not prevail. Thanks to motivated reasoning, the argument advanced by the party that we agree with is the argument that will prevail. Citizens are not learning from one another in the spirit of enlightenment and tolerance. When they can, they practice selective exposure to avoid seeing disagreeable information (Mutz & Martin, 2002; Taber & Lodge, 2006). And when they do hear arguments contrary to their own, it only reinforces their own beliefs. This effectively takes away the panacea that cross-cutting media represented for scholars like Mutz.
Furthermore, the bottom up model of citizen discourse may be perpetuating the many false and potentially dangerous conspiracy theories that are currently endemic (see for example Oliver & Wood, 2014). This bottom up paranoia, selective exposure, and inability to process counter arguments are all very serious challenges to this long-standing and optimistic view of democracy that is republicanism.
There are, however, at least two counter-arguments against this gloomy indictment of republican democracy. The first regards citizen knowledge. Political scientists from Converse on have typically been rather scathing in their portrayal of ordinary citizens’ knowledge and ability to function in a democracy. For example, Gaines et al. imagine a political science graveyard where each epitaph reads something along the lines of “if only the citizens were better informed they would be able to make political judgments more effectively” (2007, p.957). Perhaps, however, political scientists have had their argument the wrong way round all along. It is the political sophisticates (whom Converse and others would regard as the more ideal citizens) who show the most bias in processing political arguments (Taber & Lodge, 2006, p.767). Less sophisticated citizens show less bias in their processing of counter-arguments; with less arguments at hand to call upon, there is a greater chance of changing one’s mind with the presentation of contrary evidence. Even though this is born out in the data, it is still a normatively troubling solution. It’s akin to saying ignorance is bliss; cancel all the civics, history, and political science classes at once! Obviously not a good long term solution to the problem of citizen learning in a democracy.
The other approach we could take when learning about citizens’ inability to avoid motivated reasoning in forming partisan attachments is to change our conception of democracy. At some point, do we have to recognize that our democracy looks nothing like the republican democracy that nineteenth century progressive-era scholars described? And do we have to recognize that our very cognitive style of processing new information makes that form of democracy impossible to achieve?
If we do finally abandon our ideal of republican democracy, it does not mean that we give up on democracy as a system. Liberal pluralism (also called interest group democracy) is a perfectly valid form of conceptualizing citizen participation in politics, and arguably is the model that most describes America in 2020. In this model (see Baker, 2002), democracy is competing groups of self-serving, like-minded people. Although the groups are competing for political power, they are doing so under a fair, common set of rules. Importantly, this model does not have the inflated expectations of citizen competence that republicanism has. Citizens do not need to be knowledgeable, selfless, or seeking the common good. They just need to know enough to determine which party they like. Under this model, strong partisanship and polarization are not negatives, they are the norm. While I personally find liberal pluralism to be normatively unappealing, it is much more representative of the current state of our democracy, and it may be how we need to think of democracy moving forward. If citizens are unable to actually process arguments from the other side in a fair way, then deliberation is useless, and we should just allow citizens to form strong partisan identities without the tsk-tsking from political scientists, and to fight for their parties through elections.
Achen, Christopher. (1989). Prospective Voting and the Theory of Party Identification" Presented at the annual meeting of the American Political Science Association, Atlanta.
Achen, C. (2002). Parental Socialization and Rational Party Identification. Political Behavior, 24(2), 151-170. Retrieved October 8, 2020, from http://www.jstor.org/stable/1558353
Baker, C.E. (2002). Media, Markets, and Democracy. Cambridge University Press.
Bartels, L.M. (2002). Beyond the Running Tally: Partisan Bias in Political Perceptions. Political Behavior 24, 117–150. https://doi-org.proxy2.library.illinois.edu/10.1023/A:1021226224601
Downs, Anthony. (1957). An Economic Theory of Democracy. New York: Harper.
Gaines, Brian J., James H. Kuklinski, Paul J. Quirk, Buddy Peyton and Jay Verkuilen. (2007). “Same Facts, Different Interpretations: Partisan Motivation and Opinion on Iraq.” Journal of Politics 69: 957-74
Gerber, A., & Green, D. P. (1998). Rational Learning and Partisan Attitudes. American Journal of Political Science, 42(3), 794–818. JSTOR. https://doi.org/10.2307/2991730
Hogg, M. A., & Abrams, D. (1990). Social motivation, self-esteem and social identity. In D. Abrams & M. A. Hogg (Eds.), Social identity theory: Constructive and critical advances (pp. 28-47). New York: Harvester Wheatsheaf.
Huddy, L., Mason, L., & Aarøe, L. (2015). Expressive partisanship: Campaign involvement, political emotion, and partisan identity. American Political Science Review, 109(1), 1-17.
Kunda, Z. (1990). The case for motivated reasoning. Psychological bulletin, 108 3, 480-98 .
Lupia, Arthur & Mccubbins, Mathew. (1998). The Democratic Dilemma: Can Citizens Learn What They Need to Know? Cambridge University Press.
Mutz, D. (2001). Facilitating Communication Across Lines of Political Difference: The Role of Mass Media, with Paul M. Martin. American Political Science Review, March 2001, 95 (1): 97-114.
Oliver, J. E., & Wood, T. J. (2014). Conspiracy theories and the paranoid style (s) of mass opinion. American Journal of Political Science, 58(4), 952-966.
Taber, C. S., & Lodge, M. (2006). Motivated Skepticism in the Evaluation of Political Beliefs. American Journal of Political Science, 50 (3), 755-769.
Zaller, J. (1992). The Nature and Origins of Mass Opinion (Cambridge Studies in Public Opinion and Political Psychology). Cambridge: Cambridge University Press. doi:10.1017/CBO9780511818691.
Perhaps revealing my own confirmation bias, but we’ll let that slide for the purposes of this discussion. In fact, my behaviour reminded me a little bit of this classic Onion article which actually summarizes motivated reasoning far better than any academic text: https://local.theonion.com/man-leaves-comments-section-he-s-having-argument-in-to-1845244086↩
I follow the general historical trend of the literature as identified by Gerber and Green (1998, p.794-798), and Bartels (2002), with some additions of models which they neglected.↩
This is also called the instrumental approach by scholars like Huddy, Mason, and Aaroe (2015).↩
Although I suppose one does not actually elect a warlord in my black comedy (although not by much sadly) example. But even without voting I could see them arguing that the Democrats drove him to his warlordery.↩
This is less clear in 2020 than it was in 2007. A staunch Republican may believe, for example, that it is a fact that thousands of Muslims were celebrating in the streets of New York after September 11, as President Trump has detailed. A staunch Democrat, in contrast, may consider it a fact that President Bush assisted in the attacks of September 11 in order to go to war in Iraq to seize oil resources (see for example Oliver & Wood, 2014). This is an era of “alternate facts.”↩