The partisan contours of conspiracy theory beliefs

Abstract

The “conspiracy theories are for losers” argument suggests that out-of-power groups use conspiracy theories to sensitize minds, close ranks, and encourage collective action. Two necessary conditions of this argument are that (1) group members subscribe mostly to conspiracy theories that malign out-groups or bolster their in-group, and (2) group members must recognize whether conspiracy theories emanate from their own group, an opposing group, or are outside of partisan conflict. Using representative survey data from the 2012 Cooperative Congressional Election Study, we show that conspiracy accusations follow the contours of partisan conflict: partisans accuse opposing groups, rather than co-partisans or non-partisans, of conspiring. Using MTurk data, we show that partisans can differentiate between the conspiracy theories coming from members of each party. We suggest that many conspiracy beliefs behave like most partisan attitudes; they follow the contours of partisan conflict and act as calling cards that send clear signals to co-partisans.

Any complete account of contemporary American politics must consider how conspiracy theories shape individuals’ opinions and behaviors (Miller and Saunders, 2016). A growing body of research ties conspiracy beliefs to partisanship (Einstein and Glick, 2015; Hartman and Newmark, 2012; Pasek et al., 2015), consistently finding that partisans believe conspiracy theories accusing opposing groups (Miller et al., 2016; Oliver and Wood, 2014; Radnitz and Underwood, 2015). This said, there may be more to conspiracy theories besides a predilection for dubious ideas combined with animosity toward the opposition. One alternative account is that conspiracy theories follow a “strategic logic” and that they are for political “losers” (Uscinski and Parent, 2014). In their account, Uscinski and Parent argue that conspiracy theories resonate most when out-of-power, outsider groups (which they call losers) accuse in-power, establishment groups (“winners”) of conspiring. As they describe it, conspiracy theories must

conform to the present distribution of power to resonate widely. In this way, conspiracy talk has a strategic logic. Sharing conspiracy theories provides a way for groups falling in the pecking order to revamp and recoup from losses, close ranks, staunch losses, overcome collective action problems, and sensitize minds to vulnerabilities (p. 132).

But, in order for conspiracy theories to have this strategic logic they must function as signals, which in turn requires that partisans—at the very least—differentiate who is sending the signal and who is being accused of conspiring. However, no research to date investigates if and how well partisans can differentiate among conspiracy theories in this way. We attempt to answer two main questions about the relationship between conspiracy theories and partisanship. First, do partisans’ accusations of conspiracy follow the contours of modern party conflict and accuse the “correct” (i.e., opposing) groups of conspiring? If partisan motivated reasoning determines the conspiracy theories that partisans believe in, then partisans should (1) be more likely to accuse opposing partisan groups of conspiring, (2) be less likely to accuse co-partisan groups of conspiring, and (3) be less likely to accuse non-partisan groups of conspiring. Using representative survey data, we show that partisans do, in fact, focus their conspiracy beliefs on partisan out-groups.

Second, to what extent do partisans differentiate their party’s conspiracy theory signals from those emanating from the opposing party? If conspiracy theories are effective signals in partisan conflict, then partisans should be able to (1) declare “ownership” of co-partisan conspiracy theories accusing the opposition, (2) attribute ownership of conspiracy theories accusing co-partisans to the opposing party, and (3) not attribute co-partisan ownership to conspiracy theories not addressing partisan conflict. Using data gathered via MTurk, our results demonstrate bipartisan consensus about which party (if any) owns which conspiracy theories. This indicates that partisans can differentiate the embedded senders and targets in different conspiracy theories, suggesting that such theories could be used as signals.

Combined, our findings suggest that partisan conspiracy theories are not merely markers of psychological quirks, alienation, or psychopathology, but rather of shared core associations—similar to issue ownership (Petrocik, 1996)—with a major political party. Conspiracy theories clearly communicate partisan content and therefore could be used as strategic signals. The content embedded in conspiracy theories signals to partisans who the sender is, who the villain is, and what the potential danger is. This demonstrates the potential for conspiracy theories to generate collective action.

Partisanship and conspiracy theories

Miller et al. (2016) attribute the connection between partisan identity and conspiracy theory endorsement to partisan motivated reasoning. Partisanship is the “perceptual screen” through which the vast majority of Americans process political information (Campbell et al., 1960), and the foundation of perhaps the most important political identity one can take on. Group identities can push people to view their own group as upright and virtuous and opposing groups as biased and nefarious—even when there is little evidence to support such conclusions (Claassen and Ensley, 2016). It follows, then, that individuals are motivated by their partisan identities to believe in certain conspiracy theories because they impugn members of opposing groups.

The connection between partisan motivated reasoning and certain conspiracy beliefs is indisputable; however, motivated reasoning is a subconscious psychological process (Druckman, 2012). Individuals are not consciously choosing to discount incoming information that is not congruent with previously held beliefs, or to seamlessly integrate information that is congruent with previously held beliefs; rather, they are succumbing to an automatic psychological process. However, partisans may also consciously make associations between conspiracy theories and political parties, even conspiracy theories that their own “team” promulgates. Indeed, Uscinski and Parent (2014) argue that sharing conspiracy theories is a form of strategic communication and Atkinson et al. (2017) go further by suggesting that conspiracy theories send signals that can mobilize voters.

If conspiracy theories communicate partisan information, then it follows that partisans would be both willing and able to identify and “own” those conspiracy theories that denigrate opposing groups. This idea is similar to that of partisan issue ownership in that people associate issues with parties, regardless of their affect toward the party or their positions on the issues (Petrocik, 1996). The associations between issue and party can be positive or negative, but there is widespread, cross-party agreement on which issues go with which party. These associations are a basic reflection of partisanship coloring people’s information processing.

Motivated reasoning would not, on its own, account for this phenomenon. Just as Republicans would not be subconsciously motivated to cede ownership of educational issues to Democrats, Republicans would not be subconsciously motivated—particularly because of the negative connotation associated with believing conspiracy theories—to accept ownership of the “Birther” conspiracy theory. ¹ The key empirical question, then, is the extent to which partisans “own” conspiracy theories—a phenomenon that partisan motivated reasoning would not fully account for, but that the “conspiracy theories are for losers” theory would expect.

The theoretical connection between conspiracy beliefs and partisan motivated reasoning that others have suggested relies on the direct, immediate, and explicit association partisans make between a conspiracy theory and a political party or partisan identity (Miller et al., 2016). But, in order for conspiracy theories to function as signals, partisans must be able to differentiate who is sending the signal and who is being accused of conspiring. In other words, partisans should be able to tell from a conspiracy theory if it is one of their party’s, one of the other party’s, or if it has nothing to do with either party. However, no research to date investigates if they can explicitly, consciously associate conspiracy theories with political parties.

There are established theoretical reasons to suspect that people’s conspiracy beliefs do not follow partisan contours or that people would not “own” the conspiracy theories associated with their party. First, numerous studies associate conspiracy beliefs with pre-political psychological traits (Brotherton, 2015). If conspiracy beliefs are pre-political, then we would not expect conspiracy beliefs to follow clear partisan patterns. Second, conspiracy theories are negatively valenced and partisans have good reason to repudiate them: they are often referenced as paranoid, superstitious, and psychotic (Uscinski and Parent, 2014).

We examine the connection between conspiracy theories and the partisan groups that they are associated with to answer two questions implied but never explicitly tested in the prevailing theories of conspiracy beliefs: (1) Do partisans’ beliefs about out-groups follow the contours of partisan group conflict, and (2) can partisans differentiate among partisan and non-partisan conspiracy theories and take ownership of their party’s conspiracy theories?

Who is conspiring against us?

To begin, we consider the extent to which individuals—because of partisan motivated reasoning—view groups associated with the out-party as conspirators. If partisan motivated reasoning drives partisans’ accusations of who is and who is not conspiring, then we should observe that conspiracy beliefs follow the contours of party conflict. We present two null hypotheses and two alternatives:

• H _{null 1}: There is no difference between Republicans’ and Democrats’ likelihood of believing partisan groups are conspiring against them.

• H₁: Partisans are more likely to believe groups associated with the opposing party, as opposed to groups associated with their own, are conspiring against them.

• H _{null 2:} Partisans are not more likely to believe partisan groups are conspiring than non-partisan groups.

• H₂: Partisans are more likely to believe partisan groups are conspiring than non-partisans groups.

Survey data

To test these hypotheses, we employed data from the 2012 Cooperative Congressional Election Study survey (CCES). The CCES is nationally representative and fielded online by Yougov. The data we employed were collected during the month prior to the election. Respondents were asked to answer the following question: “Which of these groups are likely to work in secret against the rest of us? Check all that apply.” Respondents could select as many or as few of these groups as they liked. Response options included two options associated with the Republican Party and its partisan coalition groups: “Republicans or other conservative groups,” and “Corporations and the rich”; two options associated with the Democrat Party and its partisan coalition groups: “Democrats or other liberal groups,” and “Communists and Socialists,” and one option associated with no party “Freemasons or some other fraternal groups.” ²

The only other variable employed in this analysis is partisanship. Partisanship was measured using responses to the standard two-question measurement strategy. We consider partisan leaners to be partisans. ³

Results

We began our analysis with a test of the differences in the proportions of self-identified Democrats and Republicans who designated each group as likely to work in secret against the rest of us. A dotplot of average proportions for Democrats and Republicans, along with 95% confidence bands, is depicted in Figure 1. ⁴

Figure 1.

Proportion of Democratic and Republican respondents who believe that the following groups “are likely to work in secret against the rest of us.”

Beginning at the top of Figure 1, only 8% of Republican respondents believe “Republicans and other conservative groups” are conspiring, but 37% of Democratic respondents believe as much. Approximately 26% of Republican respondents believe “Corporations and the rich” are conspiring, but 57% of Democrat respondents believe such. As for Democratic groups, 39% of Republican respondents believe “Democrats and other liberal groups” are conspiring, while only 6% of Democrats do. Approximately 59% of Republican respondents believe “Communists and Socialists” are conspiring, while only 20% of Democratic respondents do. These findings allow us to reject our first null hypothesis that there is no difference between Republicans’ and Democrats’ likelihood of believing partisan groups are conspiring in favor of the alternative.

Unlike with the four partisan-linked groups, Republican and Democratic respondents are equally likely to suspect Freemasons and other fraternal groups of conspiring. Indeed, 13% of Republicans and 11% of Democrats suspect the Freemasons of conspiring. Even though 8% of Republicans believed “Republicans and other conservative groups” and 6% of Democrats believed that “Democrats and other liberal groups” are conspiring, partisans are still more likely to believe that the four other groups in Figure 1 are conspiring than they are to believe as much about the Freemasons or other fraternal groups. This allows us to safely reject our second null hypothesis in favor of H₂.

These findings provide robust evidence that partisanship guides the individual attribution of conspiratorial intent to political groups. We, therefore, find support for our hypotheses that there are partisan differences in the likelihood of believing partisan out-groups are conspiring. This extends previous research on the impact of partisan motivated reasoning on conspiracy beliefs by demonstrating the extent to which it guides individuals’ perceptions of who is conspiring against them. We now consider if partisans “own” their conspiracy theories.

Do partisans “own” their conspiracy theories?

Rather than probe individuals’ beliefs about groups, we confronted individuals with conspiracy theories and asked them which party was most likely to promote it. Our specific null and alternative hypotheses were as follows:

• H_null3: Partisans will not connect conspiracy theories to the party whose members tend to believe them.

• H₃: Partisans will connect conspiracy theories to the party whose members tend to believe them.

If we found support for H₃, then it would suggest that conspiracy theories can act as signals and potentially encourage collective action. It would also suggest that partisans, more than able, are willing to “own” their co-partisans’ conspiracy theories.

MTurk Data

This data came from a sample of 1543 US adults gathered via the MTurk platform from April 19^th to April 22^nd, 2016. Individuals were offered $0.80 to complete the survey; average time to completion was nine minutes. Typical with MTurk samples, respondents were slightly more liberal, educated, and younger than the adult US population; however, there is little reason to suspect that a more representative sample would yield significantly different results (Berinsky et al., 2012).

Respondents were provided the following: “Below is a list of controversial ideas related to American politics. Please indicate whether you think each idea is more likely to be promoted by Democrats, Republicans, or neither Democrats nor Republicans.” A list of eight conspiracy theories that have been employed in recent studies followed (e.g. Miller et al., 2016). Three of these are believed largely by Republicans, three largely by Democrats, and two by small numbers of both parties.

Results

Table 1 includes the proportion of Democrats and Republicans who attribute each conspiracy theory to Democrats, Republicans, or neither. Conspiracy theories regarding Obama’s birthplace, death panels, and global warming are all perceived by large majorities of both parties as being more likely to be promoted by Republicans than by Democrats. Members of both parties were nearly evenly split in assigning ownership of the non-partisan Kennedy assassination and “chemtrail” theories. Conspiracy theories regarding the 9/11 terror attacks, invasion of Iraq, and Hurricane Katrina are perceived by substantial majorities of both parties as being more likely to be promoted by Democrats than Republicans. We reject the null hypothesis in favor of H_3.

Table 1.

Percentage of self-identified Democratic and Republican respondents attributing a given conspiracy theory to Democrats or Republicans. Cell entries are proportions with 95% confidence intervals in brackets.

	Conspiracy Theory	Democratic Respondents			Republican Respondents
		Democrats	Republicans	Don’t Know/Neither	Democrats	Republicans	Don’t Know/Neither
Republican	Barack Obama was born outside of the United States.	2.88 [1.88, 4.38]	93.83 [91.82, 95.36]	3.29 [2.21, 4.86]	3.85 [2.19, 6.67]	84.94 [80.50, 88.51]	11.22 [8.15, 15.25]
	The healthcare law passed in 2010 authorized the use of death panels to make end-of-life decisions for people on Medicare.	9.79 [7.72, 12.31]	73.39 [69.86, 76.64]	16.82 [14.13, 19.89]	13.06 [9.51, 17.68]	60.07 [54.05, 65.80]	26.87 [21.87, 32.53]
	Global warming is a hoax, perpetuated by environmental scientists who have their own political agendas.	3.81 [2.62, 5.50]	84.46 [81.59, 86.95]	11.72 [9.54, 14.31]	6.62 [4.30, 10.06]	69.54 [64.08, 74.49]	23.84 [19.35, 29.00]
Non-Partisan	Lee Harvey Oswald did not act alone in assassinating John F. Kennedy.	15.81 [13.05, 19.01]	15.12 [12.42, 18.27]	69.07 [65.18, 72.70]	21.54 [16.93, 26.99]	14.62 [10.80, 19.49]	63.85 [57.79, 69.50]
	Vapor trails left by aircrafts are actually chemical agents deliberately sprayed in a secret program directed by government officials.	16.01 [13.34, 19.08]	20.29 [17.31, 23.61]	63.71 [59.86, 67.38]	24.82 [20.03, 20.31]	13.50 [9.92, 18.11]	61.68 [55.75, 67.28]
Democratic	Senior officials had knowledge of the 9/11 terrorist attacks before they occurred.	30.41 [26.99, 34.06]	16.74 [14.06, 19.81]	52.84 [48.98, 56.66]	42.66 [37.01, 48.50]	13.64 [10.11, 18.15]	43.71 [38.03, 49.55]
	The US invasion of Iraq was not part of a campaign to fight terrorism, but was driven by oil companies and Jews in the US and Israel.	48.36 [44.58, 52.15]	14.48 [12.00, 17.35]	37.16 [33.57, 40.90]	61.70 [55.86, 67.22]	12.41 [9.03, 16.83]	25.89 [21.08, 31.35]
	The federal government intentionally breached flood levees in New Orleans during Hurricane Katrina so that poor neighborhoods would be flooded and middle-class neighborhoods would be spared.	37.69 [34.01, 41.52]	11.21 [8.99, 13.90]	51.09 [47.21, 54.95]	48.94 [43.14, 54.78]	10.92 [7.77, 15.13]	40.14 [34.56, 45.99]

Each of the Democratic conspiracy theories had much higher “neither” and “don’t know” rates than did the Republican conspiracy theories. This may be attributed to the fact that Republican conspiracy theories were simply more relevant when the survey was fielded, because a Democrat occupied the White House. Also, fewer individuals felt confident in assigning a particular political party to either of the non-partisan conspiracy theories, and therefore answered “don’t know” or “neither.”

Interestingly, the rates at which partisans actually believe in each of the conspiracy theories employed are much lower than the rates at which they perceive one party or the other as being associated with the conspiracy theories (see appendix). For example, 23% of Americans (42% and 10% of Republicans and Democrats, respectively) believe to some degree that Obama was born outside of the US (Miller et al., 2016), but 94% of Democrats and 86% of Republicans believe that Republicans “own” this theory. Partisans do not need to agree with the conspiracy theories to recognize it as their own.

Discussion

What role do partisan identities play in shaping conspiracy beliefs, and conversely, what role do conspiracy theories play in motivating partisanship? Using two unique measurement strategies we show that although conspiracy theories are often attributed to cognitive hiccups, psychological traits, or psychopathologies, they actually follow the contours of more familiar partisan battles in the age of polarization.

What is the explicit role of partisanship in driving conspiracy beliefs? We find that individuals attribute conspiratorial activities to partisan out-groups. Democrats, for example, believe that Republicans and their coalition are conspiring against us, while Republicans believe as much about Democrats. Conversely, partisans are unlikely to believe that groups that are uninvolved in partisan conflict are conspiring against them. Not only are partisans unlikely to believe that non-partisan groups (e.g. Freemasons) are conspiring against them, but they are also unlikely to assign any partisan “ownership” of non-partisan conspiracy theories (e.g. vapor trails). However, partisan ownership of specific conspiracy theories is actually bipartisan: individuals are very likely to recognize co-partisan ownership of specific conspiracy theories. For example, both Democrats and Republicans recognize that Republicans frequently propagate conspiracy theories denying Barack Obama’s citizenship and the veracity of global warming.

Taken together, our findings suggest that partisans understand which conspiracy theories are “owned” by which party—they know where partisan battle lines have been drawn. Further, conspiracy theories may be an important component of the identity of the two parties: believing that climate change is a hoax is part of what it means to be a Republican, just as believing that the Iraq War was driven by oil companies is a significant component of Democratic identities for many. That members of partisan groups are so willing to admit their own party’s culpability in promoting certain conspiratorial ideas suggests these beliefs are not a marker of conspiracism, but partisanship. Many conspiracy theories function more like associative partisan attitudes than markers of an alienated psychology.

Our second major contribution is in drawing attention to the scientific measurement of conspiratorial thinking via specific conspiracy beliefs. There is a sharp divide in the partisan nature of individual conspiracy beliefs when the individuals and groups at the center of a given conspiracy are or are not attached to a specific political party or ideology. Taken together, we have evidence that the conspiracy theory category can be broken down into at least two types: partisan and non-partisan. Beliefs in partisan conspiracy theories are highly correlated with partisanship, while beliefs regarding non-partisan conspiracy theories are not. These types of conspiracy theories may be more related to generalized conspiracy thinking. That the correlates of conspiracy theories may vary by conspiracy “type” should cause scholars of conspiratorial thinking to be cautious in their choice of conspiracy beliefs to query on surveys.

Our final contribution relates to correcting people’s conspiracy beliefs. Research shows that attempting to correct partisan conspiracy beliefs with new information not only works poorly, but may strengthen such beliefs (Nyhan and Reifler, 2010). Given that these conspiracy beliefs are intertwined with partisanship, it may be necessary to reverse the partisan cues driving individuals toward these beliefs. Our results provide additional support for Berinsky’s (2017) finding that Republican beliefs about death panels could be reversed by telling Republicans that their own elites disavow the belief. Our results suggest that researchers may be able to reverse partisan conspiracy beliefs by unattaching them from individuals’ partisan identities, as Berinsky (2017) did.