Association Arguments: Dog Whistling in the Dark

Daniel A. Bonevac

The current state of our political discourse might tempt any fair-minded observer to give up on the prospects for reasoned discussion, debate, and deliberation. All are essential to democratic decision making. Increasingly, however, arguments that sound not merely fallacious, unconvincing, incomplete, or unsound, but simply ridiculous, dominate our public discourse. They sound that way, moreover, to people across a fairly wide spectrum of political views. Their willingness to distance themselves from such argumentation encourages me to resist the temptation to give up on the prospects for rational deliberation. But restoring respect for reason will take more than expressions of dismay. We must identify the forms of reasoning that underlie bizarre-sounding arguments. There are such forms of reasoning. They lose power when, but only when, exposed.

The Association Argument

Let’s begin with a form of argument that has become pervasive over the past two years among editorial writers, television personalities, and guests. They use a form of straw man argument that leads to more or less any conclusion you like. Here’s the pattern of what I’ll call the association argument. A, someone the speaker doesn’t like, says “Some Fs are Gs,” where G is something bad. (“Some immigrants are criminals,” for example.) The speaker then “reasons”:

A says that some Fs are Gs.

A is associating Fs with Gs.

A is insinuating that all, or most, Fs are Gs, or at least that Fs are generally Gs.

A is insulting Fs qua Fs.

A is F-ophobic.

A is therefore morally untrustworthy.

A is also wrong: Fs are not generally, mostly, or all Gs.

What A says is “fake news.”

A is therefore epistemically untrustworthy.

Therefore, A merits both moral and epistemic condemnation.

This is quite a flexible form of argument. If some Fs are Gs, and all Fs are Hs, then some Hs are Gs. So, the argument form generalizes: A isn’t just attacking Fs qua Fs; A is also attacking Hs qua Hs! Here’s how it works with some specific examples:

Donald Trump says that some people from Mexico are bringing crime across the border. He is associating Mexicans with criminal behavior. He is thus insinuating that Mexicans are criminals. He is thus insulting Mexicans qua Mexicans. That is both racist and xenophobic. It’s also not true. Most Mexicans living in the United States are productive, law-abiding citizens. Trump is therefore a racist xenophobe who peddles fake news.


Trump says that some Muslims are terrorists. He is associating Muslims with terrorism. He is insinuating that Muslims are terrorists. He is thus insulting Muslims qua Muslims. That is both racist and Islamophobic. It’s also not true. Most Muslims living in the United States are productive, law-abiding citizens. Trump is therefore a racist Islamophobe who lies.


Trump says that some reporters are irresponsible, reporting fake news. He is associating the press with irresponsibility and dishonesty. He is insinuating that reporters are irresponsible and dishonest. He is thus insulting reporters qua reporters. That is anti-press, anti-First Amendment, and an attack on the Constitution. It’s also not true. Most reporters are responsible journalists. Trump is therefore a threat to the Constitution and has no respect for the truth.

Notice the general pattern. Someone says something a fair-minded observer might think is uncontroversially true. (Some Mexican immigrants commit crimes; some Muslims are terrorists; some reporters are irresponsible.) Somehow we get from there to the conclusion that the speaker is an immoral liar.

Of course, association arguments aren’t all directed at Trump. Some are directed at the Republican Party:

Republicans charge that some FBI officials are corrupt, motivated by political considerations. They are associating the FBI with corruption and political bias. They are insinuating that FBI officials are generally corrupt and biased. They are insulting FBI officials qua FBI officials. That is an attack on the FBI, on the Justice Department, and on law enforcement in general. It’s also not true. Most FBI agents are upstanding, fair, and even handed in performing their duties. Republicans are therefore a threat to the rule of law as well as to the truth.

Some association arguments assail other groups:

Climate skeptics allege that some climate scientists are using unreliable climate models with little predictive power. They are associating the climate scientists with unreliability and empirical inadequacy. They are insinuating that climate scientists are bad scientists. They are thus insulting climate scientists qua scientists. That is anti-science. It’s also not true. Most climate scientists do fine work. Climate skeptics are therefore a threat to science engaged in distortion.

This pattern of association argument doesn’t work, of course, if G is good. If Trump says that some immigrants are good people, even heroes, that never suggests that Trump is pro-immigrant or deserves any kind of praise. In fact, that he says that some immigrants are good people implicates that not all are, thus reinforcing the insinuation that immigrants engage in bad behavior. In general, the argument pattern applies only if A is from someone on the other side. I haven’t seen any of these:

Barack Obama says that some news outlets (e.g., Fox News) are irresponsible—so he is anti-press and a threat to the Constitution.

Democratic activists say that some laws are unjust—so they’re threats to the rule of law.

Feminists say that some women have false consciousness—so they’re anti-woman.

Social scientists say that some people are racist—so they are anti-human.

Association arguments seem clearly fallacious. It should be obvious that “Some Fs are Gs” does not imply that all Fs are Gs, that most Fs are Gs, or that Fs are generally Gs. “Most Fs are not Gs” therefore does not refute “Some Fs are Gs.” Indeed, outside of politics, no one would find such an attempted refutation convincing. Consider:

A: Some numbers are rational.

B: That’s absurd! Most numbers are irrational. You’re a threat to mathematics!

A: Some people who get the flu need to be hospitalized.

B: Most people who get the flu don’t need hospitalization. Why do you hate sick people?

B’s responses are obviously ridiculous in those contexts. In politics, however, such responses seem commonplace. What is going on here?

In part, I suspect, people reject any attack on or even qualification of their core values. Those on the Left ostensibly value respect for people, including and especially immigrants, Muslims, and other groups they see as having been treated as “the other.” So, they react strongly against any criticism of those groups, no matter how mild or reasonable those criticisms might be. In part, leftists have a Manichean worldview. They inherit and generalize the Marxist dichotomy of bourgeoisie and proletariat, dividing the world into oppressors and oppressed, privileged and marginalized. Leftists stand with the marginalized. Any criticism of the marginalized, no matter how mild, must be resisted. Leftists cannot admit that any significant number of immigrants are criminals; that any significant number of Muslims are terrorists; that any significant number of government officials are corrupt; that any significant number of climate models are flawed, and so on. To do so would detract from the Left’s paramount goal of redistributing power. All members of marginalized groups, as well as those who speak out on their behalf, must be viewed as exemplars of moral and epistemic excellence. All immigrants, documented or undocumented, must be seen as industrious pillars of American society. All Muslims must be seen as peaceful. Anything they say must be believed. Any apparent counterexamples must be discounted as the result of existing power imbalances or ignored.

Finally, the speed of the contemporary news cycle, and a desire to shape coverage and to discount opponents’ arguments before they get a chance to articulate them, encourages people on all sides of the political spectrum to use association arguments. Nuance is difficult and takes time to develop. Broad generalizations are easy and fit into soundbites and five minute interviews. A crude ad hominem, by itself, would be too transparent. But association arguments lead from seemingly reasonable claims (“Most immigrants are productive and law-abiding,” “Most FBI agents are fair and exceptionally dedicated”) to ad hominem conclusions.


Other factors, too, explain attraction to what appears to be an easily exposed straw man fallacy. Theorists on the Left have developed a theory of propaganda, insinuation, and dog-whistling that encourages associational forms of thinking.1 It relies on a baroque, if rather imprecise, theoretical apparatus. But the key moves are easily understood.

Jason Stanley, professor of Philosophy at Yale, understands propaganda as informational manipulation involving deception, misdirection, stereotyping, or appeals to emotion.2 The concept of stereotyping plays a large role in justifying association arguments. Much of our thinking, especially but not only in politics, uses stereotypes of groups playing especially salient roles. Those stereotypes are not always negative; many are positive or neutral. But they encapsulate generalizations about members of groups. We think about policies in part by thinking about their effects on various groups. We react to policy proposals by activating stereotypes. When Donald Trump says that some immigrants are criminals, the argument goes, that activates stereotypes of immigrants and criminals and associates the two. Someone primed by his speech thus thinks in terms of overlapping stereotypes for those groups, becoming more likely to favor immigration restrictions. Perhaps they even incorporate crime into their immigrant stereotype. This explains the first move in the argument pattern, from “A says that some Fs are Gs” to “A associates Fs with Gs.”

Stanley’s position also explains why this form of argument only applies if A is bad and G is negative. He finds propaganda objectionable only when it serves what he calls a flawed ideology, one that purports to justify inequalities. So A’s association of Fs with Gs is bad only if G is negative—thus implying that Fs deserve less than what we might initially have thought. If A is not a designated spokesperson for the Fs, by saying something negative about Fs the implication is that A or some other group is more deserving than the Fs.3

The difficulty with this attempted justification of the inference is that it encourages us to think in broad, stereotypic patterns. Eschewing nuance, we cannot consider, for example, that some immigrants are criminals who should be kept out of the country while others are valuable contributors who ought to be admitted. We cannot urge that press accounts be viewed with proper skepticism, trusted only once subject to careful scrutiny and confirmation. We cannot differentially assess the scientific foundation of climate models, arguing that some rest on more secure data sets and theoretical assumptions than others. We are forced to extremes.

More precisely, we are forced to extremes if we are not on the side of the person advancing the association argument. Stanley can use nuance, drawing careful distinctions, because he does so in the service of equality. His opponents, however, by definition draw distinctions in the service of what are seen as flawed ideologies. That only egalitarians are capable of nuanced thought emerges, absurdly, not as an empirical generalization but as an a priori truth.

A more refined justification for association arguments concerns insinuation and “dog whistling.” Sometimes we communicate a general public message as well as another, more controversial message that only a subset of the audience understands. In general, that’s surely correct. Advertisers understand the strategy well, devising ways of communicating something that a portion of the audience cares about without alienating the rest. Insinuation could provide some reason to use association arguments. Perhaps A says “Some Fs are Gs” but consciously means to insinuate that all, most, or typical Fs are Gs. Perhaps A has no such conscious intention, but speaks to an audience some members of which already believe that Fs are generally Gs. Then A’s speech, the argument runs, serves as a dog whistle. The dogs—in this case, those predisposed to believe that Fs are Gs—hears the stronger claim while most of the audience hears only the weaker one. On this view, responding “But most Fs are not Gs” does nothing to refute the weaker claim, “Some Fs are Gs,” that the larger audience hears. But it does address the stronger claim the “dog whistle” evokes in the subgroup predisposed toward that view.

A consequence of this analysis, of course, is that the response is bound to strike the larger audience as a straw man argument unworthy of respect. The response that the speaker is guilty of racism, sexism, xenophobia, etc., will likely strike the larger audience as histrionic. It might nevertheless be effective against the dog whistle.

But there is a deeper problem. Understanding what A insinuates to B through a given speech act is difficult, requiring a thorough understanding of the mental states of A and B. Leftists interpreting a conservative’s speech to a conservative audience can understand what’s being insinuated to some subgroup only if they have a deep knowledge of the beliefs, desires, fears, goals, and predispositions of each subgroup. And leftists, in general, do not understand what conservatives think or why they think it.4

Much of the current work on propaganda, dog whistling, and insinuation thus faces a problem. The general description of the phenomenon is fair enough. But the examples used to illustrate the analysis—all of which seem to be from the Right—are bound to strike conservatives as unconvincing. (Do leftists or even moderates never engage in propaganda or insinuation? Are no dogs Democrats?) Leftists project their own Manichean view onto their opponents. In their universe, you’re either for the oppressors or for the oppressed. Dog whistle analyses invariably attribute racist, sexist, xenophobic, Islamophobic, and homophobic attitudes to people on the Right, even when conservatives are not thinking in these terms. The Left is obsessed with those categories, and attributes the same obsession to the Right. The result is a crudely reductive analysis unrecognizable to anyone not on the Left of the political spectrum. The problem is fully general. Attributing insinuations requires a subtle comprehension of speaker and audience beliefs, desires, goals, presuppositions, and predispositions. Few people have that kind of understanding, especially of those who see the world very differently. Maybe Joe Wurzelbacher has it right: “If you can hear the whistle, you’re the dog.”5


Association arguments also find support in George Lakoff’s theory of framing.6 Lakoff, a linguist by training, presents a version of Georges Sorel’s idea that we think about politics primarily in terms of social myths, scripts, or narratives.7 Frames are mental structures associated with certain words, that shape how we think. Lakoff argues that conservatives and progressives use different frames for thinking about a variety of political issues. Those frames lead naturally to divergent conclusions about what ought to be done. Even that core idea lends support to association arguments, for frames consist of patterns of associated ideas. Someone pointing out that some Fs are Gs is activating a frame using the terms F and G and linking them. That brings along with it a way of seeing issues that includes many other things. So, when Republicans speak of illegal aliens, for example, they activate frames involving criminality (“illegal”) and foreignness (“aliens”), which provoke fear and suggest that it would be desirable to eliminate crime. Speaking of undocumented workers instead activates frames about paperwork (“undocumented”) and productivity (“workers”), suggesting that people themselves are good but face bureaucratic hurdles. From Lakoff’s perspective, it matters little what a speaker asserts about Fs and Gs; what matters is the use of those terms and the implications of their associated frames.

Lakoff sees conservatives and progressives as organizing the frames they use around two broad views of family relationships. Conservatives analyze politics in terms of a strict father model of the family; progressives, in terms of a nurturant parent model. This is already Manichean; Lakoff is on the side of nurturing, and he denies that moderates have any model at all. Notice Lakoff’s framing—the strict father versus the nurturant parent—and ask yourself why he doesn’t contrast the strict father with the nurturing mother, or present both in neutral terms, the strict versus nurturing parent. Doing either would suggest that any reasonable model would include both discipline and nurturing. Indeed, any reasonable understanding of government needs to understand the proper role for its coercive power—for law is essentially coercive—as well as the proper role for its development of people’s capacities and enabling of their exercise of those capacities. Government provides an infrastructure for liberty, and that includes establishing the rule of law as well as building or facilitating physical and social infrastructures. Nothing Lakoff says helps us to understand how to make the inevitable tradeoffs that requires. Instead, his approach implies that tradeoffs are unnecessary.

The crudity of association arguments is their virtue, from Lakoff’s point of view. They capture the core of a frame and expose it. His examples are not only biased but simplistic and polarizing (“War” vs. “occupation,” for example, or “entitlement” vs. “earned benefit”). Their empirical foundation is obscure. Lakoff seems to pull associations out of the air; his works are remarkably free of citations to any relevant research. John Parrott is right to worry that “‘framing’ is just another word for ‘propaganda.’”8 It fits Stanley’s characterization of propaganda as informational manipulation, in part through stereotyping, for frames are simply collections of stereotypes. Nuance, precision, evidence, and rational analysis have no place in Lakoff’s vision.


I have argued that association arguments rest on both a Manichean framework that divides the world into good and evil, oppressor and oppressed, privileged and marginalized, and on conceptions of propaganda, insinuation, and framing that suggest and even explicitly advocate such thinking. The results, for those not indoctrinated into one or both of these foundations of association arguments, sound bizarre. Jonathan Chait, discussing “dog whistle” theories that find racial subtexts in Republican positions on a wide variety of issues, observes:

the liberal racial analysis collapses onto itself . . .Impressive though the historical, sociological, and psychological evidence undergirding this analysis may be, it also happens to be completely insane.9

The lunacy of dog whistle logic is not the worst of it, however. It justifies marginalizing anyone who doesn’t use it. Politics has its own version of Gresham’s Law: bad argumentation drives out good. In this case, it does so intentionally, and thinks itself virtuous for doing it.

Suppose that it’s true that some conservatives hold the positions they hold for morally unsavory reasons. Association arguments allow us to infer that conservatives generally hold their positions for morally unsavory reasons. So, conservatives and their positions deserve moral as well as epistemic condemnation. Permitting conservatives to advance arguments for their views, no matter how anodyne, no matter how far removed from race, class, gender, etc., permits them to reinforce morally and epistemically unsavory attitudes. Permitting conservative speech is thus encouraging racism, sexism, xenophobia, anti-government, and anti-science prejudices. But we have an obligation not to encourage such prejudices. So, the leftist concludes, we have an obligation not to permit conservative speech. The same applies, of course, to the speech of moderates or even traditional liberals who would accept qualifications and complications of the binary, Manichean view.

Ironically, then, the left’s obsession with marginalization and its conception of propaganda lead not only to the suppression of speech but to the marginalization of anyone not on the Left. That has already happened in many institutions of higher education and government agencies. It is happening in Silicon Valley and throughout the corporate world, aided by human resources departments and compliance offices. The Left’s sympathy with the marginalized shows no sign of extending to those marginalized by their own intolerance.

The prospects for reason are dim, but they are not hopeless. Exposed for what they are, association arguments are easy to recognize as an enemy of reason. But our best hope lies in the absurdity of the conclusions to which they lead. They license inferences to almost any conclusion. Their very flexibility may be their undoing. No matter how “woke” you happen to be, eventually they will propose marginalizing you.

  • Share
Most Commented

April 16, 2021


Social Justice 101: Intro. to Cancel Culture

Understanding the illogical origin of cancel culture, we can more easily accept mistakes, flaws, and errors in history, and in ourselves, as part of our fallen nature....

April 19, 2021


Critical Race Theory and the Will to Power

A review of "1620: A Critical Response to the 1619 Project" by NAS President Peter W. Wood....

May 30, 2018


The Case for Colonialism

From the summer issue of Academic Questions, we reprint the controversial article, "The Case for Colonialism." ...

Most Read

August 23, 2021


Testing the Tests for Racism

What is the veracity of "audit" studies, conducted primarily by sociologists, that appear to demonstrate that people of color confront intense bias at every level of society?...

May 30, 2018


The Case for Colonialism

From the summer issue of Academic Questions, we reprint the controversial article, "The Case for Colonialism." ...

March 29, 2019


Homogenous: The Political Affiliations of Elite Liberal Arts College Faculty

A study on the partisanship of liberal arts professors at America's top universities. ...