Political Vitriol and Political Violence

This is from a statement by the National Jewish Democratic Council, quoted in the New York Times:

It is fair to say—in today’s political climate, and given today’s political rhetoric—that many have contributed to the building levels of vitriol in our political discourse that have surely contributed to the atmosphere in which this event transpired.

The statement is careful, to be sure: “many have contributed” to a “discourse” that “surely contributed to the atmosphere” in which the shooting of Gabrielle Giffords and the others took places. This is a far cry from saying that vitriol led to violence. But it takes a step in that direction, and it is a step too far, given what we know at this point in time.

How might we approach this link between vitriol and violence from a social science perspective? According to Dick Armey, we shouldn’t really bother:

In the final analysis, when we get the final answer, why did this fellow do this? The answer will come from psychology, not from sociology or political science. If we really want to understand deviance and danger in this country, we should apply the correct field of study, the correct disciplines and tools of understanding with rigor and responsibility, not just exercising pop sociology out of our hip pocket.

I’m all for rejecting pop sociology and of course psychology is relevant, but I do think other social sciences can say something as well.

Is there any evidence that vitriol leads to violence? Yes. See this paper (pdf) by Nathan Kalmoe, a doctoral student in political science at the University of Michigan:

Does violent political rhetoric fuel support for political violence? Political leaders regularly infuse communication with metaphors of fighting and war. Building from theoretical foundations in media violence research, I field a nationally-representative survey experiment in which subjects are randomly assigned to different forms of the same political advertisements. I find that even mild violent language increases support for political violence among citizens with aggressive predispositions, especially among young adults.

Kalmoe exposed subjects to the text of two fictional political advertisements. Some subjects saw ads that use the verb “fight.” Some saw the one that used phrases like “stand up” or “work for.” Some saw a combination. Here is the text of one ad, with the variations made obvious:

Americans today are fighting/struggling to keep their jobs and their homes. All you ever asked of government is to stand on your side and fight/stand up for your future. That’s just what I intend to do. I will fight/work hard to get our economy back on track. I will fight/work for our children’s future. And I will fight/work for justice and opportunity for all. I will always fight/work for America’s future, no matter how tough it gets. Join me in this fight/effort.

Subjects were then asked questions about political violence, such as whether they agreed or disagreed with this statement: “Sometimes the only way to stop bad government is with physical force.”

  • Finding #1: The vast majority of people disagreed with these statements. That is, they did not support violence.
  • Finding #2: Seeing one or both of the “violent” political ads had NO overall effect on support for political violence.
  • Finding #3: Seeing violence political ads DID have an effect among those with a predisposition to aggression, as measured with a standard psychological battery. Among those with the greatest predisposition to aggression, being exposed to a violent political ad increased their support for political violence by about 20 points on a 100-point scale. Among those with the least predisposition to aggression, being exposed to a a violent ad actually decreased their support for violence.
  • Finding #4: This conditional relationship—between seeing violent ad and a predisposition to aggression—appears stronger among those under the age of 40 (vs. those older), men (vs. women), and Democrats (vs. Republicans).

Although this study concerns only attitudinal support for violence, not actual acts of violence, its findings seem, on their face, to suggest that Jared Lee Loughner could have been motivated by violent political rhetoric. Such rhetoric exists, obviously. Some of it concerned Giffords in particular. And Loughner was a man under the age of 40 with some apparent predisposition to aggression. But that’s not a conclusion we can draw at all, and this is why the National Jewish Democratic Committee’s comment struck me as going too far.

Again, social science can help—and especially conventional theories about how information affects attitudes and behavior. For information—vitriolic political discourse in this case—to influence Loughner’s attitudes and behavior, he would have had (1) to be exposed to that discourse and (2) to accept or believe what he was hearing. Was he? We do not know.

To prove that vitriol causes any particular act of violence, we cannot speak about “atmosphere.” We need to be able to demonstrate that vitriolic messages were actually heard and believed by the perpetrators of violence. That is a far harder thing to do. But absent such evidence, we are merely waving our hands at causation and preferring instead to treat the mere existence of vitriol and the mere existence of violence as implying some relationship between the two.

For more on this general subject, see this thread at Chris Blattman’s blog, John Pitney’s book on military rhetoric in politics, Lee’s old post on militant extremism, and my post on Scott Roeder. I thank Brendan Nyhan for reminding me of the Blattman thread.

Matt Grossman, who I thank for sending me the Armey quote, also sent along these possible topics, for which there might be some extant research of value:

1) the relationship between polarization and political violence
2) how individual attitudes change after losing competitive elections
3) the lack of ideological coherence in the political views of radicals
4) changes in the use of violent or military metaphors in politics
5) how radicals interpret political messages in the media
6) the relationship between polarization and civility by elites and political tolerance or the acceptance of violence

I welcome comments about these topics or anything else in comments.

13 Responses to Political Vitriol and Political Violence

  1. James January 9, 2011 at 4:34 pm #

    First of all, “I will fight hard to get our economy back on track” is hardly equivalent to a US Congresswoman stating “I want armed revolution” and “I want people armed and dangerous,” national political leaders broadcasting “Lock and load” and Obama is a marxist anti-colonialist Kenyan and an enemy of our country, and such. The Republicans have raised the level of vicious rhetoric far beyond “I will fight for your children’s future.” So I don’t think this study is all that relevant to what we are talking about.

    In addition, I doubt that any research can produce “proof” of causation — that’s just not something that controlled research is capable of doing. We can better look at history for comparable situations — the rightwing vitriol preceding the Kennedy assassination; the rightwing targeting of OB/GYNs and the assassination of Dr. Tiller and the murder of other doctors, the bombing of abortion clinics; the vitriol preceding the Oklahoma City bombing. Is there any “proof” that the former “caused” the latter? Of course not.

    Neither is it accurate or responsible to deny that this kind of addictive hate speech and action *might* influence unbalanced people into committing acts of violence. We already have that Nazi guy shooting the guard at the Holocaust Museum, the guy in Tennessee who shot up the liberal church, the guy in Oakland going after the ACLU.

    It’s entirely predictable that the Republicans will deny that their inflammatory tactics have any effect on mentally unstable people, but I frankly expected better than that from a political scientist.

  2. Zach January 9, 2011 at 6:24 pm #


    How about Pres. Obama’s comment that “if they bring a knife, we bring a gun” (paraphasing)? Did that not contribute to violent rhetoric? Did that not inflame mentally unstable people? Of course not, because it defies the ideological frame you have set up to define this situation.

  3. John Sides January 9, 2011 at 6:38 pm #

    James: You are asking me to make claims about causation — about how “inflammatory tactics” affect “mentally unstable people” — while admitting that we cannot make claims about causation. As a political scientist, that is something I will not do.

    And fair warning to James and Zach and all future commenters: I will — for the first time in the history of this blog — actually delete comments in this thread if they engage in more bickering about which party or politician has used violent rhetoric. There are plenty of other websites for shouting matches. This post is, to the best of my ability, about social science, and that is what this comment thread will be about as well.

  4. Emily January 9, 2011 at 8:15 pm #

    It’s interesting to try to contemplate the probable antecedents for incidents of individual violence, because of course so much of the work on political violence is about mass political violence. Mass political violence and extremist media get linked all the time. In the context of ethnic cleansing, for example, there’s never been much disputing that extremist media helped produce mass murder in Rwanda and the former Yugoslavia. Of course, there were substantially different political factors at work there as well.

    However, if most people aren’t going around killing their neighbors, and it’s just the occasional individual who goes on a politically-motivated killing spree, then you’re talking about minority of people who might be highly susceptible to extremist media influence, a tiny subset of the subset Kalmoe finds. At this point, you’re also in the territory of people who are probably susceptible to influence by non-political violent media as well. I also wonder about the effects of priming of the immigration issue in AZ, which has a heavy ethnic conflict component to it, and the salience of the gun debate in AZ.

    The sad irony of the general hypothesis implied by Matt’s research direction #2 is that we should now be entering a period where partisans on both sides should feel somewhat soothed by recent victory. I guess not in the political microclimate of AZ8.

  5. Nazgul35 January 9, 2011 at 9:49 pm #

    Ignoring the already ample amount of cases of violence in the past two years (as well as prevented acts) that are clearly drawn to the violent rhetoric being embraced by a political party (and movement) that isn’t named.

    I’ll just look at them and whistle.

  6. Chris January 9, 2011 at 9:53 pm #

    To add to the list of potential research topics raised by this event, I wonder if there a role being played by social norms in regards to assassinating members of congress? When I looked at the list of members of congress who have been assassinated, I found it interesting that we don’t see any members assassinated after RFK (that is, if we set aside the case of Leo Ryan as “special” since he was killed in another country), until this attempt on Giffords, about 40 years after RFK is killed. I wonder if it is possible that the assassinations that occurred during the 1960s (JFK, RFK, MLK, etc) imprinted that generation with a generally held societal norm against assassinating political leaders? In fact, looking even further back in history, we see that RFK’s assassination came about about 33 years after Long’s–again, potentially pointing to Long’s assassination having imprinted that generation with a societal norm against this form of political violence. (Additionally, the assassins in these three cases are all under the age of 30, so they would not have been imprinted by the norm generated by the prior assassination.) Of course, exploring this further would require data on any serious attempted assassinations; plus, having just these three cases is not much to work with. Moreover, we do see assassination attempts on the President fairly consistently. But, if there is a case to be made that dramatic assassinations imprint that generation with norms against assassinating political leaders, it would certainly be interesting to know how and why the experience causes the norm to generate and imprint (as well as not translate to later generations)–was it something about the government’s response to the events, or the media’s portrayal, etc.?–not only theoretically, but so that we can ensure today’s generation adopts a norm against participating in this kind of violence.

    P.S. I don’t mean the tone of my questions to be at all insensitive to these events.

  7. LFC January 10, 2011 at 10:56 am #

    The post says that “we don’t know” whether Loughner was exposed to vitriolic political discourse.

    We don’t? Has he been living under a rock or in a cave for the last 2 years? If not, I think it’s reasonable to assume he was exposed to the discourse. That’s not the same, admittedly, as showing it influenced him, but I do think we can assume exposure.

  8. Thomas January 10, 2011 at 11:37 am #

    LFC, not just two years, but nearly 10. Loughner’s high school friends describe him as a 9/11 Truther, so he certainly was exposed to some of the most vile political discourse we’ve ever seen in this country, left or right (though in this case left), and apparently he believed what he heard. We know that the 9/11 Truthers in Tucson confronted Giffords on more than one occasion, so perhaps we have the trigger. It’s as good a theory as we’re likely to find.

  9. Joshua Tucker January 10, 2011 at 4:56 pm #

    @Chris: One counter-point to this would be the assassination attempts against Ford and Reagan. So unless it’s a norm against _congressional_ assasination attempts, we can’t really speak of this kicking in until the 1980s.

  10. Joshua Fryer January 11, 2011 at 4:17 am #

    I believe you misrepresented the findings of this paper. My reading of it found that there was NO significant difference between men and women OR between republicans and democrats. The only statistically significant difference was found in age.

  11. JRon January 11, 2011 at 2:29 pm #

    It does seem it would not be inappropriate to also consider the effects of violent political rhetoric and propaganda in other countries as well. I realize there will be a different set of historical baggage in each case, but we as Americans are not so unique for that to be irrelevant.

    There are many many cases of political violence in other gun-filled countries that also follow demonization of public figures.

    Pretending that people here have a different ability to be influenced, especially the mentally disturbed, simply because we don’t consider ourselves to be a third-world country appears to me to be a sort if willful blindness to how humans everywhere can react and overreact.

    Our current sort of demonization by constant propaganda has recent precedents all over the world, none of which have ended prettily.

  12. Chris January 11, 2011 at 3:14 pm #

    @Joshua Tucker: Thanks for responding to my comment. I did point out this issue in my comment–that in fact we do see attempts against Presidents fairly consistently and that this could potentially counter the argument that a norm emerges against assassinating political leaders in each generation that experiences assassination. But, perhaps we could make a case that Presidents are “different” in some relevant and important ways than members of Congress? Maybe violence toward Presidents are “special” cases? So maybe a norm against violence directed toward congressional leaders exists, but does not extend to Presidents. And, again, this whole argument would also depend on whether or not there have been any serious attempted assassinations of congressional leaders, and I don’t have that data. But, I still think it is a potentially interesting perspective on why we don’t see more assassinations of congressional leaders.

  13. John Sides January 12, 2011 at 5:27 pm #

    Joshua Fryer: In discussing this part of the paper (“Finding #4” in my post), note that I wrote “appears stronger.” That was my colloquial way of conveying the statistical uncertainty underlying whether the conditional relationship between the treatment and aggression is different among age, gender, and partisan groups. I think I described the study accurately.