Less than two weeks ago, a gunman armed with an AR-15 assault rifle killed five Dallas police officers. After a three-hour standoff with SWAT officers , the Dallas Police Department (DPD) sent in a remote-controlled robot carrying explosives that detonated and killed the gunman.

That’s the first time police have killed a suspect using a bomb-laden robot – and doing so has been widely debated since.

Some feel that the police response was completely reasonable, since the shooter had already killed several officers and was threatening to kill more. New York Police Commissioner William J. Bratton said, “This is an individual that killed five police officers … God bless ’em (DPD).” DPD Chief David O. Brown explained, “Other options would have exposed the officers to grave danger.”

Others felt that the DPD response was excessive, compared with such options as negotiating, waiting or even having a police sniper shoot the suspect. One New York Times article noted, “The decision to deliver a bomb by robot stunned some current and former law enforcement officials, who said they believed the new tactic blurred the line between policing and warfare.”

Many echoed these concerns. William Cohen, a former Exponent employee who helped design the MARCbot, wondered, “Where are the police going to draw the line when trying to decide between continuing to negotiate and doing something like this?”

Social science does not offer much research about what the public thinks are appropriate, reasonable, just and right responses by police and protesters to each other’s tactics. So we looked into it.

How we examined public opinion on protests and policing 

To find out what Americans believe is appropriate, last October we contracted with Qualtrics to collect data on a nationally representative sample of 1,200 respondents, including an oversample of African Americans numbering 600. We deployed a stratified quota sample matching census characteristics on sex, race (for non-African Americans), census region, and household income. The paper is here.

In a series of paired comparisons (of 21 police actions and 24 protester actions), we asked respondents:

Which action is more intense/severe: for the police to [insert random police action here] or for protestors to [insert random protester action here]?

We ended up with a comparison of every tactic from one side being compared with every tactic from the other side (the order of police and protester was also randomized). This enabled us to not only create a scale of what tactics are deemed more or less severe for protesters and police, but also to create a scale of what tactics are deemed to be more or less severe compared with what the other does.

We used this information to understand what people would call “proportional” response — where the action is roughly believed to be comparably severe — as opposed to “disproportional” response — where one side’s action is roughly believed to be far more (or less) severe than the other’s. The idea is that communities will support proportional responses as reasonable and just, but will reject and object to disproportional responses as unreasonable and unjust.

How do we calculate “proportionality”? Consider the following two actions: Police arrest people and protesters march in the streets. These two would be considered disproportional if, across our model’s simulations, police arresting people were consistently estimated to be more (or less) severe than marches.

The figure below shows which protester actions are considered proportional to which police actions, arrayed in increasing order of severity. To find out what is or is not considered proportional, identify a specific protester tactic (like taking hostages, which is third from the top) and then move along the row to identify whether the tactics on the bottom of the figure are seen as proportional or disproportional.

The light gray band (running from the lower-left to upper-right corners of the figure) identifies ranges of proportionality.

We were surprised by some of what Americans consider “proportional” policing 

What we see in the figure is that Americans classify many tactics on each side as proportional to one another and thus reasonable and just. For example, on the lower end of the scale, the police showing up in large numbers are believed to be proportional to a protest rally, marching in the street, yelling insults, blocking government buildings and disrupting speeches.

According to our respondents, if protesters are destroying police cars, throwing rocks and stones or Molotov cocktails, proportional police responses would include using false information, using batons and shields to beat down challengers, and, somewhat shockingly, using live ammunition to shoot people and forcibly disappearing people.

While finding it acceptable to shoot people is somewhat comprehensible given American familiarity with standoffs, SWAT teams and sniper fire, we did not expect to find respondents saying that it was acceptable to forcibly disappear individuals. Indeed, we wonder if respondents thought it meant that people were just being removed from a place or space and allowed to live, rather than being removed, never to return.

But while respondents often consider it appropriate for police to forcibly disappear people or shoot with live ammunition, they don’t feel the same way about torture – which they see as proportional only when a suspect had taken a hostage, set fires or blown up buildings. This contextualizes the findings of other research that suggests that support for torture is limited, albeit without offering the context in which torture would be used. To evaluate any side’s tactic within a dyadic conflict between challengers and governments, one needs to consider what the other side is doing.

Overall, Americans don’t like public protest

What you can also see in the figure is that the U.S. population seems quite hostile to protest, and quite willing to see police use lethal tactics in response. For example, our respondents thought torture would be a proportional response to looting; declaring martial law would be proportional to graffiti; and spying on people would be proportional to yelling insults.

Americans seem fine with police actions that far exceed what the law, jurisprudence or even police procedure would allow — tactics reminiscent of those the FBI used in its 1950s through 1970s COINTELPRO program, which spied on, infiltrated, and disrupted left-leaning antiwar, civil rights, and religious-based political groups.

Why are Americans seemingly so anti-protest and willing to let police engage in what many would classify as extreme police measures? This may have something to do with how challengers and challenges are portrayed in the media in particular or in culture writ large, or with a punitive sensibility that permeates through different institutions in America. The truth is, we do not yet know.

But Americans also believe that many protest tactics are reasonable

Of course, our respondents did feel that many protestor tactics were reasonable – and that an array of police responses would be disproportionately severe, unjust, and wrong. For example, a large police presence is believed to be disproportional to burning the flag, blocking streets or defacing property with graffiti.

The last one was surprising, given that respondents felt that martial law was an appropriate response to graffiti, but this is precisely where further digging into individual preferences will be useful. After all, ours is the first study of its kind.

Opinions vary based on demographics and attitudes

Of course, these opinions vary depending on group. For example, on average, African Americans tend to see all police actions as more severe than do white and Hispanic respondents. Respondents over 45 saw all police actions as less severe than those in the 18-to-24 age group.

Attitudes matter as well. Those who support the Black Lives Matter movement, and those who have had bad experiences with the police, are more likely to perceive all police actions as more severe than others. Respondents who feel threatened in their neighborhoods, in general, perceive police actions to be less severe than those who feel more secure. Those who identify as politically liberal perceive police actions as more severe than those who identify as conservative.

Overall, it’s easy to conclude that Americans will accept killer police robots in response to targeted assassination. If robots with explosives are used under different circumstances, however, this may shift.

Christian Davenport is a professor of political science at the University of Michigan as well as research professor at the Peace Research Institute Oslo. Find him on Twitter @engagedscholar.

David Armstrong is an associate professor of political science at the University of Wisconsin-Milwaukee.

Thomas Zeitzoff is an assistant professor in the School of Public Affairs at American University. Find him on Twitter @zeitzoff.