sign in sign up
<<
Home , Militarization, Militarization of police, Riot, SWAT

ftutbtenctzn n a nocmn nhgl policed www.pnas.org/cgi/doi/10.1073/pnas.1805161115 highly in enforcement law and racial citizens of between policing trust levels already-anemic of the the eroding in potentially these (11–13), applied minorities that allege disproportionately critics officer are while enhance (10), tactics units violent deter accounts. and journalistic militarized safety and that crit- absence anecdotal and argue the by advocates Proponents In informed both largely 9). of are (8, arguments ics the constraints analysis, data scientific police by of of study hampered units the States militarized been all), United their at has them of the document activities in they and (if agencies presence enforcement the law document thou- local way in the of in prevalent heterogeneity sands more of are Because color. they of whether of communities or consequences tactics the police demonstrating militarized systematic little is control social exercises (7). state minorities the racial which over by means another this as militariza- to police tion characterized due have part maintain scholars In race to critical (3–6). served history, hierarchies have social drugs” class-based on and “ race- the like efforts and policing frisk” which and in “stop ways interven- the the documented in have earlier, in research years of forces” ing century decades “occupying And a as (2). half communities police these Black urban than color, described More of Baldwin citizens surprising. James many less of to scholars were perhaps to images and But policing, force. and occupying race wartime a into by transformed deployed armor, (1). battle and agencies in police Black local clad unarmed and officers state at images and aim broadcast , taking enforcement’s brown networks law vehicles, news of armored national character of days, the For by response. alarmed and surprised A reputation bureaucratic militarization police liberties civil choice. and false that safety a suggest public is results between the trade-off policing, often-cited militarized mass the of the case in reputation the police In diminish public. may militarized reports seeing news oversample that in large show police respondents—I a American includes using African Finally, which of crime. of local reduce experiments—one or survey safety policing officer militarized enhance that to on demonstrate control-fails data I panel after militarization, nationwide police even using Further, local residents, rates. with crime American local communities for African in ling milita- deployed of that often shares census show more large are geocoded I units rare Maryland, and police from a minorities rized deployments Using racial team enforcement. SWAT at while law of targeted crime, in are trust violent tactics erode deters these and allege milita- officers claim critics protects Advocates policing enforcement. the law rized over local concern of in widespread units militarization stoked 2018) police 24, has armed March communities review heavily for American of (received presence 2018 visible 2, July increasingly approved The and IL, Evanston, University, Northwestern Hagan, John by Edited 08544 NJ Princeton, a reputation Mummolo police Jonathan harm reduce may or but safety crime police enhance to fails Militarization eateto oiis rneo nvriy rneo,N 84;and 08544; NJ Princeton, University, Princeton , of Department u ept rlne n ioosntoa eae there debate, national vigorous and prolonged a despite But suddenly have to appeared police American people, some To hoigo ihe rw n21,mn mrcn were Americans many 2014, in police Brown the Michael to of MO shooting Ferguson, in marched thousands s | ulcsafety public a,b,1 | crime | aeadpolicing and race | b odo isnSho fPbi n nentoa far,PictnUniversity, Princeton Affairs, International and Public of School Wilson Woodrow nruhy900U a nocmn gnis swl sthe as well as agencies, enforcement original law an US use 9,000 teams SWAT for then roughly active controlling of I in presence after the rates. measuring even crime panel nationwide and including geography indicators holds of social that levels high relationship with multiple a communities at Americans, in African deployed militarized of often that concentrations more show I are deploy. units characteristics they police the which and in communities used the are tactics” of units and police ways weapons militarized the which “special characterize in to of Maryland census in esti- rare deployments team a to (SWAT) of use perceptions and public first and I policing safety, police. officer crime, militarized on effects by its mate affected communities the claims. these depend of power validity state empirical of the exercise for on the crucially militarization and safety, police public of rights, civil implications The (14). communities 1073/pnas.1805161115/-/DCSupplemental at online information supporting contains article This Dataverse Harvard in deposited been 1 under have materials distributed replication ( All is deposition: Data article BY-NC-ND) (CC 4.0 access License NonCommercial-NoDerivatives open This Submission. Direct PNAS a is article This interest. of conflict no declares author and The data, analyzed research, paper. performed the research, wrote designed J.M. contributions: Author more police for seeing support that depress and show crime I of police— perceptions reports, inflate equipped can traditionally news to in officers—relative militarized officers police of at rates assaulted. the or lowers killed or are rates crime officers local which reduces deploy- or team obtaining SWAT that a over evidence ing no Using concerns find I safety. mitigate bias, variable officer greatly omitted promotes that and militarized comparisons whether rates within-agency test crime to lowers data, policing deployment SWAT Maryland https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/VYPUVC mi:[email protected]. Email: iiaie oiigmyb nteitrs fbt oieand police both of interest the curtailing in citizens. enforce- be that law may suggest policing toward findings militarized opinion these together, erodes Taken reports ment. news militarized in seeing police average. that on suggest reduction, experiments crime survey in violent However, or benefits claims safety officer detectable to of no terms and—contrary administrators—provide color, often police of more by communities are administrative teams in of (SWAT) deployed tactics” militarized array that and show an weapons I “special experiments Using original costs and limitations. sources the unclear data data remain but practices to safety, police public due controversial and of trade- benefits liberties a and as civil framed between often off are including policing, tactics, “militarized” police so-called heavy-handed over debates National Significance hssuylvrgspeiul nvial aat describe to data unavailable previously leverages study This ial,uigsre xeiet htrnol sinimages assign randomly that experiments survey using Finally, y . Cetv omn Attribution- Commons Creative www.pnas.org/lookup/suppl/doi:10. NSLts Articles Latest PNAS . | f6 of 1 ).

POLITICAL SCIENCES funding and presence. This analysis includes a large oversample panel includes roughly 9,000 agencies, each observed in 2000, of African American respondents—an important feature given 2004, and 2008. About 29% of agencies vary on SWAT sta- the high rate at which militarized police units deploy in Black tus during this period (SI Appendix, Table S2). I then use an neighborhoods. agency-month panel in Maryland to test whether the increased On average, militarized police units do not appear to pro- deployment of SWAT teams affects crime and officer safety vide the safety benefits that many police administrators claim. outcomes. And police may suffer reputational damage when they deploy To estimate the effects of militarized policing on public militarized units. These results suggest that the often-cited trade- perceptions of , I conducted two survey experi- off between public safety and civil liberties is, in the case of ments: one using a convenience sample from Amazon’s Mechan- militarized policing, a false choice. ical Turk (n= 1,566) and one conducted by Survey Sampling International (SSI) (n= 4,465). (Survey experiments were Defining Police Militarization approved by Stanford University’s Institutional Review Board, Police militarization is a continuum defined by a combination of protocol no. 32534. All respondents supplied informed .) equipment, tactics, and culture that centers on violent conflict (7, These experiments provided brief news articles accompanied 15, 16). In recent decades, local police agencies have militarized by a randomly assigned image that conveyed different levels their departments to varying degrees, adopting weapons, attire, of militarization, allowing for an estimate of the causal effect tactics, and organizational structures developed for theaters of of seeing militarized police on attitudes toward law enforce- war. The proliferation of militarized policing is due in part to ment relative to seeing more traditionally equipped police forces. an expansion of the and federal initiatives that Because of the prevalence of SWAT deployments in Black com- supplied localities with excess equipment and funds to munities, the SSI data include an oversample of roughly 1,850 purchase arms (17, 18). Heterogeneity in agency capacity makes African Americans to test whether treatment effects varied with it difficult to precisely code police agencies as “militarized” or respondent race. not. Recently publicized data on military gear disbursements have been used in some studies to estimate the effects of mili- Where and Why SWAT Teams Deploy tarization on police , crime, and officer safety (19–21). The rare census of SWAT activity available in Maryland offers But these data convey only the receipt of equipment from one of a valuable opportunity to study the stated reasons for, and geo- several programs that help supply agencies with militarized gear graphic correlates of, militarized policing. Maryland also exhibits (22). The data also appear incomplete (see SI Appendix, section large variation in the racial composition of localities [e.g., the 1A for details). percentage of Blacks in zip-code tabulation areas (ZCTAs) As an alternative approach, this paper analyzes the effects of ranges from 0% to more than 95%], allowing for an evaluation a substantively important threshold on the militarization con- of the relationship between militarized tactics and neighborhood tinuum: the use of SWAT teams. Both popular and scholarly racial composition with minimal extrapolation. debates over police militarization have focused on the activi- Before examining that relationship, Table 1, generated using ties of SWAT teams, their pronounced role in conducting the the pooled Maryland SWAT census, first displays the reasons drug war (17), and their high-profile crowd-control efforts (23). SWAT teams are deployed. Table 1 shows that roughly 90% SWAT teams often receive advanced combat training and exhibit of SWAT deployments in that state over 5 fiscal years were a command structure modeled on military special forces units conducted to serve search warrants. Previous work has shown (15). In general the formation of a SWAT team represents a that the use of SWAT teams to serve warrants, a practice heightened commitment to the use of militarized equipment and which escalated as a result of the war on drugs (17), is an tactics. extremely disruptive event in the lives of citizens and often While it is plausible that SWAT deployments deter violent involves percussive , battering rams, substantial prop- crime—and, in turn, improve officer safety—previous research erty damage, and in rare cases deadly altercations stemming on other heavy-handed tactics, including some that were much from citizens’ mistaken belief that they are experiencing a home more widely applied such as stop and frisk, has found little evi- invasion (28, 29). Table 1 also shows that less than 5% of dence of resulting crime reductions (24–26). It is therefore cru- deployments involved a “barricade” scenario, which typically cial to empirically evaluate the assumed benefits of militarized involves an armed suspect refusing to surrender to police. Vio- policing. lence to people and animals is rare, and gun shots are fired 1.2% of the time—roughly 100 deployments during this period. Data and Methods While the data suggest that indiscriminate violence is less com- To characterize where and why SWAT teams deploy, I obtained mon than some anecdotal reports suggest, they also show that data on every SWAT deployment in the state of Maryland over a the vast majority of SWAT deployments occur in connection 5-y period via a public records request. These data exist because with nonemergency scenarios, predominately to serve search of an unusual statute requiring every Maryland agency to uni- warrants. formly record all SWAT activity. Because the statute has since sunset, the data represent a rare, complete accounting of mil- itarized police units’ activities and contain the date, postal zip Table 1. SWAT deployments, Maryland FY2010–2014 code, and agency of each SWAT deployment between FY2010 Reason, legal authority % Outcome % and FY2014, as well as the reasons for and outcomes of each 91.06 Property taken 84.38 deployment (n ≈ 8,200 deployments). Barricade 4.92 Forcible entry 68.36 To estimate the effects of police militarization on crime and Other 2.67 63.69 officer safety, I use a nationwide panel measuring the pres- 0.89 Shot fired 1.20 ence of active SWAT teams. I generated the national panel by Exigent circumstances 0.45 Person injured 1.15 merging (27) three waves of the federal Census of State and Officer injured 0.42 Local Law Enforcement Agencies (CSLLEA) surveys—which Animal killed 0.17 measure whether agencies supply SWAT services—with FBI Animal injured 0.12 data on violent and the FBI’s Law Enforcement Offi- Person killed 0.11 cers Killed and Assaulted (LEOKA) database. The resulting

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1805161115 Mummolo Mummolo panel nationwide a use I claims, these and test To crime (10). violent police deter protect tactics militarized that claim Proponents Safety Officer and Crime militarized on Effects experiencing of risk pronounced Black policing. that a claim descriptive face the residents with However, consistent generaliz- are comparisons. causal results when a cross-sectional these infer warranted these confidently is from cannot also caution relationship We statistically so Maryland. not beyond state, ing is one comes slope analysis only This the controls. from with although model county level, the in county significant the at tern this 10.53% 1, during a Fig. residents with 100,000 period. associated per is deployments ZCTA SWAT in a increase in Blacks of percentage a the in 1, activity Appendix Fig. SWAT SI in of slope volume The the ZCTA. predicts 1, strongly (Fig. still indicators rates Blacks social crime aforementioned local the and ZCTAs. for to controlling mapped George’s be after can Departments—which Even Prince that data Police the crime incident-level County publish and Montgomery Department and County Police City Baltimore after Right household 1, dle even and (Fig. correlation, education, levels positive unemployment, income local strong for a controlling shows and per Maryland deployments logged. both SWAT 1, FY2010–2014, of during shows Fig. in unit volume that 1 residents the in Fig. Black residents and claim, 100,000 of unit this percentage geographic racial test with the To a of line between 30). treatment in relationship 6, (4, the be disparate police would indicating by that minorities evidence pattern a of color, decades of communities in scales. log and on plotted weighted are Locally Variables shown. unit. are fits geographic regression corresponding linear in indicators social ( agencies. Maryland ties. large three ( from Americans. ZCTAs African of trations 1. Fig. rtc leeta iiaie atc r oeotnapplied often more are tactics militarized that allege Critics Left WTdpomnsaemr rqeti ra ihhg concen- high with areas in frequent more are deployments SWAT o Left Top ounsosbvraerelationships. bivariate shows column sgnrtduigdt rmtrelreagencies—the large three from data using generated is alsS n S9 and S8 Tables , otmLeft Bottom and o Right Top o Right Top and Top ideRight Middle ideRight Middle otmRight Bottom mle hta1%ices in increase 10% a that implies ) o)Mrln CA.( ZCTAs. Maryland row) .Fg 1, Fig. ). sgnrtduigZTsin ZCTAs using generated is Bottom Right ideLeft Middle ( β ,tepretg of percentage the ), hw iia pat- similar a shows 1 = o)Mrln coun- Maryland row) ouncnrl for controls column . 05 , Middle P and < 0 . Mid- 001 row) ; hr sltl aito nteeotoe oepan.Estimates (i.e., explain). rare to so outcomes are these they in because variation little partly is zero, there near insignificant and statistically felonious, precise and a accidental are to both deaths, shrinks differen- officer for effect trend Estimates 4.3%. to the agencies time, allow over we tially once promoting in However, and trust (14). to investigations lowering crime violently criminal policing hindering militarized reacting thereby with suspects police, or consis- of (29) in is teams evidence top result SWAT increase anecdotal This The 6.5% with average. measures. on tent significant safety crimes, statistically violent officer within-agency a and implies crime coefficient violent on team of influence the reduce specifications). to (see logged compar- values are cross-sectional extreme outcomes plague All that (31). to bias isons agencies variable concerns each comparing omitted mitigate greatly By allowing over approaches time. also both time, over but over uniquely themselves effects and trend fixed effects to fixed including agency period model mod- time second two and a estimate (DID) agency difference-in-differences I violent containing generalized robustness, estimator on a gauge outcome: focus policing per To militarized els I safety. between link outcomes. public deployment asserted and these increased the of the affects because whether agency-month crime teams violent an test SWAT on and to team of assaults Maryland SWAT and in a fatalities panel acquiring officer of and effects crime the estimate to bann WTta ncieadofie aey ucmsaelogged. are Outcomes safety. intervals. officer confidence and 95% of crime are effect Bars on the team estimating SWAT regressions a from obtaining coefficients are Shown safety. cer 2. Fig. a acquiring that safety. officer evidence promotes no or crime is lowers there indistin- team SWAT and often and zero, positive from In generally signifi- guishable specification. are statistically trends effects time estimated a result the sum, this in shows although significance assaults, statistical one noninjurious misses and in increase noisier, 3.2% cant are assaults for i.2dsly siae fteefc fotiigaSWAT a obtaining of effect the of estimates displays 2 Fig. bann WTta osntrdc rm ripoeoffi- improve or crime reduce not does team SWAT a Obtaining IAppendix SI etos2–4 sections , NSLts Articles Latest PNAS fralternative for | f6 of 3

POLITICAL SCIENCES image randomly varied across respondents (Fig. 4). Images were randomly assigned within racial groups of respondents in the sec- ond survey (which contained an oversample of African American respondents), with equal probability. The control image (Fig. 4A) featured five male, traditionally uniformed officers (e.g., blue uniforms, brimmed caps, and standard side arms). The “low” militarization conditions (Fig. 4B and Fig. 4C) showed five male officers with “ gear” and batons or with and assault rifles. The “high” militarization condition (Fig. 4D) added an armored vehicle. (Because it elicited very similar treatment effects to the other low militarization image in Fig. 4C, the Fig. 4B image was dropped from the SSI survey to enhance statistical power.) In each case, the caption beneath the photo read, “Above: Five city police officers stand guard during a local protest.” All images were tightly cropped to ensure that any differences in responses were due to the appearance of police officers and not the sur- rounding area in which they were deployed. Following the article, respondents answered questions measuring perceived crime lev- els, support for police spending, and confidence in police. All images and text were borrowed or adapted from real online news content (32–36).

Experimental Results. Fig. 5 shows average differences in responses between each treatment condition and the con- trol condition in the unweighted samples. All outcomes were rescaled to range from 0 to 100 so effects can be interpreted as percentage-point increases on each outcome’s respective scale. Militarized images in the M-Turk survey caused clear increases Fig. 3. SWAT team deployments exert no detectable effects on crime or between roughly 8 points and 15 points in perceived levels of officer safety. Shown are coefficients on log(#SWAT Deployments + 1) from crime in the vignette city. The high militarization condition in panel regressions using data from Maryland, FY2010–2014. All outcomes are the SSI survey caused a statistically significant 2.2-point increase logged. Bars are 95% confidence intervals. in the perceived level of crime in the vignette city and, strikingly, a 3.2-point drop in respondents’ desire for more police patrols in their own neighborhoods. Fig. 3 displays the effects of increased SWAT deployments in The high militarization treatment also caused support for Maryland on crime and officer safety. Fig. 3, Left shows results police funding in the to fall by roughly four points from models using all deployments, summed within agency in the M-Turk survey and two points in the SSI survey. Support months and logged, while Fig. 3, Right uses deployments only for barricade situations to test whether using SWAT in emer- gencies might be especially effective. (I thank an anonymous reviewer for this suggestion.) (Officer fatalities are omitted as AB an outcome here because they almost never occur in these data; SI Appendix, Table S7.) As Fig. 3 shows, point estimates are negative but near zero, none are statistically significant, and all estimates imply a 10% increase in deployments exerts changes smaller than 1% in all outcomes. Some small declines in officer assaults are detectable if the data are weighted by agency size (SI Appendix, Table S35), suggesting the Maryland results are more sensitive to model specification than the national analysis which makes it difficult to draw firm conclusions. Using the available data, the benefits of increased deployments appear to be either CD small or nonexistent. The adoption and deployment of SWAT teams is not ran- domly assigned, and we should therefore make causal interpreta- tions with some caution (see SI Appendix, section 2B, for placebo tests assessing the risk of endogeneity). The Maryland results may also not generalize to other states. But the lack of any robust association between militarized policing and public and officer safety shown here calls the validity of these claimed benefits into question.

Effects on Police Reputation I conducted two survey experiments to estimate effects on pub- lic perceptions of police, the first one via Amazon’s Mechanical Fig. 4. Experimental manipulation [from Top Left (A) to Bottom Right Turk (M-Turk) and the second one via SSI. Respondents read a (D)]. Reprinted with permission from Reuters Pictures/Jessica Rinaldi, Shut- mock news article concerning an unnamed police chief seeking a terstock/JPL Designs, Associated Press/Jeff Roberson, and The New York budget increase. Text remained constant while the accompanying Times/Redux Pictures/Whitney Curtis.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1805161115 Mummolo xesv ro okidctn iprt ramn fBlack of Mummolo treatment disparate indicating work prior extensive magnitude. in small be differences to any likely that are suggest effects, effects 6 disparate Fig. in in of estimates detection point small the responses. the for but heterogeneous allow of may general, samples evidence In Larger little significant. reveal statistically vignette results not the is these in the difference crime as this perceived such but on signs, city, effects. militarization opposite in high have effects of difference group effects the the as cases well African some In as and separately, white self-reports) non-Hispanic via measured race/ethnicity on respondent (with estimated respondents American survey SSI Respondent. the of Race by public. Effects mass the in enforcement to effects law help images. of could all militarized views time over negative and to items cement exposure news direction, similar brief to of single exposure Repeated terms a only in M-Turk after results the appear repli- M-Turk generally from results the sizes SSI cate the effect limitations, cau- large these be Despite should the survey. we effects endorsing so treatment replicate, about militarization not tious of did low sample the effects M-Turk smaller and the The substantially from sample, large. also SSI at are the treatment in enforcement militarization law high not the agency and the to question suggesting confined be vignette, in may experimental effects images’ the confidence militarized of in lower some portrayed did police survey the SSI high in the the in although treatment significant, statistically militarization not but negative erally (see funded in fund- the discussion well that A police impression already of the fell. was gave supportive agency also equipment less militarized article because were suggest ing news respondents responses open-ended the treated of analysis that in text and department reading close the funding for intervals. confidence (blue) 95% M-Turk are in Bars images surveys. militarized (red) SSI of and effects are Shown reputation. police 5. Fig. hsrltvl nfr atr frsos ssrrsn given surprising is response of pattern uniform relatively This gen- are States United the in police in confidence on Effects uvyeprmna eut:sen iiaie oiemytarnish may police militarized seeing results: experimental Survey IAppendix SI eto 2E section , i.6sostetetefcsin effects treatment shows 6 Fig. ). IAppendix SI i.S10 Fig. , and iiaie oiigi h ntdSae 1,1)hsrie con- raised has 16) (15, States United the of in normalization policing The militarized (3). preserve hierarchies to social serve that class-based ways and in race- dispropor- color of been citizens to historically applied tionately have strategies policing Aggressive Conclusion and Discussion work (38). participation prior civic stunted since with investigations associated troubling, criminal are inhibit is and police of This views negative ways. that shows unintended enforcement, milita- in law that on is likely costs analysis reputational present impose this can the of policing from rized scope learn the we Adju- what beyond (37). But is study. hostility explanations racial these work to between previous dicating exposure of with the by probability predicted consistent well increased not also is discrimination perceived are that showing discrim- results of residents memories These conjure Black their ination. they while because with negatively enforcement, clash react law may they of because perception images baseline react militarized may to respondents effects white negatively example, produce For to magnitude. group similar different each of that within is operating explanation are One mechanisms groups? two white Black the among among in among effects than had response lower they mean points ( respondents percentage the confidence in 21 States, much condition was police United how control participants the in asked the in trust when in police data, of Indeed, SSI levels color. the depressed of communities and in police by Americans ofiec intervals. confidence 6. Fig. rates crime local for adjusting concentrations after high even with Americans, areas African in of deployed often more are units This claims. police. techniques these analytical protect evaluate and systematically sources and tactics to data of crime the array violent an defend marshals deter study administrators enforce- can law they enforcement toward claiming law opinion public but in eroding used ment, is being and is strategy ways policing similar heavy-handed new, a that cerns ossetwt ncoa vdne(1,mltrzdpolice militarized (11), evidence anecdotal with Consistent ohtrgnosefcsb aeo epnet asae95% are Bars respondent. of race by effects heterogeneous No P < 0 . 001 .Wyte,d eosresc similar such observe we do then, Why ). NSLts Articles Latest PNAS | f6 of 5

POLITICAL SCIENCES and other community traits. But I find no firm evidence that no detectable public safety benefit. While SWAT teams arguably SWAT teams lower an agency’s rate or the rates at remain a necessary tool for violent emergency situations, restrict- which officers are killed or assaulted. Using survey experiments, ing their use to those rare events may improve perceptions of I show that citizens react negatively to the appearance of mili- police with little or no safety loss. tarized police units in news reports and become less willing to ACKNOWLEDGMENTS. I thank Justin Grimmer, Alexandra Blackman, fund police agencies and less supportive of having police patrols Courtenay Conrad, Ted Enamorado, Andrew Guess, Jens Hainmueller, in their own neighborhoods. Andrew Hall, Dorothy Kronick, Amy Lerman, Neil Malhotra, Steve Mello, Tali Given the concentration of deployments in communities of Mendelberg, Naomi Murukawa, Terry Moe, Erik Peterson, Jonathan Rodden, color, where trust in law enforcement and government at large Gary Segura, Mike Tomz, Sean Westwood, Vesla Weaver, Lauren Wright, and Yiqing Xu for their guidance throughout this project. I thank Helene Wood, is already depressed (14, 38), the routine use of militarized Madeleine Marr, and Grace Masback for research assistance. This work was police tactics by local agencies threatens to increase the his- funded by Stanford University, Princeton University, and National Science toric tensions between marginalized groups and the state with Foundation Grant 15-571.

1. Institute for Intergovernmental Research (2015) After-action assessment of the 21. Masera F (2016) Bringing war home: Violent crime, police killings and the over- police response to the August 2014 demonstrations in Ferguson, Missouri. COPS militarization of the US police. Available at https://papers.ssrn.com/sol3/papers.cfm? Office Critical Response Initiative (Office of Community Oriented Policing Services, abstract id=2851522. Accessed May 1, 2018. Washington, DC). 22. Beavers E, Shank M (2014) Get the military off of main street: Ferguson shows the risks 2. Baldwin J (1966) A report from occupied territory. The Nation. Available at https:// of militarized policing. The New York Times. Available at https://www.nytimes.com/ www.thenation.com/article/report-occupied-territory/. Accessed May 1, 2018. 2014/08/15/opinion/ferguson-shows-the-risks-of-militarized-policing.html. Accessed 3. Alexander M (2010) The New Jim Crow: Mass Incarceration in the Age of May 1, 2018. Colorblindness (The New Press, New York). 23. Szoldra P (2014) SWAT team called on peaceful protesters in Ferguson? Police arrest 4. Garland D (1996) The limits of the sovereign state: Strategies of crime control in journalists, fire and rubber . Business Insider. Available at www. contemporary society. Br J Criminol. 36:445–471. businessinsider.com/swat-ferguson-protest-2014-8. Accessed May 1, 2018. 5. Gottschalk M (2008) Hiding in plain sight: American politics and the carceral state. 24. Rosenfeld R, Fornago R (2012) The impact of police stops on precinct and Annu Rev Polit Sci 11:235–260. burglary rates in , 2003-2010. Justice Q 31:96–122. 6. Wacquant L (2009) Prisons of Poverty (Univ of Minnesota Press, Minneapolis). 25. MacDonald J, Fagan J, Geller A (2016) The effects of local police surges on crime and 7. Gama F (2016) The racial politics of protection: A critical race examination of police in New York City. PLoS One 11:1–15. militarization. Calif L Rev 104:979–1008. 26. Mummolo J (2017) Modern police tactics, police-citizen interactions and the prospects 8. Balko R (2014) Shedding light on the use of SWAT teams. The Washington Post. Avail- for reform. J Polit 80:1–15. able at https://www.washingtonpost.com/news/the-watch/wp/2014/02/17/shedding- 27. Enamorado T, Fifield B, Imai K (2018) Using a probabilistic model to assist merging light-on-the-use-of--teams/. Accessed May 1, 2018. of large-scale administrative records. Available at https://ssrn.com/abstract=3214172. 9. President’s Task Force on Policing (2015) Final Report of the Presi- Accessed August 1, 2018. dent’s Task Force on 21st Century Policing (Office of Community Oriented Policing 28. Redden M (2015) Documents reveal the fearmongering local cops use to score Services, Washington, DC). Available at https://ric-zai-inc.com/Publications/cops-p311- military gear from . Mother Jones. Available at www.motherjones.com/ pub.pdf. politics/2015/08/new-documents-reveal-fearmongering-local-cops-use-score-military- 10. Madhani A (2014) St. Louis county chief defends . USA Today. gear-pentagon. Accessed May 1, 2018. Available at https://www.usatoday.com/story/news/nation/2014/09/16/ferguson-st- 29. Sack K (2017) Door-busting drug raids leave a trail of blood. The New York louis-county-michael-brown-militarization/15736907/Newsweek. Accessed May 1, Times. Available at https://www.nytimes.com/interactive/2017/03/18/us/forced-entry- 2018. warrant-drug-raid.html. Accessed May 1, 2018. 11. ACLU (2014) War comes home: The excessive militarization of American polic- 30. Eberhardt J, Goff PA, Purdie VJ, Davies PG (2004) Seeing black: Race, crime, and visual ing. Available at https://www.aclu.org/sites/default/files/assets/jus14-warcomeshome- processing. J Personal Soc Psychol 87:876–893. report-web-rel1.pdf. Accessed May 1, 2018. 31. Angrist J, Pischke J (2009) Mostly Harmless Econometrics: An Empiricist’s Companion 12. Camp J (2016) Incarcerating the Crisis: Freedom Struggles and the Rise of the (Princeton Univ Press, Princeton). Neoliberal State (Univ of Press, Oakland, CA). 32. Bosman J, Eckholm E (2014) Anonymity in police shooting fuels anger in Mis- 13. Hinton E (2016) From the War on Poverty to the War on Crime: The Making of Mass souri. The New York Times. Available at www.nytimes.com/2014/08/14/us/missouri- Incarceration in America (Harvard Univ Press, Cambridge, MA). teenager-and-officer-scuffled-before-shooting-chief-says.html? r=0. Accessed May 1, 14. Tyler T, Fagan J (2008) Legitimacy and cooperation: Why do people help the 2018. police fight crime in their communities? Ohio State J Criminal Law 6:231– 33. Jonsson P (2014) Mike Brown shooting: Images of militarized police bring out 275. critics. The Christian Science Monitor. Available at https://www.csmonitor.com/ 15. Kraska P, Kappeler V (1997) Militarizing American police: The rise and normalization USA/Justice/2014/0814/Mike-Brown-shooting-Images-of-militarized-police-bring-out- of units. Soc Probl 44:1–18. critics. Accessed May 1, 2018. 16. Kraska P (2007) Militarization and policing: Its relevance to 21st century police. 34. Molotch H (2013) How much security is too much? New York Daily News. Available at Policing 1:501–513. www.nydailynews.com/opinion/security-article-1.1320360. Accessed May 1, 2018. 17. Balko R (2013) Rise of the Warrior Cop: The Militarization of America’s Police Forces 35. Sebens S (2013) Oregon : House of spies or terror-threat preven- (Public Affairs, New York). tion? Watchdog.org. Available at watchdog.org/86470/spy-center-or-terrorist-threat- 18. Wofford T (2014) How America’s police became an army: The 1033 program. prevention-or-fusion-center-in-limbo/. Accessed May 1, 2018. Newsweek. Available at www.newsweek.com/how-americas-police-became-army- 36. Zummo M (2014) Vernon police looking to add officers. The Advertiser-News. Avail- 1033-program-264537. Accessed May 1, 2018. able at www.advertisernewsnorth.com/article/20140122/NEWS01/140129966/Vernon- 19. Bove V, Gavrilova E (2017) Police officer on the frontline or a soldier? the effect of police-looking-to-add-officers. Accessed May 1, 2018. police militarization on crime. Am Econ J Econ Policy 9:1–18. 37. Hopkins D, et al. (2016) Out of context: The absence of geographic variation in US 20. Harris M, Park J, Bruce D, Murray M (2017) Peacekeeping force: Effects of pro- immigrants’ perceptions of discrimination. Polit Groups Identities 4:363–392. viding tactical equipment to local law enforcement. Am Econ J Econ Policy 9: 38. Lerman A, Weaver V (2014) Arresting Citizenship: The Democratic Consequences of 291–313. American Crime Control (Univ of Press, Chicago).

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1805161115 Mummolo