GRCC Psychology Lecture Series | Joseph Cesario
Resumo
TLDREl contenido aborda la complejidad de los tiroteos policiales fatales en los Estados Unidos, examinando el posible sesgo racial en la decisión de disparar por parte de los oficiales. Hasta 2015 no existían bases de datos completas sobre estos eventos, lo que creó inexactitudes en el entendimiento de su frecuencia y circunstancias. Investigaciones recientes han utilizado bases de datos compiladas por organizaciones periodísticas para evaluar tales eventos, encontrando que el sesgo racial en las decisiones de disparo es una cuestión complicada y no siempre suscribible a simples disparidades de población. Aun cuando los datos muestran disparidades a nivel de población global, cuando se ajustan por la exposición al crimen violento, estas disparidades parecen disminuir. Sin embargo, ciertos tipos de disparos, especialmente casos de malentendidos como confundir una billetera con un arma, presentan grandes incertidumbres debido a la falta de datos. La exposición al crimen y la violencia es clave para comprender estos eventos, más aún que la simple raza del oficial o la víctima. Los resultados revelan la importancia de la exposición al crimen en estas situaciones policiales y ponen en duda ideas simplificadas sobre el sesgo racial.
Conclusões
- 📊 Antes de 2015 no había datos completos sobre tiroteos policiales en EE.UU.
- 📉 Disparidades raciales en tiroteos disminuyen al ajustar por crimen violento.
- 🤔 No se puede simplificar el sesgo racial con los niveles de población.
- 🔍 Diferencias en crimen violento por raza afectan las estadísticas de disparos.
- ❓ Incertidumbre sobre sesgos raciales en casos específicos de tiroteos.
- 🔫 Los tiroteos fatales suelen ocurrir en contextos de criminalidad.
- 🧩 La raza del oficial no es un factor determinante en el disparo.
- 📈 El contexto del lugar influye más que la raza del oficial en los tiroteos.
- 📝 La colecta de datos por medios es crucial para estos análisis.
- 🧐 Los estudios sugieren que entendimientos comunes sobre el sesgo pueden ser erróneos.
Linha do tempo
- 00:00:00 - 00:05:00
En este video se discute la complejidad de las interacciones entre la policía y los ciudadanos, especialmente en el contexto de tiroteos fatales. Se debate sobre la noción de que los tiroteos fatales por parte de la policía están mal entendidos y que encontrar evidencia de sesgo racial en las decisiones de disparar no es tan simple como parece. A través de algunos casos emblemáticos de EE.UU., se plantea que las suposiciones de sesgo racial en tiroteos policiales son complejas y a menudo se basan en datos insuficientes.
- 00:05:00 - 00:10:00
Históricamente, no había bases de datos completas sobre los tiroteos fatales de la policía en EE.UU. hasta que organizaciones como The Washington Post y The Guardian comenzaron a documentarlos en 2015. Antes de esto, muchas policía no reportaban estos eventos, y se descubrió que había un subregistro del 50%. Este descubrimiento fue en parte porque grandes ciudades decidieron no reportar para no sobresalir en estadísticas negativas. Sin embargo, la situación está cambiando con nuevas legislaciones.
- 00:10:00 - 00:15:00
La interacción entre un oficial de policía y un ciudadano es compleja y dinámica. Puede empezar como una decisión discrecional o no discrecional, y cada interacción puede desarrollarse de manera diferente. Ejemplos de diferentes interacciones resaltan cómo incluso un pequeño cambio en la situación puede llevar a resultados diferentes, algunos de los cuales pueden terminar en eventos fatales. Esto subraya que no todos los tiroteos son iguales.
- 00:15:00 - 00:20:00
Al enfrentar un tiroteo policial, es importante considerar la influencia de ambas partes en la interacción: tanto el oficial como el ciudadano. Se destaca que la evidencia de sesgo racial puede no ser consistente a lo largo de diferentes etapas de interacción, y es posible que los prejuicios raciales no aparezcan en todas las decisiones involucradas en el uso de la fuerza letal.
- 00:20:00 - 00:25:00
El análisis de los tiroteos fatales muestra que el uso de la fuerza letal por parte de policías está vinculado con situaciones de crímenes violentos. Se discuten ejemplos donde los ciudadanos estaban armados o eran sospechosos de un crimen violento en progreso. Se resalta que una fuerza letal justificada se basa en la percepción de amenaza por el oficial.
- 00:25:00 - 00:30:00
Los datos sugieren que la mayoría de las situaciones que llevan a un tiroteo fatal no son aleatorias, sino situaciones de crímenes violentos. En 2015, muchas víctimas de tiroteos estaban armadas. Este análisis indica que los oficiales disparan principalmente en estos contextos, y que el número de espectadores inocentes afectados es bajo.
- 00:30:00 - 00:35:00
El uso de proporciones poblacionales para medir disparidades raciales en tiroteos puede ser engañoso, como se demuestra con ejemplos de tratamiento de cáncer y carreras académicas en Física. Para comprender las disparidades, es importante considerar el contexto específico, como quiénes están realmente en las situaciones donde se utiliza la fuerza letal.
- 00:35:00 - 00:40:00
Los ciudadanos de diferentes grupos raciales no están involucrados en situaciones de delitos violentos en proporciones iguales a su representación poblacional. Esto afecta cómo se analiza el uso de la fuerza letal, sugiriendo que tasas de crimen violentas son un mejor marcador para evaluar disparidades que meras cifras de población.
- 00:40:00 - 00:45:00
Los datos del FBI y el CDC se utilizan para entender mejor la exposición de los distintos grupos raciales a situaciones de crimen violento, que son contextos donde la fuerza letal es más relevante. Estos datos indican que no hay un sesgo anti-negro significativo en los tiroteos, una vez se considera el contexto del crimen violento.
- 00:45:00 - 00:50:00
Al examinar casos más ambiguos, como tiroteos de ciudadanos desarmados, los datos son inciertos y limitados, lo que lleva a un estado de desconocimiento. Esto subraya la complejidad de hacer aseveraciones concluyentes sobre el sesgo racial en situaciones específicas de uso de la fuerza letal.
- 00:50:00 - 00:56:24
Por último, se analiza la raza de los oficiales involucrados en los tiroteos. No hubo relación entre la raza del oficial y del ciudadano disparado. Los datos sugieren que las tasas de criminalidad local influyen más en estos eventos, y la raza del oficial no es un factor determinante. Esto apunta a que el contexto del crimen violento es clave para entender los tiroteos policiales fatales.
Mapa mental
Vídeo de perguntas e respostas
¿En qué contextos suelen ocurrir los tiroteos policiales fatales?
Las situaciones de tiroteos policiales fatales ocurren principalmente en contextos de crímenes violentos, donde los oficiales sienten amenazas para sus vidas o las de los demás.
¿Siempre hay sesgos raciales en los tiroteos policiales fatales?
No, no siempre hay pruebas claras de sesgos raciales en los tiroteos policiales fatales una vez que se consideran las tasas de criminalidad por grupo racial.
¿Por qué es engañoso sólo usar proporciones de población para analizar los tiroteos policiales raciales?
Existen desigualdades en las oportunidades que tienen los diferentes grupos raciales para estar en situaciones donde podría ser relevante el uso de fuerza mortal por parte de la policía.
¿Qué relación se encontró entre la raza de los oficiales y los tiroteos de ciudadanos de minorías?
Se encontró que el nivel de crimen en un condado está más relacionado con la raza de la persona que es fatalmente disparada que la raza de los oficiales involucrados.
¿Cuándo comenzaron a existir bases de datos completas sobre tiroteos policiales fatales?
Las bases de datos más completas sobre tiroteos policiales fatales se crearon a partir de 2015 gracias al trabajo de organizaciones periodísticas como The Guardian y The Washington Post.
¿Implica el sesgo racial que más ciudadanos negros sean disparados, independientemente del contexto?
No, ya que las investigaciones muestran que los tiroteos ocurren principalmente en situaciones de criminalidad violenta y no sólo por el simple sesgo racial.
¿Por qué hay incertidumbre en algunos resultados sobre el sesgo racial y los tiroteos?
La incertidumbre prevalece debido a la falta de datos adecuados para algunas situaciones específicas de tiroteo, como aquellas que involucran malentendidos de objetos como armas.
¿La raza del oficial afecta a quién disparan?
La raza del oficial no se relaciona directamente con el hecho de que dispare más a personas de minorías raciales, sino más con el contexto de criminalidad del lugar.
¿Por qué antes de 2015 no había datos completos sobre tiroteos policiales?
Antes de 2015 no existían bases de datos completas debido a que no era un requerimiento federal reportar tiroteos a nivel nacional.
¿Hay evidencia clara de sesgo racial en los datos de tiroteos policiales analizados?
El trabajo analítico presentó que incluso en casos de tiroteos no está claro si hay sesgos raciales debido a que muchos factores contextuales influyen.
Ver mais resumos de vídeos
- 00:00:06it's a pleasure
- 00:00:10interests you in that you have
- 00:00:13on the research what I want to do is
- 00:00:16maybe give an overview of some of the
- 00:00:18work that we've done in our lab on the
- 00:00:20topic of race and police shootings
- 00:00:22including some analyses of real-world
- 00:00:25data from police shooting events in the
- 00:00:28United States and also some experimental
- 00:00:30laboratory data if we have time as well
- 00:00:33and where we're gonna end up getting to
- 00:00:36hopefully at the end of this is a
- 00:00:38general theme that the the nature of
- 00:00:40fatal police shootings is often
- 00:00:42misunderstood
- 00:00:44in many people's minds and that finding
- 00:00:47evidence for racial bias in officers
- 00:00:50decisions to shoot is actually a much
- 00:00:52more complicated thing than it might
- 00:00:54appear on the surface okay and at the
- 00:00:56same time also where we're gonna end up
- 00:00:58is that there are some things that we
- 00:00:59just don't know we don't have good
- 00:01:01enough data for and we are and some
- 00:01:04questions in a state of uncertainty and
- 00:01:06and we have to just be comfortable with
- 00:01:08being in a state of unknown and
- 00:01:10uncertainty for a while so when we think
- 00:01:13about police shootings at least in the
- 00:01:16United States most of us think of a
- 00:01:19number of widely publicized and really
- 00:01:22well-known highly tragic cases that have
- 00:01:25happened in the u.s. okay if you're of a
- 00:01:29certain age probably roughly my age or
- 00:01:32older one of the first people that might
- 00:01:34come to mind is amadou diallo who was
- 00:01:36shot by the NYPD unarmed African
- 00:01:40immigrant who was reaching for his
- 00:01:41wallet and was shot in the late 90s and
- 00:01:45more recently more in the modern or in
- 00:01:48the last 10 to 15 year is Michael Brown
- 00:01:52of course Ferguson Missouri also unarmed
- 00:01:55when shot and a few years ago Philander
- 00:01:59Oh Castillo okay who was shot in
- 00:02:00Minnesota again during a traffic stop
- 00:02:03was reaching for his wallet and the
- 00:02:05officer shot and killed him okay
- 00:02:09much of the discussion around fatal
- 00:02:11police shootings concerns cases like
- 00:02:14this okay really high-profile
- 00:02:16extraordinarily tragic heartbreaking
- 00:02:19cases of fatal shootings in the u.s.
- 00:02:22well it goes along with that is often
- 00:02:24the assumption that these
- 00:02:25are the common kind of fatal police
- 00:02:28shooting okay
- 00:02:29that in other words that these are
- 00:02:30representative of what fatal police
- 00:02:32shootings really look like so we can ask
- 00:02:36question of what do we know about the
- 00:02:38roughly thousand or so US citizens who
- 00:02:40are shot and killed by the police each
- 00:02:43year and what do we know about the
- 00:02:46circumstances surrounding those
- 00:02:47shootings and what do we know about the
- 00:02:49role of racial bias or of an
- 00:02:52individual's race on the likelihood that
- 00:02:55that person was shot him it turns out
- 00:02:58that up until a few years ago we
- 00:03:00actually weren't able to answer these
- 00:03:02kinds of questions and any kind of
- 00:03:04comprehensive nationwide level okay
- 00:03:07that's because prior to 2015 in case you
- 00:03:09didn't know there were no complete
- 00:03:12databases of fatal police shootings
- 00:03:14that's because in an almost unbelievable
- 00:03:19condition here in the u.s. police
- 00:03:22departments are not required by law to
- 00:03:24report fatal police shootings to the
- 00:03:26federal government okay
- 00:03:28so when a police department has a fatal
- 00:03:30shooting of a citizen they're under
- 00:03:32they're encouraged to report but they're
- 00:03:34under no technical obligation to report
- 00:03:36that to the to the federal government
- 00:03:38and so people had suspected that federal
- 00:03:41databases of fatal police shootings were
- 00:03:43under counts of actual shoots shootings
- 00:03:46but no one really had any hard data on
- 00:03:48just how bad those federal databases
- 00:03:51were in 2015 news organizations like The
- 00:03:55Washington Post and The Guardian
- 00:03:56pictured here did some reporting and
- 00:03:59developed public databases of all fatal
- 00:04:02police shootings in the US and so they
- 00:04:04did investigative journalism they had
- 00:04:07citizens contact them when there was a
- 00:04:09shooting in their city and they went out
- 00:04:11and gathered information about that and
- 00:04:12compiled all of those in one place and
- 00:04:15there are a little bit different from
- 00:04:16one another to databases aren't exactly
- 00:04:18the same but they are as far as we know
- 00:04:20the most comprehensive and complete
- 00:04:22databases of fatal police shootings and
- 00:04:25it turned out you could we could now
- 00:04:27look and ask how much were police
- 00:04:30departments under reporting in the US
- 00:04:32and they it turned out they were under
- 00:04:33reporting by about 50% okay so in at the
- 00:04:36federal level we only knew
- 00:04:38about half of all of the actual
- 00:04:40shootings that had occurred in the US
- 00:04:42right um there's some rumor that the
- 00:04:47reason why or one of the major reasons
- 00:04:49why that was the case was at the three
- 00:04:51major cities in the u.s. particularly
- 00:04:53Los Angeles New York and Chicago all had
- 00:04:56an unspoken agreement that they wouldn't
- 00:04:57fully report because none they didn't
- 00:04:59know which was worse none of them wanted
- 00:05:01to be number one in this statistic so
- 00:05:03they all just simply failed to report
- 00:05:05and that's that's a large number of
- 00:05:07shootings from the database but it's not
- 00:05:09all of them so it was the case that many
- 00:05:12departments perhaps most apartments were
- 00:05:14under reporting the the police shootings
- 00:05:16that they were having okay so that that
- 00:05:20is an incredible statistic it's still
- 00:05:22not the case that they're forced to
- 00:05:23they're working on new legislation to
- 00:05:25get this but this is really one of the
- 00:05:27the most embarrassing aspects of United
- 00:05:30States federal law enforcement policy so
- 00:05:34now that we have these more complete
- 00:05:36databases we can look to them and try to
- 00:05:38answer some questions about the nature
- 00:05:40of fatal police shootings okay by way of
- 00:05:44getting to some answers it's a day of
- 00:05:48technical problems that's fine I can use
- 00:05:52the computer here by way of getting to
- 00:05:54some answers we want to first I'll keep
- 00:05:57something in mind okay when we try to
- 00:06:00ask the question about fatal police
- 00:06:01shootings so that we get some clarity
- 00:06:03about what exactly we are and what
- 00:06:05exactly we aren't able to answer them so
- 00:06:09one important point is that the process
- 00:06:13that leads from initial contact between
- 00:06:16a police officer and a citizen to the
- 00:06:19endpoint of a fatal shooting is a very
- 00:06:22complicated dynamic oftentimes
- 00:06:26vulnerable volatile very uncertain
- 00:06:30process okay and this makes the topic a
- 00:06:32bit difficult to study it is not a
- 00:06:34straightforward movement from initial
- 00:06:36contact to shooting and it's not a
- 00:06:39straightforward series of events even
- 00:06:41from one shooting to the next and so we
- 00:06:43have to appreciate some of the
- 00:06:44uncertainty that's part of this okay if
- 00:06:46we think about you know some kind of
- 00:06:48hypothetical police-citizen interaction
- 00:06:51and there's some initial contact that
- 00:06:53gets made between the officers and the
- 00:06:55citizens sometimes that's discretionary
- 00:06:58where the police officer might make the
- 00:07:00decision to stop someone because they
- 00:07:02look like a suspect or they think
- 00:07:04they're doing something suspicious or
- 00:07:05whatever other times it's
- 00:07:06non-discretionary okay so the officers
- 00:07:09might be responding to 911 911 call they
- 00:07:11don't have a choice and to do that okay
- 00:07:13they have they're compelled to do that
- 00:07:15so there can be discretionary there can
- 00:07:17be non-discretionary initial contacts
- 00:07:19okay once that initial contact is made
- 00:07:22though there's some interaction that
- 00:07:25begins okay some ongoing interaction
- 00:07:27that's gonna unfold over time right it
- 00:07:30might be initial information gathering
- 00:07:32on the part of the officer and during
- 00:07:34that interaction many different things
- 00:07:36can occur that you could imagine that at
- 00:07:38some point the officer is going to make
- 00:07:40the decision to pack down the citizen
- 00:07:42that they're talking with okay for again
- 00:07:44for whatever reason we don't always have
- 00:07:46insight into into that decision okay but
- 00:07:49there's some complicated process and
- 00:07:50complicated interaction going with a
- 00:07:53number of different decision steps along
- 00:07:55the way okay sometimes it ends there and
- 00:07:58that's the end of the the police citizen
- 00:08:00interaction other times as it unfolds it
- 00:08:04can branch off you could think about the
- 00:08:05many different ways that that
- 00:08:06interaction could branch off and follow
- 00:08:08different paths okay just give you a
- 00:08:11couple of examples from recent some
- 00:08:13police shootings and some not police
- 00:08:15shootings okay recently in New York an
- 00:08:18individual was stopped he was a part of
- 00:08:21the group of citizens who was a highly
- 00:08:23publicized event of dumping water on
- 00:08:26some police officers he was stopped for
- 00:08:28a different reason
- 00:08:30the police officers were talking with
- 00:08:32him it was not I wouldn't describe it as
- 00:08:34a friendly interaction okay but it
- 00:08:36wasn't an overtly combative interaction
- 00:08:38either when they went to cuff him then
- 00:08:40he took off running okay now this ended
- 00:08:43up he was you know grabbed and and
- 00:08:46handcuffed and that was the end of it
- 00:08:47and in a different case this is a still
- 00:08:51from body cam image of a event of a
- 00:08:53shooting from Salt Lake City a couple
- 00:08:55years ago an officer was called on a 911
- 00:08:59call got dispatched that there was a gun
- 00:09:01in a convenience store he showed up he
- 00:09:03tried to talk with this
- 00:09:04I believe it was a 19 year old again not
- 00:09:07overly combative but the officer told me
- 00:09:10put his hands up because he thought he
- 00:09:12was the one with the handgun the citizen
- 00:09:14told him no and reached for his
- 00:09:16waistband and the officer shot him
- 00:09:18thinking that he was going for a gun
- 00:09:19that was tucked into his waistband
- 00:09:21it turned out he was unarmed ok so again
- 00:09:23an initial contact that went in a very
- 00:09:25different direction one was Kimura
- 00:09:27combative one was a less combative event
- 00:09:30in Chicago last year this was an event
- 00:09:34where again officers stopped a citizen
- 00:09:36that he was a suspect that they were for
- 00:09:40I believe a murder actually the
- 00:09:42interaction again was going very well
- 00:09:43until they went to pat him down and then
- 00:09:45they discovered he had a gun on his hip
- 00:09:47he then became combative and ran when he
- 00:09:51ran he got stopped by a car spun around
- 00:09:53grabbed his hip and the officer shot him
- 00:09:56ok in that case he didn't grab his gun
- 00:09:58but he would had his hand where his gun
- 00:10:01was and the officers knew that he was
- 00:10:03armed and and sometimes you get much
- 00:10:06clearer cases such as this case where
- 00:10:09again a citizen was stopped he had a gun
- 00:10:11ran from the officers had the gun out
- 00:10:14and turned around to face the officers
- 00:10:15and he also was shot so if you think
- 00:10:19about that series of events ok that long
- 00:10:24complex just highly variable set of
- 00:10:28events leading from initial contact to a
- 00:10:30possible shooting we really want to
- 00:10:32appreciate just the complexity that's
- 00:10:34involved there ok no two shootings are
- 00:10:37going to be exactly alike right within
- 00:10:41this what we also want to keep in mind
- 00:10:43and this is often hard to remember is
- 00:10:46that there are really two parties in any
- 00:10:49one of these interactions ok and each of
- 00:10:51those parties can influence the way in
- 00:10:54which that interaction unfolds at any of
- 00:10:56these points right and so we have the
- 00:10:58officer and we care a lot about the
- 00:11:00officer and we typically study the
- 00:11:01officer and focus on the officer but
- 00:11:03there's also the citizen right both of
- 00:11:05them together are coming together in
- 00:11:07this interaction and influencing how it
- 00:11:10unfolds not all officers are the same
- 00:11:12some are more authoritarian than others
- 00:11:14some are more combative and
- 00:11:16disrespectful than others
- 00:11:17not all citizens are the same okay some
- 00:11:19are more combative and disrespectful
- 00:11:21than others they are more or less
- 00:11:23compliant than others each of those at
- 00:11:25any point can impact how this
- 00:11:27interaction carries out so within all of
- 00:11:33this it's very firm about and also the
- 00:11:36other thing to keep in mind is that if
- 00:11:38we do find evidence of racial bias let's
- 00:11:41say on the part of police officers at
- 00:11:43one stage it doesn't mean there's racial
- 00:11:45bias at every one of these stages okay
- 00:11:47and if we don't find it at one stage it
- 00:11:48doesn't mean that there isn't some
- 00:11:50somewhere else okay so an officer might
- 00:11:52be racially biased let's say in terms of
- 00:11:54the quality of the interaction that they
- 00:11:57have with the citizen where maybe
- 00:11:58they're dis more disrespectful say
- 00:12:00toward minority than non-minority
- 00:12:01citizens okay they might show racial
- 00:12:04bias there and not in some other stuff
- 00:12:06okay so we're gonna try as we move
- 00:12:08through some of the data to answer a
- 00:12:10number of these steps okay but where
- 00:12:12we're gonna start with is that decision
- 00:12:15to use deadly force and the endpoint
- 00:12:18decision here in the moment where the
- 00:12:20officer has to make a decision to fire
- 00:12:23his or her gun or not okay what do we
- 00:12:26know about that decision in the role
- 00:12:27that race might play in it and when we
- 00:12:30ask about racial bias we what we really
- 00:12:33mean is the question that was a famous
- 00:12:35question that was raised a number of
- 00:12:36decades ago of whether officers have
- 00:12:39quote one trigger finger for blacks and
- 00:12:41a separate trigger finger
- 00:12:42for whites okay in other words is it the
- 00:12:45case that officers would not shoot
- 00:12:47people if they were white relative to if
- 00:12:50they were black that's essentially the
- 00:12:52question of racial bias in that decision
- 00:12:55to shoot okay
- 00:13:00so as I said the major question that we
- 00:13:03have then is to try to understand race
- 00:13:04bias in that decision okay in the
- 00:13:07decision to use deadly force and
- 00:13:08understand the degree to which it
- 00:13:11impacts police officers okay you might
- 00:13:15think that's really not a question that
- 00:13:18needs to be asked because we all already
- 00:13:20know in fact that there's a racial
- 00:13:22disparity and who gets shot and we
- 00:13:24already know that police officers are
- 00:13:26biased and that's what you know account
- 00:13:29for that racial disparity in terms of
- 00:13:31citizen race and being fatally shot by
- 00:13:34the police well what we want to do
- 00:13:37actually is look at what the evidence
- 00:13:38for that is and try to understand that
- 00:13:40evidence in a little bit more detail and
- 00:13:42see how solid that really might be and
- 00:13:45what we're gonna conclude here at the
- 00:13:48end of this section is that
- 00:13:50understanding racial by old asperities
- 00:13:53okay understanding whether one group is
- 00:13:55shot more than we would expect relative
- 00:13:57to another group is always relative okay
- 00:14:00we always have to have some standard or
- 00:14:02some comparison that we're dealing with
- 00:14:05okay we're comparing the rate at which
- 00:14:06some group is shot relative to what okay
- 00:14:09and that becomes a really key question
- 00:14:11in understanding racial disparities and
- 00:14:15so just let's take a look at some
- 00:14:17numbers okay these are fatal shooting
- 00:14:20data in the u.s. from 2015 we're gonna
- 00:14:23restrict our dumb numbers here just to
- 00:14:26black or white citizens just for ease of
- 00:14:29analyzing the data but I you know we can
- 00:14:32look at other groups as well if you're
- 00:14:34interested in we can talk about that at
- 00:14:36the end of him and these are only fatal
- 00:14:39shootings so as hard as it is to get
- 00:14:42data and learn something about fatal
- 00:14:43shootings it's even harder to get to any
- 00:14:46data to learn something about non-fatal
- 00:14:47shootings okay so only only conclusions
- 00:14:51that we can draw a really concern the
- 00:14:53case where a citizen is actually killed
- 00:14:55we know almost nothing about when
- 00:14:57citizens aren't killed but they are
- 00:14:58shocked okay and and officers do not
- 00:15:01kill everybody that they they shoot okay
- 00:15:04there's a lot of shootings that occur
- 00:15:05and that don't result in a fatality
- 00:15:07we're just looking at fatal shootings
- 00:15:09here because those are the data that we
- 00:15:10that we have that we can look at and so
- 00:15:13in 2015 there were 261 black citizens in
- 00:15:18the u.s. who were shot fatally by the
- 00:15:20police they imagine each of the people
- 00:15:22up there is a hundred people and and
- 00:15:25there were 526 white citizens who were
- 00:15:29shot by the police in 2015 and so in
- 00:15:32terms of just a simple odds right based
- 00:15:35on those raw numbers we would say that
- 00:15:37whites were twice as likely to be shot
- 00:15:39relative to blacks in the United States
- 00:15:41in 2050
- 00:15:43okay now obviously we all recognize
- 00:15:45right off the bat that that's not a
- 00:15:47particularly meaningful number okay to
- 00:15:49know that whites were twice as likely
- 00:15:51there's 526 versus 261 because of course
- 00:15:55there are more white citizens in the US
- 00:15:56than black citizens okay so even if
- 00:15:59officers were showing a lot of racial
- 00:16:01bias we would still expect more whites
- 00:16:03to be shot just given that there's a
- 00:16:05larger pool of whites to draw from okay
- 00:16:07relative to black citizens in the u.s.
- 00:16:10so we need to do some kind of adjustment
- 00:16:13for that right in other words we need to
- 00:16:15have some benchmark right what to what
- 00:16:18do we benchmark those 261 and those 526
- 00:16:22citizens shot we have to have some
- 00:16:23comparison and the way that this is
- 00:16:28typically done the standard way to do
- 00:16:30for this to do this is to adjust for or
- 00:16:34compared to overall population levels
- 00:16:37okay that is a standard kind of
- 00:16:39comparison to look at disparities of
- 00:16:41really any kind but especially racial
- 00:16:44disparities in the decision to shoot
- 00:16:46okay and when we do that what we're
- 00:16:48asking is our black citizens our white
- 00:16:51citizens more likely to be shot given
- 00:16:54each groups overall population level
- 00:16:57okay so given the overall number of
- 00:16:59black and white citizens in the u.s.
- 00:17:01right benchmarked against that are black
- 00:17:04or white citizens more likely to be to
- 00:17:07be fatally shot okay that's really the
- 00:17:10question that we're asking when we
- 00:17:11benchmark against population values okay
- 00:17:14so again if we take those 261 in the 526
- 00:17:18citizens okay we can benchmark that
- 00:17:22again against the I'm not putting the
- 00:17:25exact numbers roughly 40 million black
- 00:17:28citizens in the US and 200 million white
- 00:17:30citizens in the US okay and we can ask
- 00:17:33what are the odds that a person is shot
- 00:17:35if they're black or white given the
- 00:17:39overall population values okay and the
- 00:17:42way that we would do that which is just
- 00:17:43a standard way again that every
- 00:17:45calculation would do that is you would
- 00:17:47compute the odds of being shot if you're
- 00:17:49a black citizen relative to the odds
- 00:17:51that you're a shot of being a white
- 00:17:52citizen and divided one of the odds over
- 00:17:55the other okay and then
- 00:17:56gives you an odds ratio that tells you
- 00:17:58that one group is acts times as likely
- 00:18:00as another to have something happen
- 00:18:02again this is absolutely ubiquitous if
- 00:18:05you were doing work on the odds of dying
- 00:18:07from a heart attack or something you
- 00:18:08would do exactly the same sort of
- 00:18:10calculation okay and so basically you
- 00:18:13could say you would take the 261 and
- 00:18:16compute the odds that a black citizen is
- 00:18:18getting is going to be shot given the
- 00:18:20population of 40 million that works out
- 00:18:23two point zero zero zero zero zero six
- 00:18:26five okay so you're looking at roughly
- 00:18:28seven in a million okay you do the same
- 00:18:32thing for whites 526 out of 200 and
- 00:18:35million again you you basically divide
- 00:18:37those two numbers you get about three
- 00:18:39out of a million okay
- 00:18:40and what you can see is that relative to
- 00:18:42each groups overall population it's
- 00:18:45clear that black citizens are more
- 00:18:46likely to be shocked okay
- 00:18:48the odds are two-and-a-half times
- 00:18:49greater than a black citizen will be
- 00:18:51shot relative to white citizen seven
- 00:18:53million out versus seven out of a
- 00:18:55million versus three 1/2 out of a
- 00:18:57million okay and that's the standard
- 00:18:58odds ratio that's a well-known value if
- 00:19:01you look at things like unarmed
- 00:19:02shootings that number is even higher
- 00:19:04something closer to three and a half
- 00:19:06times more likely to be shot unarmed if
- 00:19:08you're a black citizen relative to if
- 00:19:11you're a white citizen okay so when we
- 00:19:12say that one group is two-and-a-half
- 00:19:14times more likely to have some event
- 00:19:16that's how we would calculate that okay
- 00:19:19if you prefer you could do the same
- 00:19:22thing in terms of percentages okay you
- 00:19:23could say out of everybody who's shot by
- 00:19:26the police
- 00:19:26black satisfy the police excuse me black
- 00:19:30citizens represent twenty three percent
- 00:19:31of that group okay of that roughly
- 00:19:34thousand people or so that's about
- 00:19:36twenty three percent of that group in
- 00:19:37the US population
- 00:19:39black citizens are roughly thirteen
- 00:19:40percent there's a disparity right
- 00:19:42thirteen is we would expect thirteen
- 00:19:44percent and we get twenty three so again
- 00:19:46there's a disparity there we do the same
- 00:19:48thing with whites and again computer
- 00:19:49odds ratio okay to give us roughly the
- 00:19:52same value there okay so when we do that
- 00:19:58the conclusion is is absolutely clear
- 00:20:01there's no question at all about this
- 00:20:03that relative to the population values
- 00:20:06of each group or the population levels
- 00:20:08of each group blacks are over
- 00:20:10presented in being fatally shot by the
- 00:20:12police okay and so we would state that
- 00:20:15as black citizens are two-and-a-half
- 00:20:17times likely to be shot compared to
- 00:20:19whites given each groups overall number
- 00:20:22in the US population okay and again as I
- 00:20:29said this is a just absolutely standard
- 00:20:31way of doing this kind of work
- 00:20:32anytime you're showing a per capita
- 00:20:34event that you're basically doing the
- 00:20:37same thing as what I just showed you any
- 00:20:39time you know two per 100,000 or two per
- 00:20:41million whatever it might be
- 00:20:43comparing percentages all of these cases
- 00:20:45are basically doing the same thing of
- 00:20:47calculating a ratio based on or
- 00:20:51benchmarked against some population
- 00:20:54value okay but notice what we're saying
- 00:20:59when we make the statement that black
- 00:21:01citizens or any group would be
- 00:21:03two-and-a-half times more likely to be
- 00:21:04shot okay given their overall
- 00:21:07representation in the population okay I
- 00:21:10said earlier that all calculations of
- 00:21:14this type are relative to some standard
- 00:21:16there's always some comparison that we
- 00:21:19have to make in that we have to justify
- 00:21:20the particular comparison that we're
- 00:21:22using okay here the benchmark as we said
- 00:21:25is population proportions it's the forty
- 00:21:27million and the two hundred million
- 00:21:29citizens in the u.s. broadly okay when
- 00:21:33we use that what we're saying is that we
- 00:21:35expect that group members should be
- 00:21:37shocked at the same proportion that they
- 00:21:40exist in the general population okay
- 00:21:42that's the implicit statement that we're
- 00:21:44making when we do this kind of
- 00:21:46comparison and if we deviate from that
- 00:21:48expectation then we say that there's a
- 00:21:50disparity right we expect that black
- 00:21:52citizens should be shot it should be 13%
- 00:21:55of those shot because they're 13% of the
- 00:21:57population we deviate from that and so
- 00:21:59there's a disparity there and but what
- 00:22:03we want to ask actually in digging a
- 00:22:04little deeper is whether it's correct
- 00:22:06in fact and whether it's accurate to
- 00:22:08compare the proportion of a groups
- 00:22:10outcome to that groups overall
- 00:22:13representation in the population okay is
- 00:22:16that actually the right benchmark for us
- 00:22:18to understand this outcome in to gain an
- 00:22:20understanding of racial bias in the
- 00:22:23decision to
- 00:22:24shoot okay okay well why might this be
- 00:22:27misleading let's take a couple non
- 00:22:29shooting examples just to illustrate the
- 00:22:33general problem with this calculation
- 00:22:35okay imagine that there's a new cancer
- 00:22:38treatment out okay it's the best cancer
- 00:22:40treatment that there is it's the you
- 00:22:41know gold star for cancer treatment and
- 00:22:43we want to research whether groups are
- 00:22:45receiving that treatment in a
- 00:22:47proportionate way okay is there a
- 00:22:49disparity in who is receiving that
- 00:22:51treatment okay so let's say we look at
- 00:22:53everybody who has who gets cancer
- 00:22:55treatment in a year and we find that 13%
- 00:22:58of all those people who got the best
- 00:23:00cancer treatment were black citizens
- 00:23:02okay we're black patients so we look and
- 00:23:07we compare that to the thirteen percent
- 00:23:09in the u.s. population right by that
- 00:23:12metric or by that benchmark we would say
- 00:23:15that there is not a disparity right
- 00:23:17things look great here okay black
- 00:23:19citizens are 13% of the population and
- 00:23:21they're also 13% of those receiving the
- 00:23:24top treatment everything looks fine okay
- 00:23:26so by that benchmark we would say that
- 00:23:28the the world of cancer treatment does
- 00:23:31not show any disparities any racial
- 00:23:33disparities I'm imagine that we found
- 00:23:37out for whatever reason might be the
- 00:23:39case we don't need to know the reason
- 00:23:41right now but imagine that we found out
- 00:23:42that actually although black citizens
- 00:23:45make up 13% of the US population
- 00:23:48they actually make up 50% of cancer
- 00:23:50patients now at this point we would I
- 00:23:54think rightly conclude that there was a
- 00:23:56racial disparity here okay that black
- 00:23:58patients were not getting the top
- 00:24:00treatment in accordance with what is the
- 00:24:03appropriate benchmark the people who
- 00:24:05have cancer okay
- 00:24:06so in this case it's quite clear that we
- 00:24:09would want to compare the proportion who
- 00:24:11got cancer treatment to the proportion
- 00:24:12who actually have cancer not to the
- 00:24:14general US population the general US
- 00:24:16population is irrelevant in this case
- 00:24:18okay because the outcome isn't needed
- 00:24:21for everybody right in this case there's
- 00:24:24a very clear way in which that's true
- 00:24:25which is that people generally don't
- 00:24:27elect to have cancer treatment if they
- 00:24:28don't have cancer okay so so we have a
- 00:24:31clearer sense of who the appropriate
- 00:24:33pool is and it's not the total US
- 00:24:35population
- 00:24:37let's take another example which is
- 00:24:39disparities sucks disparities in stem
- 00:24:42okay we can look at physics majors in
- 00:24:45undergraduate populations okay let's
- 00:24:47look and see as is the case that females
- 00:24:51represent about 20% of physics majors
- 00:24:53okay so we look at all physics majors if
- 00:24:56we consider that to be the outcome
- 00:24:57you know enrolling as a physics major we
- 00:24:59find that women are only 20% of physics
- 00:25:02majors and we look to the u.s.
- 00:25:04population women are about 51 percent of
- 00:25:07the u.s. population alright that's a
- 00:25:09disparity we have a clear disparity here
- 00:25:11there's a 30% gap in terms of what we
- 00:25:14would expect to be true right we should
- 00:25:17expect a 51% enrollment of women in
- 00:25:20physics
- 00:25:20whereas physics majors but in this case
- 00:25:24actually the disparity is much worse
- 00:25:27than that okay and that's because women
- 00:25:30make up 70% of the undergraduate
- 00:25:32population ok so again the disparity is
- 00:25:37not a 30% disparity the disparity is
- 00:25:39actually a 50% disparity then as we saw
- 00:25:42in the previous example the relevant
- 00:25:44pool of individuals is not the u.s.
- 00:25:46population it's people enrolled in
- 00:25:48college okay because you have to be
- 00:25:50enrolled in college to be a physics
- 00:25:51major right so the u.s. population
- 00:25:54benchmark as in the prior example can be
- 00:25:57misleading right if the outcome is not
- 00:26:01equally available to everybody or
- 00:26:03equally relevant to everybody then it's
- 00:26:05not always the case that the u.s.
- 00:26:07population level is the right benchmark
- 00:26:09okay here it's the undergraduate
- 00:26:11population okay both of these cases
- 00:26:15illustrate some potential pitfalls with
- 00:26:18choosing the US population as the right
- 00:26:21benchmark okay and they have a major
- 00:26:24assumption that's built in there right
- 00:26:26and that assumption is that members of
- 00:26:29different groups have equal
- 00:26:30opportunities for experiencing that
- 00:26:32outcome okay in the case of cancer
- 00:26:35treatment it's obviously it's obvious
- 00:26:37that that's not true okay because only
- 00:26:39people with cancer elect to have cancer
- 00:26:41treatment okay in this physics major the
- 00:26:44same is true because not everybody in
- 00:26:46the US population is becoming a physics
- 00:26:48major you have to be enrolled as an
- 00:26:49undergraduate okay
- 00:26:51so that the assumption there when we use
- 00:26:54that population benchmark is that
- 00:26:56members of different groups have that
- 00:26:58same opportunity and if we apply this
- 00:27:02logic to the case of fatal police
- 00:27:05shootings when we benchmark shooting
- 00:27:08percentages or raw shooting numbers
- 00:27:11against the US population what we are
- 00:27:15saying in fact what's required for us to
- 00:27:17do that is to say that members of
- 00:27:20different racial groups are in those
- 00:27:22situations where deadly force is
- 00:27:24relevant at rates that are proportionate
- 00:27:27to their overall population just as in
- 00:27:31the cancer and the stem example what
- 00:27:34that calculation requires is that if
- 00:27:37black citizens are 13% of the US
- 00:27:39population and if white citizens are 70
- 00:27:41percent of the population then it must
- 00:27:43be the case that if we look at all of
- 00:27:46those decisions all those situations
- 00:27:47where deadly force is relevant for the
- 00:27:50officer okay where the officer is in the
- 00:27:53case of making a shooter no shoot
- 00:27:54decision that 13 percent of those
- 00:27:57individuals are black citizens and 70
- 00:27:59percent are white that has to be the
- 00:28:03case if we're going to look to the
- 00:28:05population proportions I'm so
- 00:28:09benchmarking any type of you benchmark
- 00:28:11some number against the US population
- 00:28:13you are requiring that the experience of
- 00:28:17that or the opportunity for that event
- 00:28:19to occur happens in proportion to that
- 00:28:22groups overall representation okay if it
- 00:28:25doesn't as with the stem in the cancer
- 00:28:27examples you're gonna get a misleading
- 00:28:29understanding of what the world is like
- 00:28:31and if it's the case that different
- 00:28:36groups occupy deadly force relevant
- 00:28:40decisions at different rates and then
- 00:28:43the more appropriate benchmark is not
- 00:28:46the number of citizens in the general US
- 00:28:49population okay it's the number of
- 00:28:51citizens from each group who are in
- 00:28:53those deadly force situations okay
- 00:28:56that's the appropriate benchmark if we
- 00:28:59want to know something about racial bias
- 00:29:01on the part of the police officer and
- 00:29:03I'll give some examples here
- 00:29:04in a second of why that's actually
- 00:29:07important and if said differently
- 00:29:09you can't experience a policing outcome
- 00:29:12without encountering the police and so
- 00:29:15if members from different groups
- 00:29:17encounter the police at different rates
- 00:29:19we have to take that into account right
- 00:29:21we can't just look at their overall
- 00:29:23population values otherwise it's a
- 00:29:25misleading quantity okay given this what
- 00:29:33we can do is ask whether police are
- 00:29:35likely to use deadly force in every
- 00:29:36single situation okay if it's the case
- 00:29:39that police are likely to use deadly
- 00:29:40force as likely in one situation as
- 00:29:43another then actually we're okay with
- 00:29:45using the population values right if
- 00:29:48it's the case that police are more
- 00:29:49likely to use deadly force in some
- 00:29:51situations than others then that
- 00:29:53population value might be misleading
- 00:29:55okay and so we want to know are there
- 00:29:58some situations for which deadly force
- 00:30:00is relevant for an officer and some for
- 00:30:03which it is not okay and it's very clear
- 00:30:05and I'll show you the data on this okay
- 00:30:07that deadly force is not relevant for
- 00:30:10every single experience that a police
- 00:30:12officer has or every single kind of
- 00:30:14policing scenario for an officer okay
- 00:30:16and that's because deadly force
- 00:30:18scenarios are situations are strongly
- 00:30:21strongly tied to violent crime
- 00:30:23situations okay
- 00:30:25so as a qualitative way of looking at
- 00:30:27this we can actually look I just started
- 00:30:30at 2015 at the database that we have now
- 00:30:32from the Guardian and went through the
- 00:30:34first ten shootings of 2015 okay so
- 00:30:37let's take a look I gave you give you a
- 00:30:39quick summary of what those first ten
- 00:30:41shootings are the first shooting there
- 00:30:45was a 911 call about a domestic
- 00:30:47disturbance so that's non-discretionary
- 00:30:48the police have to show up the police
- 00:30:50show up to this house the man here
- 00:30:53emerged from his house pointing his gun
- 00:30:55at police officers and police officer
- 00:30:58shot and killed the police officer who
- 00:30:59showed up shot and killed him and second
- 00:31:02we don't have a picture of this
- 00:31:03individual again a 9-1-1 call came in
- 00:31:06there was an armed man police showed up
- 00:31:09the man threatened to shoot the officers
- 00:31:11pointed his gun at them refused to drop
- 00:31:13his gun and the officers shot and killed
- 00:31:15him
- 00:31:16okay third case police respond to 9-1-1
- 00:31:20call okay there was a violent dispute a
- 00:31:22violent domestic dispute they showed up
- 00:31:24they the man was tased okay he wasn't
- 00:31:27complying with the police orders so he
- 00:31:29was tased he reached for his waistband
- 00:31:31and was shot in this case he didn't have
- 00:31:33a gun on him the officer thought he was
- 00:31:36reaching for a gun when he went for his
- 00:31:37waistband this one is a still actually
- 00:31:43from the dashcam video of this event man
- 00:31:46was pulled over in a traffic stop at
- 00:31:48night the officer thought he was acting
- 00:31:50really nervously he got him out of the
- 00:31:52car did a pat down as he was patting him
- 00:31:54down discovered a gun the moment the
- 00:31:56officer discovered the gun the guys
- 00:31:57became combative and tried to grab his
- 00:31:59gun and the officer shot him while he
- 00:32:02was reaching for his gun
- 00:32:05this one again there's no no picture
- 00:32:07this individual a man pulled up to a
- 00:32:08restricted police lot waited for the
- 00:32:10police to come to his car when he did he
- 00:32:12pulled out a gun and only aimed it at
- 00:32:14the officers and the officer shot him
- 00:32:15this was a tragic case also of suicide
- 00:32:19by cop we know this because he left a
- 00:32:21note in his car for the police officer
- 00:32:23who was going to eventually killed him
- 00:32:25saying I'm sorry I used you I just want
- 00:32:28to die and so on okay we'll talk we
- 00:32:30could talk about that also there's a
- 00:32:31surprisingly high number of those cases
- 00:32:33in the US I think the number six man
- 00:32:37identified by the police as a robbery
- 00:32:39suspect he was driving a stolen vehicle
- 00:32:41they stopped him they ID'd him as a
- 00:32:43robbery suspect he then opened his door
- 00:32:45and started firing his gun at the police
- 00:32:46officers officers killed him and police
- 00:32:49respond to a 911 call of a man armed
- 00:32:51with a gun threatening bar patrons and
- 00:32:53assaulting his girlfriend he was shot
- 00:32:56when he refused to drop his weapon again
- 00:33:009-1-1 call man threatening his son the
- 00:33:02police show up the man took out his gun
- 00:33:04and pointed it at officers was shot and
- 00:33:06killed him police were trying to carry
- 00:33:09out a warrant for this man he was wanted
- 00:33:10for sexual assault of a teenage girl he
- 00:33:13aimed his gun at the police officers and
- 00:33:15was shot okay finally again 9-1-1 call
- 00:33:18officers respond the man here had a
- 00:33:21knife he stabbed the police dog when
- 00:33:23they when they showed up with the dog
- 00:33:24and then he raised his knife over his
- 00:33:26head and approached the officers in the
- 00:33:28officers shot
- 00:33:29and the point about going through each
- 00:33:33of these cases is to illustrate that
- 00:33:35this is not a random sample of
- 00:33:37situations that everybody in the u.s.
- 00:33:39finds themselves in okay and it's not a
- 00:33:42random sample of what is relevant for
- 00:33:44everybody in the US okay these
- 00:33:46situations are by and large violent
- 00:33:50crime okay they involve violent crime
- 00:33:52right and police training as well as the
- 00:33:54legal justification for police to use
- 00:33:56deadly force is tied to threats okay
- 00:33:59they're tied to threats to the officers
- 00:34:01life or another citizens life that's
- 00:34:04when police are justified in using
- 00:34:05deadly threat or a deadly force and so
- 00:34:10the question is not our police or our
- 00:34:13black or white citizens more likely to
- 00:34:15be shot given their overall population
- 00:34:18proportions it's our black and white
- 00:34:19citizens more likely to be shot given
- 00:34:21their encounters with the police in
- 00:34:23these decision relevant situations okay
- 00:34:26and the quantitative data backed this up
- 00:34:28as well so if you look at the data on
- 00:34:30again that now we can analyze that we
- 00:34:33have these large you know nearly
- 00:34:35complete databases and what you see is
- 00:34:38that deadly force use again is strongly
- 00:34:40tied to physical threats from citizens
- 00:34:44so somewhere between 85 and 90 percent
- 00:34:46of citizens who are fatally shot are
- 00:34:48actually armed at the time of being shot
- 00:34:52out of the roughly eight hundred black
- 00:34:56or white citizens so again about a
- 00:34:57thousand are shot every citizens are
- 00:34:59shot every year there were eleven
- 00:35:00hundred and twenty fifteen of those
- 00:35:02eight hundred are black or white
- 00:35:04citizens if you look at those black or
- 00:35:07white citizens in the US who are fatally
- 00:35:09shot only about 45 or so per year could
- 00:35:13be described or classified as being both
- 00:35:16unarmed and not aggressing against an
- 00:35:19officer at the time of being shot okay
- 00:35:21so in the phalangeal Casteel case for
- 00:35:24example that would be one of those cases
- 00:35:27in some of the smaller data sets that
- 00:35:30were analyzed prior to the the 2015
- 00:35:34year's 2015 2016 the those research were
- 00:35:37finding the same sorts of things roughly
- 00:35:4090% of fatal shootings involved what
- 00:35:43defined as a clear deadly threat again
- 00:35:45someone has a gun they're currently
- 00:35:47shooting at someone so on and so forth
- 00:35:50and so it is a case that when you
- 00:35:53analyze fatal police shootings you do
- 00:35:54see that officers are overwhelmingly
- 00:35:56shocked in crime related situations okay
- 00:36:00and almost always shoot the person who
- 00:36:02is involved in criminal activity
- 00:36:05mistakes do happen and we'll talk about
- 00:36:07those but innocent bystanders being shot
- 00:36:09by the police is really quite rare okay
- 00:36:15so if police are more likely to use
- 00:36:18deadly force in violent crime related
- 00:36:22situations okay where violent crime is
- 00:36:25currently happening or where they've
- 00:36:27showed up to a scene where it's
- 00:36:28suspected to have been happening okay if
- 00:36:31that's the case then the question that
- 00:36:34we want to know okay if we're going to
- 00:36:38use the population proportion as a
- 00:36:39benchmark is to know whether citizens
- 00:36:42from different groups are in those
- 00:36:44situations at the same rates okay so is
- 00:36:47it the case that black and white
- 00:36:48citizens and when again we could look at
- 00:36:50the full range of data we have data on
- 00:36:52Hispanic and Asian citizens as well okay
- 00:36:54but what we want to know again for this
- 00:36:56for the purposes of what we're analyzing
- 00:36:57is it the case that black and white
- 00:37:00citizens are involved in violent crime
- 00:37:02situations to the same degree okay and
- 00:37:06this is a very complicated let's just
- 00:37:09appreciate the complicated nature of why
- 00:37:12this is true okay nobody's saying that
- 00:37:14there's a simple answer for why this is
- 00:37:15true or a simple solution for it okay
- 00:37:18but it is very well-known that people
- 00:37:20from different racial groups don't
- 00:37:22occupy violent crime situations at equal
- 00:37:25rates okay and so the data are just
- 00:37:28enormous ly clear on this there's no
- 00:37:30question at all about it if you look to
- 00:37:32the data and we'll talk about police
- 00:37:34bias in arrests and reporting's also in
- 00:37:37a little bit okay
- 00:37:38but as just one example you could look
- 00:37:40at the Centers for Disease Control data
- 00:37:43Centers for Disease Control have nothing
- 00:37:45to do with police data the police don't
- 00:37:47report the data to them these are
- 00:37:48autopsy reports okay and what we can do
- 00:37:51is look because we know that over 90% of
- 00:37:54any death
- 00:37:56if it's a if it's a murder and not
- 00:37:58negligent manslaughter is within race we
- 00:38:01can get a pretty good understanding of
- 00:38:02who suspects are based on who victims
- 00:38:05are okay and in 2015 for example if you
- 00:38:07just look at all assaults and so these
- 00:38:10are raw numbers in terms of all assaults
- 00:38:12white citizens dying again almost always
- 00:38:15at the hands of another white citizen
- 00:38:17there were 5000 cases for black citizens
- 00:38:20again dying at the hands of black
- 00:38:21citizens was almost 9,000 cases okay
- 00:38:25in 2020 15 if you look at death by
- 00:38:28firearm discharge specifically so all
- 00:38:31assaults is any kind of assault if you
- 00:38:33look at just firearm deaths those
- 00:38:34numbers are even more dramatically
- 00:38:36different
- 00:38:37okay 3,000 versus 7,500 and again those
- 00:38:40are raw numbers so when you actually
- 00:38:42look at a per capita rate the difference
- 00:38:45in in violent crime experience is
- 00:38:47massive for black and white citizens
- 00:38:49okay there's a very strong difference
- 00:38:51again this is victimization in terms of
- 00:38:53them being victimized but because we
- 00:38:55know that almost all of the are about 90
- 00:38:57percent or so of the suspects in those
- 00:39:00cases are the same race we can get a
- 00:39:02good sense of what the suspects are also
- 00:39:05okay to just be clear this is this a
- 00:39:09really important point
- 00:39:10before we move on also that the number
- 00:39:13of people who do things like commit
- 00:39:15murder is unbelievably low okay and so
- 00:39:19it's useful when we talk about crime and
- 00:39:21race to remind ourselves that almost
- 00:39:24everybody from every group is exactly
- 00:39:26the same okay it doesn't matter what
- 00:39:28your racial group is you're not going to
- 00:39:30commit murder right and so if you look
- 00:39:31at this this is arrest rates for black
- 00:39:33and white citizens all right point zero
- 00:39:36zero two and point zero one percent of
- 00:39:40black and white citizens are arrested
- 00:39:43for murder and nine others remain sorry
- 00:39:45you can't see it but this is actually a
- 00:39:46pie chart with that sliver cut out of it
- 00:39:49okay that's how small of a group that is
- 00:39:52right the number of people who would
- 00:39:54commit murder is unbelievably low black
- 00:39:58and white citizens are the same 99 you
- 00:40:01know above 99 percent of them okay we're
- 00:40:04all the same right but there is a
- 00:40:06difference there is a probabilistic
- 00:40:08difference
- 00:40:09and in that difference is bigger when
- 00:40:11you know there's a larger slice of the
- 00:40:13pie when you look at violent crime in
- 00:40:14general rather than just murder but
- 00:40:16there is an important difference there
- 00:40:21okay well would skip this slide and move
- 00:40:25on to this okay so as we said then
- 00:40:27because black and white citizens are not
- 00:40:30involved in violent crime at the same
- 00:40:33rates okay in terms of being in line
- 00:40:35with their population levels this is
- 00:40:37gonna make the population benchmark
- 00:40:39misleading okay it's like looking at all
- 00:40:42people instead of cancer patients right
- 00:40:44we want the group of people for whom the
- 00:40:46outcome is actually relevant okay and so
- 00:40:49the more informative question is
- 00:40:50actually our black or white citizens
- 00:40:53over-represented and in fatal police
- 00:40:56shootings given each groups involvement
- 00:40:58in violent crime because that's when
- 00:41:00fatal police shootings happen okay or
- 00:41:02said differently given the rates at
- 00:41:05which black and white citizens are
- 00:41:07exposed to the police okay
- 00:41:09in those situations because remember you
- 00:41:11can't experience a policing outcome
- 00:41:13without being exposed to the police and
- 00:41:16then is it the case that officers are
- 00:41:18more likely to shoot black citizens okay
- 00:41:21now the problem with this of course is
- 00:41:24we don't know any groups racial or any
- 00:41:27groups crime rate in in reality the true
- 00:41:30crime rate is unknown of any group okay
- 00:41:33what we can do is try to look at as many
- 00:41:35different pieces of data as possible and
- 00:41:37then see whether they're all telling us
- 00:41:39the same answer and and so what I'm
- 00:41:43gonna show you is some first some data
- 00:41:46there are more data here and I'll
- 00:41:47explain them as we get it but what we
- 00:41:49can do to look at exposure first is we
- 00:41:51can look at the FBI's two different FBI
- 00:41:54sources and again we'll look at non
- 00:41:55policing sources in a moment okay but we
- 00:41:58can look at the FBI's summary report
- 00:42:00system data okay on murder non-negligent
- 00:42:03manslaughter rates alright our numbers
- 00:42:06for both arrests and reports of those
- 00:42:08events violent crime arrests violent
- 00:42:11crime includes murder non-negligent
- 00:42:13manslaughter rape robbery and assault
- 00:42:16okay
- 00:42:16and also weapons violations ok who's
- 00:42:19carrying a gun illegally
- 00:42:22and again this is being arrested for
- 00:42:24that right we can also look at the
- 00:42:26National Institute and that's our
- 00:42:28National Incident based reporting system
- 00:42:30also from the FBI and look at again
- 00:42:32murder non-negligent manslaughter
- 00:42:33violent crime and weapons violations so
- 00:42:35these are overlapping but somewhat
- 00:42:37distinct datasets of crime okay and what
- 00:42:40we can do then is ask what are the odds
- 00:42:42that a black or white citizen is shot
- 00:42:44given their rates of violent crime of
- 00:42:48that group okay
- 00:42:49so again instead of saying given the
- 00:42:50population proportion what do we know
- 00:42:53given the rate at which different
- 00:42:55members of different groups would
- 00:42:56encounter the police in those situations
- 00:42:59where the police might use deadly force
- 00:43:01okay or for which deadly force would be
- 00:43:03relevant okay and so these are all just
- 00:43:05listed here the first panel is
- 00:43:07benchmarking off of homicide data okay
- 00:43:11what you're gonna see is on the y-axis
- 00:43:12is the odds ratio sorry the clickers and
- 00:43:15other pointers not working and I'm not
- 00:43:17tall the odds ratio one would mean that
- 00:43:20there's an equal likelihood of a black
- 00:43:22or a white citizen being shot okay
- 00:43:24numbers below mean black citizens would
- 00:43:26be more likely numbers above would mean
- 00:43:28white citizens would be more likely okay
- 00:43:30so again where we saw before on the
- 00:43:32population level a two point five odds
- 00:43:34of black citizens being 2.5 times more
- 00:43:37likely to be shot that would be down
- 00:43:39here the bar would be low to a below one
- 00:43:42okay well when we benchmark and set it
- 00:43:45on homicide data actually that effect
- 00:43:48goes away okay and in indeed it flips to
- 00:43:51the other side where black citizens are
- 00:43:53not more likely to be shot than white
- 00:43:55citizens given each groups involvement
- 00:43:57in homicide okay we can do the same
- 00:44:00thing benchmarking on violent crime in
- 00:44:02general and there's a number of
- 00:44:03different calculations of violent crime
- 00:44:05that you can do they're all up there so
- 00:44:07you can see them what you could see is
- 00:44:08in no case do you see again evidence of
- 00:44:11anti-black disparity in police shootings
- 00:44:13once you account for violent crime rates
- 00:44:16then and again the same is true when we
- 00:44:18look at weapons violation data so again
- 00:44:21this is asking given the rates at which
- 00:44:24black and white citizens are arrested
- 00:44:27for or reported carrying illegal weapons
- 00:44:30okay we again don't see anti black
- 00:44:33disparity in
- 00:44:34fatal police shootings now an obvious
- 00:44:38problem here is if the data themselves
- 00:44:41are biased okay then that might mask any
- 00:44:44actual bias on the part of the police so
- 00:44:47if it's just the case that police are
- 00:44:48more likely to let's say pat-down and
- 00:44:51arrest black citizens relative to white
- 00:44:53citizens then the whole thing is wrong
- 00:44:55because there it's gonna be bias already
- 00:44:57built into the data okay and and that
- 00:44:59would mask anti black disparities on the
- 00:45:03part of the police making deadly force
- 00:45:05decisions what we can do to account for
- 00:45:07that is to look at two additional
- 00:45:09sources of data okay one as I said is
- 00:45:12the CDC death data okay so these are
- 00:45:15data from the Centers for Disease
- 00:45:16Control on whether people have died in
- 00:45:19to death by assault those are not
- 00:45:22subject to police bias right so we know
- 00:45:24that these are as much as we can a quote
- 00:45:27pure count of homicide rates for
- 00:45:30different from black and white citizens
- 00:45:32we can also look at the National the
- 00:45:36National victimization survey sorry
- 00:45:38blanked on what the National
- 00:45:41victimization survey okay
- 00:45:43is a nationwide representative survey
- 00:45:46it's a telephone survey people are
- 00:45:48called up every year and asked were you
- 00:45:50victimized by somebody in the last year
- 00:45:52if so tell us about it
- 00:45:53okay so again that's a self-report if
- 00:45:56you were robbed you would tell the
- 00:45:58person on the phone yeah I was robbed by
- 00:45:59a you know male in the middle of the
- 00:46:01night and so on so again those are these
- 00:46:04are not contaminated by police data or
- 00:46:07police bias and what we see when we do
- 00:46:10that again is little to no evidence for
- 00:46:12anti black disparities and being shocked
- 00:46:15by the police the CDC death data look
- 00:46:18like all the other homicide data the
- 00:46:20National Crime Victimization survey data
- 00:46:22are right there on no bias which is
- 00:46:25again the odds ratio of one and same
- 00:46:27with the reports of having a weapon used
- 00:46:29against you when we benchmark on that
- 00:46:32again it looks at one
- 00:46:34now these are as you can see at the top
- 00:46:37there it says all fatal shootings these
- 00:46:40are all cases where anyone was fatally
- 00:46:42shot by the police right what we want to
- 00:46:45do though is ask is it the case that
- 00:46:47maybe a police aren't biased in all
- 00:46:49shootings okay maybe they're biased when
- 00:46:52it comes to things like the
- 00:46:54misidentification of a wallet for a gun
- 00:46:57okay or in cases where it's more
- 00:46:59ambiguous right perhaps when someone has
- 00:47:02a gun out and they're firing it at the
- 00:47:04officer they're gonna get shot no matter
- 00:47:06what the race is but in those ambiguous
- 00:47:08cases maybe that's where we see bias and
- 00:47:11and so we can do that first by looking
- 00:47:13at cases where someone is killed when
- 00:47:18they're unarmed and they're not
- 00:47:20physically attacking an officer those
- 00:47:22would be more ambiguous cases there's a
- 00:47:24hundred of those across the two-year
- 00:47:26period and now what you see okay is that
- 00:47:28there's a little bit more evidence for
- 00:47:30anti-black bias
- 00:47:31okay anti-black disparities some of
- 00:47:33those numbers are starting to go below
- 00:47:34one and what you also see is those bars
- 00:47:37the sorry the lines above and below the
- 00:47:40the odds ratios get very big okay and
- 00:47:43we'll talk about that in in just a
- 00:47:45second when we look at those cases where
- 00:47:48someone is killed while they are holding
- 00:47:50or reaching for an object they're
- 00:47:52reaching for their waistband or they
- 00:47:53take out their wallet and it's mistaken
- 00:47:55for gun again those are 45 over the
- 00:47:57two-year period that we analyzed here
- 00:48:00again you start to see a little bit of
- 00:48:02potential anti-black disparity but now
- 00:48:04those measures of uncertainty are huge
- 00:48:06okay and what that's going to do is
- 00:48:09that's going to reflect the fact that
- 00:48:10once we start to ask about these more
- 00:48:12specific questions there just isn't a
- 00:48:14lot of data there okay we really don't
- 00:48:17know the answer of whether police are
- 00:48:19more racially biased when it comes to
- 00:48:22identifying a gun in a mistaken they
- 00:48:24mistake a wallet for a gun because there
- 00:48:26are just so few cases of that so it then
- 00:48:30at this level we have to just be in a
- 00:48:32place of uncertainty where we say we
- 00:48:35don't really know whether there's racial
- 00:48:36bias on the part of the police it might
- 00:48:38be the case it looks like there might be
- 00:48:40some evidence for that but there's just
- 00:48:42so much uncertainty in what we can know
- 00:48:44and that's not true for all the
- 00:48:46shootings when we look at all the
- 00:48:47shootings
- 00:48:48a lot of data there we know what the
- 00:48:50answer is it doesn't look like there's
- 00:48:51racial disparities a racial bias on the
- 00:48:54part of police okay once we get to this
- 00:48:56level we're just at a place of
- 00:48:58uncertainty so if we summarize that we
- 00:49:04asked our black or white citizens
- 00:49:06over-represented in fatal police
- 00:49:07shootings not given their population
- 00:49:09levels but given their and vile
- 00:49:11involvement in violent crime overall and
- 00:49:14by overall I mean when we look at all
- 00:49:16the shootings from 2015 and 2016 okay
- 00:49:20which is about 1,500 shootings total it
- 00:49:22does not look like there's anti black
- 00:49:24disparity in those shootings okay what
- 00:49:28this means actually is that even if it
- 00:49:31was the case that officers were
- 00:49:33completely blind to citizen race okay
- 00:49:36even if they somehow couldn't know what
- 00:49:39a citizen's race was we are still gonna
- 00:49:42see a population level disparity I am
- 00:49:45they're still going to be a population
- 00:49:47level disparity with black citizens
- 00:49:49being more likely to be shot given their
- 00:49:51population values because of the greater
- 00:49:54exposure into crime once we get into
- 00:49:59specific types of shootings shootings of
- 00:50:02unarmed citizens the data are just too
- 00:50:04uncertain okay we just don't have an
- 00:50:06answer to that question right now and I
- 00:50:08know generally people don't like
- 00:50:10uncertainty and don't like to be left
- 00:50:11with uncertainty but that really is the
- 00:50:13state of where those data are when we
- 00:50:15look at how and uncertain the estimates
- 00:50:18are of those odds
- 00:50:21because the data are consistent when we
- 00:50:24look at the CDC data for example it's
- 00:50:27unlikely that initial contact bias has a
- 00:50:31lot to do with those numbers okay so we
- 00:50:35are doing another analysis right now
- 00:50:36where we're separating all the shootings
- 00:50:38by those that start with a discretionary
- 00:50:40decision by the officer pull somebody
- 00:50:43over for a traffic violation versus
- 00:50:45those that don't
- 00:50:46with like 911 response calls and we'll
- 00:50:49see what the data look like but most of
- 00:50:51the police shootings are these kinds of
- 00:50:53non-discretionary mandatory responses on
- 00:50:56the part of officers and the final thing
- 00:51:02we can cut out here we is now about an
- 00:51:05appropriate time to stop or what how it
- 00:51:07is the time like so two three okay I
- 00:51:10think we'll do one more thing then what
- 00:51:14this also tells us is that the shootings
- 00:51:16that we have of what we call unarmed non
- 00:51:19aggressive citizens the phalangeal
- 00:51:21Casteel case the amadou diallo case and
- 00:51:23so on are exceptionally tragic but
- 00:51:26they're also very rare okay they are not
- 00:51:29the modal or the most common type of
- 00:51:32police shooting that we that we have
- 00:51:34then and then I'll skip all this just to
- 00:51:38get into one last finding that we have
- 00:51:42which we've looked at recently and
- 00:51:44that's data on the question of officer
- 00:51:47race okay so one thing that people have
- 00:51:50wondered some people have speculated
- 00:51:52that when it comes to shooting minority
- 00:51:55citizens particularly black citizens
- 00:51:57that white officers might be more likely
- 00:51:59to shoot minority citizens than black
- 00:52:02officers okay lots of cases of this in
- 00:52:05the media for example the race of the
- 00:52:08officer is often mentioned if the
- 00:52:09officer is white okay on the other hand
- 00:52:12people have argued that it isn't about
- 00:52:14the race of the officer it's about sort
- 00:52:16of the institution as a whole and that
- 00:52:18officers you know we don't distinguish
- 00:52:19between black or white officers when it
- 00:52:21comes to these questions okay well up
- 00:52:24until last year we actually were not
- 00:52:28able to answer the question of whether
- 00:52:29white officers are more likely than
- 00:52:31black officers for example to shoot
- 00:52:33minority citizens because again there's
- 00:52:36no there are no data bases on this right
- 00:52:39so even when there is a fatal shooting
- 00:52:40the the local police departments are not
- 00:52:43required to report the race of the
- 00:52:45officer to the federal government right
- 00:52:48so we just simply had no data actually
- 00:52:50on the race of officers involved in
- 00:52:52these kinds of shootings yeah so what we
- 00:52:55did was we spent two years trying to
- 00:52:58collect all of these data right we
- 00:53:01looked at every shooting from 2019 so
- 00:53:04again that's about a thousand or so
- 00:53:05about 1,100 total shootings and we
- 00:53:08contacted every Police Department that
- 00:53:10was that had a fatal shooting in that
- 00:53:12year and we got from those police
- 00:53:14departments after
- 00:53:15many many tries some police departments
- 00:53:18were happy to give us the information
- 00:53:19others it took a lot of effort to get
- 00:53:21from them okay but we actually got the
- 00:53:24officers race
- 00:53:25the officers sex and the officers years
- 00:53:28of experience for every officer that was
- 00:53:30involved in a fatal police shooting from
- 00:53:33that department okay and we have we got
- 00:53:35about eighty per se to five percent of
- 00:53:37all officer information which for these
- 00:53:39kinds of things is actually very high to
- 00:53:41be able to get 85% of data from police
- 00:53:44departments that doesn't always happen
- 00:53:45okay and what we were able to do is ask
- 00:53:48okay we have this database of people who
- 00:53:51were fatally shot right we know the race
- 00:53:53of the people who were fatally shot and
- 00:53:55then we could ask what among those
- 00:53:58individuals who were fatally shot okay
- 00:54:00what predicted or related to that
- 00:54:03person's race so citizen one who was
- 00:54:05shot let's say they were white okay was
- 00:54:08it more likely that there was a white or
- 00:54:10black or Hispanic officer on the scene
- 00:54:12who shot them citizen two was black
- 00:54:14wasn't more likely that they were would
- 00:54:16that there was a black or white or
- 00:54:17Hispanic officer on the scene who shot
- 00:54:19them in so on okay we can skip that
- 00:54:23question what we found actually when we
- 00:54:26looked at that was that the race of the
- 00:54:28police officer actually did not relate
- 00:54:30to the race of the citizen shot okay so
- 00:54:34it was not the case that white officers
- 00:54:36were more likely to shoot black citizens
- 00:54:38to fatally shoot black citizens and
- 00:54:40black officers were in fact at just a
- 00:54:43pure relationship level it was actually
- 00:54:46the case that black officers were more
- 00:54:47likely to have shot black citizens but
- 00:54:49that was simply due to the demographics
- 00:54:51of a County okay so in a County with a
- 00:54:53lot of black citizens you're gonna have
- 00:54:55a county with a lot of black officers
- 00:54:56okay because officers are drawn from the
- 00:54:58communities that they serve okay once
- 00:55:01you control for that there's no
- 00:55:02relationship between the race of the
- 00:55:04police officer and the race of the
- 00:55:06citizen who is shot in contrast what we
- 00:55:10did find and again this is consistent
- 00:55:12with what we saw in the in the other
- 00:55:14work that I just presented is that local
- 00:55:17or county level crime rates predicted
- 00:55:20the race of a person shot in that County
- 00:55:23okay and in predictive it quite strongly
- 00:55:25so the greater number of crimes for
- 00:55:28instance who were commit that we're
- 00:55:29made by whites in a County made it more
- 00:55:32likely that a person who is fatally shot
- 00:55:34by the police was white okay
- 00:55:36the greater number of crimes committed
- 00:55:38by black citizens in a County made it
- 00:55:40more likely that a black citizen was
- 00:55:42shot and again in this data set we
- 00:55:44expanded it to Hispanic citizens as well
- 00:55:46the greater number of crimes committed
- 00:55:48by Hispanics in a county the more likely
- 00:55:50it was that a person was Hispanic was
- 00:55:51shot or the person who shot was was
- 00:55:53Hispanic okay
- 00:55:55so again this gives further support to
- 00:55:57the idea that violent crime is really
- 00:55:59the context of fatal shootings and that
- 00:56:02that's the only thing that we have to
- 00:56:03consider when we're looking at policing
- 00:56:05behavior and then I actually will stop
- 00:56:09it there and I'm happy to answer any
- 00:56:11more any any questions that people have
- 00:56:13about the work or anything that we
- 00:56:15covered so thank you very much
- 00:56:19[Applause]
- tiroteos policiales
- sesgo racial
- Estados Unidos
- crimen violento
- bases de datos
- disparidades raciales
- investigación
- exposición policial
- contexto criminal
- evidencia
- percepción pública
- legalidad