For many years, left identity-politics activists have claimed that blacks, or sometimes “people of color”¹ in general, in the United States (and other white-majority states, to a lesser extent) are disproportionately the victims of interracial violence at the hands of whites. Media reports make lurid claims about supposed lynchings and other acts of racial violence. These claims are important because they are used to justify extreme acts of violence and looting perpetrated by black mobs. If these claims turned out to be untrue, then violence and looting by terrorist groups such as BLM and Antifa would lose their main defence: that they are supposedly a reproportioning of an existing imbalance of violence in the pursuit of justice.
The Data
The data I use for this analysis is from Criminal Victimization, 2018 [NCJ 253043], a bulletin published in September 2019 by Rachel Morgan and Barbara Oudenkerk of Bureau of Justice Statistics, an agency belonging to the US Department of Justice. Table 12 on page 12 of the report contains population estimates for racial/ethnic groups in the US, and Table 14 on page 13 contains a breakdown of violent incidents by race/ethnicity of both victim and offender.
In the US, “white”, “black”, and “Asian” are referred to as “races”, but “Hispanic” is referred to as an “ethnicity”. The most important practical effect of this for our purposes here is that while individuals cannot be both white and black (they would then be classed as “mixed”, grouped with “other” by the report), they can be both white and Hispanic, or black and Hispanic. The report treats ethnicity as the dominant characteristic over race, thus classing Hispanic whites and Hispanic blacks as “Hispanic”, and leaving “white” and “black” as remainder categories for non-Hispanic whites and non-Hispanic blacks respectively.²
The report includes victim figures for White, Hispanic, Black, and Asian. Offender figures are provided for the previous categories and also for Other and “Multiple offenders of various races”. Due to small population/sample sizes for Asians, I have analysed the three main racial/ethnic groups in the US: White, Hispanic, and Black. The basic data is as follows:
| Victim Race/Ethnicity: | White | Hispanic | Black | NET [WHB] |
|---|---|---|---|---|
| Population | 171,493,180 | 46,997,610 | 33,132,390 | 251,623,180 |
| Violent Incidents by Victim [WHB-only] |
3,137,272 | 652,891 | 500,779 | 4,290,942 |
| Observed Offender White (%) |
2,224,025 | 207,104 | 59,778 | 2,490,907 |
| Observed Offender Hispanic (%) |
365,299 | 333,422 | 44,551 | 743,272 |
| Observed Offender Black (%) |
547,948 | 112,365 | 396,450 | 1,056,763 |
Immediately we can see some suggestive details. Offending by blacks is higher than offending by Hispanics, despite Hispanics having a higher population. Though victimhood-offending should be symmetrical around the downward-sloping diagonal, there is a clear skew towards the bottom left quadrant. These facets will be explored further in the analysis proper.
The Analysis
We can create a model of the proportions in which we would expect to see intra-group and inter-group violence occur if violence were randomly distributed across the population. Assume a steady rate of violence equal to one violent incident per other member of society. The number of violent incidents that groups perform against themselves will be the triangular number of their population minus one (themselves), then doubled to account for reciprocity. The number of violent incidents that groups perform against each other will be the product of their populations. We can therefore produce expected rates of inter-group violence in the US based on the race/ethnicity population figures. Once we have those rates we can multiply them by our total number of observed incidents to produce expected numbers of violent incidents given the observed frequency of violence.
| Victim Race/Ethnicity: | White | Hispanic | Black | NET [WHB] |
|---|---|---|---|---|
| Expected Offender White (per trillion) |
29,410 | 8,060 | 5,682 | 43,152 |
| Expected Offender Hispanic (per trillion) |
8,060 | 2,209 | 1,557 | 11,826 |
| Expected Offender Black (per trillion) |
5,682 | 1,557 | 1,098 | 8,337 |
| Expected Offender White (n) |
1,993,173 | 546,228 | 385,080 | 2,924,481 |
| Expected Offender Hispanic (n) |
546,228 | 149,693 | 105,531 | 801,452 |
| Expected Offender Black (n) |
385,080 | 105,531 | 74,397 | 565,008 |
Having created these expected frequencies, we can now compare them against the observed frequencies:
| Victim Race/Ethnicity: | White | Hispanic | Black | NET [WHB] |
|---|---|---|---|---|
| Difference in Incidents White Offender (O-E) |
230,852 | -339,124 | -325,302 | -433,574 |
| Difference in Incidents Hispanic Offender (O-E) |
-180,929 | 183,729 | -60,980 | -58,180 |
| Difference in Incidents Black Offender (O-E) |
162,868 | 6,834 | 322,053 | 491,755 |
| Difference in Incidents NET Offender (O-E) |
212,791 | -148,561 | -64,229 | 0 |
| Ratio of Incidents White Offender (O/E) |
1.116 | 0.379 | 0.155 | 0.852 |
| Ratio of Incidents Hispanic Offender (O/E) |
0.669 | 2.227 | 0.422 | 0.927 |
| Ratio of Incidents Black Offender (O/E) |
1.423 | 1.065 | 5.329 | 1.870 |
| Ratio of Incidents NET Offender (O/E) |
1.073 | 0.815 | 0.886 | 1 |
These results show clearly that all racial groups are more likely to commit acts of violence against their own group. This is especially extreme for blacks, who are over 5 times as likely to be violent to each other as would be expected at this level of violent incidents. This is true even among the least violent group, whites, who only perpetrate 85% of expected violence. Hispanics are also less violent than their population would suggest, 93% of expected. The only racial group that is more violent than expected is blacks, perpetrating half a million more acts of violence than would be expected of them. Looking at victimhood, blacks and Hispanics are both less likely to be victims of violence than their population levels suggest: only whites are disproportionately the victims of violence, experiencing over two hundred thousand more violent incidents than expected.
Looking at interracial violence, the same pattern continues. White violence against blacks is just 16% of what population levels suggest, while black violence against whites is the highest excess of interracial violence, 142% of expected. Again, only black offender violence is more common than it should be, with whites and Hispanics both having lower than expected violence against their out-groups, and both being disproportionately the victims of excessive black violence.
These interracial violence figures are themselves small compared to the high levels of in-group violence, especially among blacks. We can account for this by looking at the ratio between the two sides of the expected symmetry. That is, Hispanic-on-white violence / white-on-Hispanic violence, black-on-white violence / white-on-black violence, and black-on-Hispanic violence / Hispanic-on-black violence.³ The results are as follows:
- Black-on-White violence: 9.2 times greater than white-on-black violence
- Black-on-Hispanic violence: 2.5 times greater than Hispanic-on-black violence
- Hispanic-on-White violence: 1.8 times greater than white-on-Hispanic violence
For good measure, I also ran the numbers with the small Asian sample included. The symmetry ratio comparisons suggest Asian-on-white violence at a multiplier of 1.8 times, but then Hispanic-on-Asian violence at a multiplier of 2.9 times, and black-on-Asian violence at a multiplier of around 5,000 times, though this is hard to ascertain owing to the very low sample size.
Conclusions
It is clear that the principal victims of interracial violence in the modern United States are whites. Additionally, the other side of this is not “people of color” in general, but specifically blacks. Hispanics and Asians are also net victims of black violence, and only slightly more violent than whites. It is simply untrue to suggest that blacks are in any way the victims of interracial violence in the United States, nor is there any reason to think that this pattern of excessive black violence is reversed in other white-majority countries.
As a result, it is clearly not true that the actions of the BLM / Antifa terrorist groups can be justified as a rebalancing of rates of interracial violence, because they are in fact an exacerbation of the real imbalance. Not that it would have been a valid justification anyway: the high rates of black violence likewise do not justify riots or other forms of terrorist violence by other ethnic groups. It would not be acceptable for Asians, for example, to loot black stores to gain reparations for the excessive violence they receive at the hands of blacks. Nonetheless, we can conclude that there is no problem of white-on-black violence in America, no need for whites to apologise or make amends for interracial violence of which they are the main victims, and no need to enact policies attempting to solve this non-problem.
sum θoətiz abaʊt intə-reiʃəl vaiələns
foə menii yiəriz, left aidentitii-politiks aktivistiz kleimiv ðat blakiz, oə sum-taimiz “puəsəniz ov kulə”¹ in jenrəl, in ðə yʊʊnaitəð steitiz (and uðə wait-majoritii steitiz, tʊʊw a lesər ekstent) bii disprəpʊəʃənətlii ðə viktimiz ov intə-reiʃəl vaiələns at ðə handiz ov waitiz. miidyə ripʊətiz meik lyʊərid kleimiz abaʊt supoʊzəð linciŋgiz and uðər aktiz ov reiʃəl vaiələns. ðiiz kleimiz biiy impʊətənt bikuz dii bii yʊʊzəð tə justifaiy ekstriim aktiz ov vaiələns and lʊʊtiŋ puəpətreitəð bai blak mobiz. if ðiiz kleimiz tuənid aʊt tə biiy untrʊʊ, ðen vaiələns and lʊʊtiŋ bai terərist grʊʊpiz suc az b.l.m. and antifə wʊd lʊʊz diis mein difens: ðat dii bii supoʊzəðliiy a rii-prəpʊəʃəniŋg ov an egzistiŋg imbaləns ov vaiələns in ðə pəsʊʊt ov justis.
ðə deitə
ðə deitə mii yʊʊz foə ðis analisis bii from kriminəl viktimaizeiʃən, 2018 [NCJ 253043], a bulitin publishəð in septembə 2019 bai reicəl mʊəgən and baəbrər ʊʊdənkuək ov byʊəroʊw ov justis statistiks, an eijənsii biloŋgiŋ tə ðə y.s. dipaətmənt ov justis. teibəl dein-tʊʊw on peij dein-tʊʊw ov ðə ripʊət kəntein popyəleiʃən estimətiz foə reiʃəl/eθnik grʊʊpiz in ðə y.s., and teibəl dein-fʊh on peij dein-tii kəntein a breik-daʊn ov vaiələnt insidəntiz bai reis/eθnisitiiy ov boʊθ viktim and əfendə.
in ðə y.s., “wait”, “blak”, and “eiʒən” bii rifuərəð tʊʊw az “reisiz”, but “hispanik” bii rifuərəð tʊʊw az an “eθnisitii”. ðə moʊst impʊətənt praktikəl efekt ov ðis foə wiis puəpəsiz hiə bii ðat wail indivijʊʊwəliz kan not bii boʊθ wait and blak (dii wʊd ðen bii klasəð az “miksəð”, grʊʊpəð wið “uðə” bai ðə ripʊət), dii kan bii boʊθ wait and hispanik, oə blak and hispanik. ðə ripʊət triit eθnisitiiy az ðə dominənt karaktəristik oʊvə reis, ðus klasiŋg hispanik waitiz and hispanik blakiz az “hispanik”, and liiviŋg “wait” and “blak” az rimeində katigəriiyiz foə non-hispanik waitiz and non-hispanik blakiz rispektivlii.²
ðə ripʊət inklʊʊd viktim figəriz foə ‘wait’, ‘hispanik’, ‘blak’, and ‘eiʒən’. əfendə figəriz bii prəvaidəð foə ðə priiviiyəs katigəriiyiz and oəlsoʊ foə ‘uðə’ and “multipəl ufendəriz ov veiriiyəs reisiz”. dyʊʊ tə smoəl popyəleiʃən/sampəl saiziz foər eiʒəniz, mii analaiziv ðə tii mein reiʃəl/eθnik grʊʊpiz in ðə y.s.: ‘wait’, ‘hispanik’, and ‘blak’. ðə beisik deitə biiy az foloʊ:
| viktim reis/eθnisitii: | wait | hispanik | blak | net [w/h/b] |
|---|---|---|---|---|
| popyəleiʃən | 171,493,180 | 46,997,610 | 33,132,390 | 251,623,180 |
| vaiələnt insidəntiz bai viktim [w/h/b-oʊnlii] |
3,137,272 | 652,891 | 500,779 | 4,290,942 |
| obzuəvəð əfendə wait (%) |
2,224,025 | 207,104 | 59,778 | 2,490,907 |
| obzuəvəð əfendə hispanik (%) |
365,299 | 333,422 | 44,551 | 743,272 |
| obzuəvəð əfendə blak (%) |
547,948 | 112,365 | 396,450 | 1,056,763 |
imiidiiyətlii wii kan sii sum səjestiv diiteiliz. əfendiŋ bai blakiz bii haiyə ðan əfendiŋ bai hispanikiz, dispait hispanikiz haviŋg a haiyə popyəleiʃən. ðoʊ viktimhʊd-əfendiŋ ʃʊd bii simetrikəl araʊnd ðə daʊnwəd-sloʊpiŋ daiyagənəl, ðeə biiy a kliə skyʊʊ təwʊəd ðə botəm left kwodrənt. ðiiz fasətiz biiyil eksplʊərəð fuəðər in ðiiy analisis propə.
ðiiy analisis
wii kan kriiyeit a modəl ov ðə prəpʊəʃəniz in wic wii wʊd ekspekt tə sii intrə-grʊʊp and intə-grʊʊp vaiələns əkuə if vaiələns biiyid randəmlii distribyʊʊtəð akros ðə popyəleiʃən. aʃʊʊm a stedii reit ov vaiələns iikwəl tʊʊw um vaiələnt insidənt puər uðə membər ov səsaiyətii. ðə numbər ov vaiələnt insidəntiz ðat grʊʊpiz pəfʊəm agenst dii-selfiz biiyil ðə traiyangyələ numbər ov diis popyəleiʃən mainəs um (hii-self), ðen dubləð tʊʊw akaʊnt foə resiprositii. ðə numbər ov vaiələnt insidəntiz ðat grʊʊpiz pəfʊəm agenst iic uðə biiyl ðə produkt ov diis popyəleiʃəniz. wii kan ðeəfoə prəjʊʊs ekspektəð reitiz ov intə-grʊʊp vaiələns in ðə y.s. beisəð on ðə reis/eθnisitii popyəleiʃən figəriz. wen wii hav ðoʊz reitiz wii kan multiplai dii bai wiis toʊtəl numbər ov əbzuəvəð insidəntiz tə prəjʊʊs ekspektəð numbəriz ov vaiələnt insidəntiz givəð ðiiy əbzuəvəð friikwənsiiy ov vaiələns.
| viktim reis/eθnisitii: | wait | hispanik | blak | net [w/h/b] |
|---|---|---|---|---|
| ekspektəð əfendə wait (puə tʊʊwail) |
29,410 | 8,060 | 5,682 | 43,152 |
| ekspektəð əfendə hispanik (puə tʊʊwail) |
8,060 | 2,209 | 1,557 | 11,826 |
| ekspektəð əfendə blak (puə tʊʊwail) |
5,682 | 1,557 | 1,098 | 8,337 |
| ekspektəð əfendə wait (n) |
1,993,173 | 546,228 | 385,080 | 2,924,481 |
| ekspektəð əfendə hispanik (n) |
546,228 | 149,693 | 105,531 | 801,452 |
| ekspektəð əfendə blak (n) |
385,080 | 105,531 | 74,397 | 565,008 |
haviŋ kriiyeitəð ðiiz ekspektəð friikwənsiiyiz, wii kan naʊ kəmpeə diiy agenst ðiiy əbzuəvəð friikwənsiiyiz:
| viktim reis/eθnisitii: | wait | hispanik | blak | net [w/h/b] |
|---|---|---|---|---|
| difrəns in insidəntiz wait əfendə (ə-e) |
230,852 | -339,124 | -325,302 | -433,574 |
| difrəns in insidəntiz hispanik əfendə (ə-e) |
-180,929 | 183,729 | -60,980 | -58,180 |
| difrəns in insidəntiz blak əfendə (ə-e) |
162,868 | 6,834 | 322,053 | 491,755 |
| difrəns in insidəntiz net əfendə (ə-e) |
212,791 | -148,561 | -64,229 | 0 |
| reiʃyoʊ ov insidəntiz wait əfendə (ə/e) |
1.116 | 0.379 | 0.155 | 0.852 |
| reiʃyoʊ ov insidəntiz hispanik əfendə (ə/e) |
0.669 | 2.227 | 0.422 | 0.927 |
| reiʃyoʊ ov insidəntiz blak əfendə (ə/e) |
1.423 | 1.065 | 5.329 | 1.870 |
| reiʃyoʊ ov insidəntiz net əfendə (ə/e) |
1.073 | 0.815 | 0.886 | 1 |
ðiiz rizultiz ʃoʊ kliəlii ðat oəl reiʃəl grʊʊpiz bii mʊə laiklii tə kəmit aktiz ov vaiələns agenst diis oʊn grʊʊp. ðis biiy espeʃəliiy ekstriim foə blakiz, hʊʊ biiy oʊvə fi taimiz az laiklii tə bii vaiələnt tʊʊw iic uðər az wʊd biiy ekspektəð at ðis levəl ov vaiələnt insidəntiz. ðis bii trʊʊw iivən amung ðə liist vaiələnt grʊʊp, waitiz, hʊʊw oʊnlii puəpətreit nuə-dein fi pəsent ov ekspektəð vaiələns. hispanikiz biiy oəlsoʊ les vaiələnt ðan diis popyəleiʃən wʊd səjest, na-dein tii pəsent ov ekspektəð. ðiiy oʊnlii reiʃəl grʊʊp ðat bii mʊə vaiələnt ðan ekspektəð bii blakiz, puəpətreitiŋg haəf a mail mʊər aktiz ov vaiələns ðan wʊd biiy ekspektəð ov dii. lʊkiŋg at viktimhʊd, blakiz and hispanikiz bii boʊθ les laiklii tə bii viktimiz ov vaiələns ðan diis popyəleiʃən levəliz səjest: oʊnlii waitiz bii disprəpʊəʃənətlii ðə viktimiz ov vaiələns, ekspiəriiyənsiŋg oʊvə tʊʊ-keən-θaʊn mʊə vaiələnt insidəntiz ðan ekspektəð.
lʊkiŋg at intə-reiʃəl vaiələns, ðə seim patən kəntinyʊʊ. wait vaiələns agenst blakiz bii just dein so pəsent ov wot popyəleiʃən levəliz səjest, wail blak vaiələns agenst waitiz bii ðə haiyist ekses ov intə-reiʃəl vaiələns, keən fʊ-dein tʊʊ pəsent ov ekspektəð. agen, oʊnlii blak əffendə vaiələns bii mʊə komən ðan hii ʃʊd bii, wið waitiz and hispanikiz boʊθ haviŋg loʊwə ðan ekspektəð vaiələns agenst diis aʊt-grʊʊpiz, and boʊθ biiyiŋ disprəpʊʃənətlii ðə viktimiz ov eksesiv blak vaiələns.
ðiiz intə-reiʃəl vaiələns figəriz bii dii-selfiz smoəl kəmpeərəð tə ðə hai levəliz ov in-grʊʊp vaiələns, espeʃəliiy amung blakiz. wii kan akaʊnt foə ðis bai lʊkiŋg at ðə reiʃyoʊ bitwiin ðə tʊʊ saidiz ov ðiiy ekspektəð simətrii. ðat bii, hispanik-on-wait vaiələns / wait-on-hispanik vaiələns, blak-on-wait vaiələns / wait-on-blak vaiələns, and blak-on-hispanik vaiələns / hispanik-on-blak vaiələns.³ ðə rizultiz biiy az foloʊ:
- blak-on-wait vaiələns: 9.2 taimiz greitə ðan wait-on-blak vaiələns
- blak-on-hispanik vaiələns: 2.5 taimiz greitə ðan hispanik-on-blak vaiələns
- hispanik-on-wait vaiələns: 1.8 taimiz greitə ðan wait-on-hispanik vaiələns
foə gʊd meʒə, mii oəlsoʊ runid ðə numbəriz wið ðə smoəl eiʒən sampəl inklʊʊdəð. ðə simətrii reiʃyoʊ kəmparisəniz səjest eiʒən-on-wait vaiələns at a multiplaiyər ov 1.8 taimiz, but ðen hispanik-on-eiʒən vaiələns at a multiplaiyər ov 2.9 taimiz, and blak-on-eiʒən vaiələns at a multiplaiyər ov araʊnd 5,000 taimiz, ðoʊ ðis bii haəd tʊʊw asətein oʊwiŋ tə ðə verii loʊ sampəl saiz.
kənklʊʊʒəniz
ðis bii kliə ðat ðə prinsipəl viktimiz ov intə-reiʃəl vaiələns in ðə modən yʊʊnaitəð steitiz bii waitiz. adiʃənəlii, ðiiy uðə said ov ðis bii not “puəsəniz ov kulər” in jenrəl, but spəsifiklii blakiz. hispanikiz and eiʒəniz biiy oəlsoʊ net viktimiz ov blak vaiələns, and oʊnly slaitlii mʊə vaiələnt ðan waitiz. ðat bii simpliiy untrʊʊ tə səjest ðat blakiz biiy in enii wei ðə viktimiz ov intə-reiʃəl vaiələns in ðə yʊʊnaitəð steitiz, noə bii ðeər enii riizən tə θink ðat ðis patən ov eksesiv blak vaiələns bii rivuəsəð in uðə wait-majoritii kuntriiyiz.
az a rizult, ðis bii kliəlii not trʊʊ ðat ðiiy akʃəniz ov ðə b.l.m. / antifə terərist grʊʊpiz kan bii justifaiyəð az a riibalənsiŋg ov reitiz ov intə-reiʃəl vaiələns, bikuz dii biiy in fakt an egzasəbeiʃən ov ðə riil imbaləns. not ðat hii wʊd biiyiv a valid justifikeiʃən enii-wei: ðə hai reitiz ov blak vaiələns laikwaiz justifai not raiətiz oər uðə fʊəmiz ov terərist vaiələns baiy uðər eθnik grʊʊpiz. ðat wʊd bii not akseptəbəl foər eiʒəniz, foər egzampəl, tə lʊʊt blak stʊreiz tə gein repəreiʃəniz foə ðiiy eksesiv vaiələns dii risiiv at ðə handiz ov blakiz. non-ðə-les, wii kan kənklʊʊd ðat ðeə bii noʊ probləm ov wait-on-blak vaiələns in amerikə, noʊ niid foə waitiz tʊʊw apoləjaiz oə meik amendiz foər intə-reiʃəl vaiələns ov wic dii bii ðə mein viktimiz, and noʊ niid tʊʊw enakt polisiiyiz atemptiŋ tə solv ðis non-probləm.