Birmingham, Ala.

The Markup and Gizmodo have obtained and analyzed actual predictions for more than three dozen departments that used PredPol predictive policing software for at least six months between 2018 and 2020. This data sheet provides the findings from our disparate impact analysis and public housing analysis for Birmingham, Ala. To learn more about the project read, our investigation. For more details on how we did this analysis, read our methodology.

Findings

Overview

  • Predpol’s algorithm relentlessly targeted the block groups in each jurisdiction that were most heavily populated by people of color and the poor, particularly those containing public housing. The algorithm spared block groups with more White residents the same level of scrutiny.

  • The proportion of each jurisdiction’s Black and Latino residents was higher in the most-targeted block groups and lower in the least-targeted block groups compared to the jurisdiction overall. The opposite was true for the White population: The least-targeted block groups contained a higher proportion of White residents, and the most-targeted block groups contained a lower proportion.

  • For the majority of jurisdictions in our data set (27 jurisdictions), a higher proportion of their low-income households lived in the block groups that were targeted the most. In some jurisdictions, all of their subsidized and public housing was located in block groups PredPol targeted more than the median.

  • These vast disparities were caused by the algorithm relentlessly predicting crime in the block groups in each jurisdiction that contained a higher proportion of the low-income residents and Black and Latino residents. They were the subject of crime predictions every shift, every day, and in multiple locations in the same block group.

  • We also analyzed arrest statistics by race from the FBI’s Uniform Crime Reporting (UCR) Project for 29 of the agencies in our data that were in UCR. In 90 percent of them, per capita arrests were higher for Black people than White people—or any other racial group included in the dataset, mirroring the characteristics of the neighborhoods that the algorithm targeted.

  • We analyzed arrest data provided by 10 law enforcement agencies in our data and the rates of arrest in predicted areas remained the same whether PredPol predicted a crime that day or not.

Race and Ethnicity

Compared to Birmingham, Ala., overall, the most-targeted block groups had:

  • A greater proportion of Asians residents.
  • A greater proportion of Black residents.
  • A greater proportion of Latino residents.
  • A smaller proportion of White residents.

Compared to Birmingham, Ala. overall, the least-targeted block groups had:

  • A smaller proportion of Asian residents.
  • A smaller proportion of Black residents.
  • A smaller proportion of Latino residents.
  • A greater proportion of White residents.

Targeting Level Demographic Proportion of Block Group pop.
Most Targeted Block Groups Asian 1.1
Most Targeted Block Groups Black 50.3
Most Targeted Block Groups Latino 1.1
Most Targeted Block Groups White 30.9
Median Targeted Block Groups Asian 0.0
Median Targeted Block Groups Black 56.7
Median Targeted Block Groups Latino 0.9
Median Targeted Block Groups White 22.7
Least Targeted Block Groups Asian 0.0
Least Targeted Block Groups Black 21.7
Least Targeted Block Groups Latino 0.2
Least Targeted Block Groups White 64.8
Jurisdiction Total Asian 0.4
Jurisdiction Total Black 49.5
Jurisdiction Total Latino 0.6
Jurisdiction Total White 33.9

Household Income

Compared to Birmingham, Ala. overall, the most-targeted block groups had:

  • A greater proportion of households that made less than $45K a year.
  • A greater proportion of households that made between between $75K and 100k a year.
  • A smaller proportion of households that made between $120k and 150K a year.
  • A smaller proportion of households that made $200K and above a year.

Compared to the Birmingham, Ala. overall, the least-targeted block groups had:

  • A smaller proportion of households that made less than $45K a year.
  • A greater proportion of households that made between $75K and 100K a year.
  • A greater proportion of households that made between $120K and 150K a year.
  • A greater proportion of households that made $200K and above a year.

Targeting Level Demographic Proportion of Block Group pop.
Most Targeted Block Groups $120k - 150k 0.9
Most Targeted Block Groups $75k - 100k 3.9
Most Targeted Block Groups $200k and above 0.3
Most Targeted Block Groups Less than 45k 54.7
Median Targeted Block Groups $120k - 150k 0.6
Median Targeted Block Groups $75k - 100k 3.8
Median Targeted Block Groups $200k and above 0.3
Median Targeted Block Groups Less than 45k 36.5
Least Targeted Block Groups $120k - 150k 1.4
Least Targeted Block Groups $75k - 100k 6.3
Least Targeted Block Groups $200k and above 3.7
Least Targeted Block Groups Less than 45k 26.5
Jurisdiction Total $120k - 150k 1.0
Jurisdiction Total $75k - 100k 3.7
Jurisdiction Total $200k and above 2.3
Jurisdiction Total Less than 45k 38.7

Public Housing

In Birmingham, Ala. 36 percent of public housing was on block groups the software targeted the most, 89 percent of public housing was on block groups the software targeted more than the median.

The table below provides how many predictions each block with public housing received. The final column tells us the percentage of days a block received predictions from PredPol’s software between Sep 01, 2019 and Jan 30, 2021. We confirmed these dates with the Birmingham, Ala., police department.

Census GEOID Block Predictions Num. Public Housing Units Pct. days w/ Predictions
010730131001 1001 1611 1 99.8069498
010730131001 1008 2234 9 99.8069498
010730039001 1003 1725 2 98.8416988
010730052003 3001 1209 2 98.0694981
010730024001 1015 1477 8 97.8764479
010730112091 1016 1117 2 96.7181467
010730055002 2024 1327 19 95.3667954
010730130021 1010 1083 30 95.1737452
010730020002 2026 1106 1 94.7876448
010730045002 2001 921 29 94.5945946
010730005002 2005 833 5 94.4015444
010730005002 2010 1088 8 92.8571429
010730040005 5020 1333 1 92.4710425
010730023031 1014 864 21 91.5057915
010730007002 2014 901 6 91.1196911
010730005001 1029 665 3 87.0656371
010730023031 1007 618 23 83.0115830
010730029001 1048 767 13 82.4324324
010730030023 3006 775 8 81.2741313
010730055002 2033 668 9 81.2741313
010730029001 1051 522 8 80.5019305
010730023031 1008 717 2 78.9575290
010730029001 1025 566 4 77.9922780
010730131001 1009 536 5 75.0965251
010730004005 5001 568 5 74.5173745
010730023031 1020 508 4 72.0077220
010730130022 2002 471 8 70.8494208
010730004004 4000 384 2 70.2702703
010730055002 2027 470 6 70.0772201
010730007001 1023 512 1 70.0772201
010730005002 2012 452 5 63.8996139
010730023034 4004 372 6 63.1274131
010730036006 6003 367 2 61.5830116
010730029001 1052 457 10 61.3899614
010730023034 4001 391 11 61.1969112
010730045002 2002 403 10 59.2664093
010730007002 2011 544 10 57.1428571
010730024001 1014 310 1 55.7915058
010730027003 3056 357 1 55.4054054
010730045002 2003 449 9 54.8262548
010730023034 4002 325 7 54.2471042
010730007002 2004 309 1 53.2818533
010730131001 1013 352 6 53.2818533
010730029001 1050 316 8 51.9305019
010730130021 1011 318 20 48.0694981
010730048002 2002 286 1 48.0694981
010730030023 3003 285 10 46.9111969
010730030023 3004 259 8 45.1737452
010730032002 2001 269 2 44.2084942
010730032002 2019 279 2 43.8223938
010730051011 1002 325 6 43.6293436
010730024002 2060 269 8 43.0501931
010730045002 2005 342 3 42.8571429
010730031005 5013 240 3 42.6640927
010730005002 2011 273 6 40.9266409
010730023031 1003 240 4 40.1544402
010730027001 1174 227 1 38.4169884
010730023031 1004 262 5 37.4517375
010730023031 1016 216 25 35.9073359
010730027001 1073 211 4 28.9575290
010730057021 1011 146 1 27.9922780
010730005002 2016 186 4 27.6061776
010730011002 2025 169 2 27.2200772
010730029001 1053 143 5 25.8687259
010730024001 1012 151 7 25.8687259
010730005001 1028 145 7 25.8687259
010730027002 2008 140 1 25.0965251
010730011003 3008 141 1 25.0965251
010730055002 2043 144 2 24.7104247
010730007002 2016 141 2 23.5521236
010730032002 2003 131 2 23.5521236
010730027001 1095 141 2 22.0077220
010730024001 1016 136 4 21.8146718
010730024001 1010 182 2 21.4285714
010730030023 3000 148 15 20.0772201
010730040004 4005 120 1 17.9536680
010730029001 1049 99 19 17.9536680
010730027002 2012 106 1 17.1814672
010730032002 2002 98 1 16.7953668
010730032002 2018 91 5 16.0231660
010730020002 2004 109 2 15.0579151
010730029003 3026 74 1 14.2857143
010730023033 3001 85 2 14.2857143
010730032002 2000 82 1 13.8996139
010730030023 3001 90 5 13.7065637
010730029002 2034 68 2 13.1274131
010730027001 1106 67 1 12.7413127
010730051011 1011 88 8 12.1621622
010730027002 2014 86 5 11.9691120
010730032002 2008 62 2 11.5830116
010730030023 3002 62 6 10.8108108
010730040004 4000 69 1 10.8108108
010730005001 1012 60 3 10.8108108
010730027002 2016 66 3 9.8455598
010730045002 2006 56 1 9.8455598
010730027001 1097 56 2 9.2664093
010730008001 1057 47 1 8.6872587
010730057023 3010 45 1 8.6872587
010730051011 1009 51 4 8.1081081
010730030023 3005 42 3 7.7220077
010730034003 3005 42 1 6.3706564
010730143021 1031 39 1 6.1776062
010730005001 1013 32 8 5.9845560
010730024003 3029 31 1 5.7915058
010730027002 2001 27 1 5.2123552
010730030023 3026 30 1 5.2123552
010730021001 1021 28 2 4.6332046
010730029001 1022 23 6 4.2471042
010730023031 1002 23 5 3.4749035
010730008001 1061 25 2 3.4749035
010730003001 1053 15 2 2.8957529
010730039001 1104 11 2 2.1235521
010730027001 1074 9 4 1.7374517
010730051011 1004 9 1 1.7374517
010730005002 2019 7 1 1.3513514
010730133003 3008 6 12 1.1583012
010730024005 5008 6 1 1.1583012
010730024004 4013 7 1 1.1583012
010730005001 1014 5 1 0.9652510
010730027002 2007 3 5 0.5791506
010730023031 1019 4 10 0.5791506
010730023032 2013 3 2 0.5791506
010730005002 2002 2 2 0.3861004
010730023031 1001 2 7 0.3861004
010730032002 2016 2 4 0.3861004
010730130022 2001 2 1 0.3861004
010730048001 1008 2 1 0.3861004
010730005002 2006 1 1 0.1930502
010730051011 1008 1 5 0.1930502
010730001005 5036 1 1 0.1930502

Maps

Predictions

Density Map

The map below aggregates all the predictions Birmingham, Ala., received in our analysis window into a 2D grid. Each square of the grid represents an area approximately 500 ft. x 500 ft., the size of the PredPol prediction box. The color represents the number of predictions that occurred within the square. The more predictions, the darker the square.

Prediction Count Description
0 - 189 Least Predictions
189 - 377
377 - 564
564 - 752
752 - 940
940 - 1130 Most Predictions

Sources: Markup, Predpol

The grid drawn on this map provides an approximate aggregation of the prediction data. The actual prediction box in the reports provided to departments will vary from the ones shown above.

Choropleth

This map shows the predictions aggregated to the level of the Census block group. Aggregating prediction data to the geographic area of a Census block group introduces additional complexity to the analysis, and hence this map should be interpreted with some caution. See the limitations section of the methodology for more details.

Source: Markup, Predpol

Race and Ethnicity

Black

Source: 2018 five-year ACS.

Latino

Source: 2018 five-year ACS.

White

Source: 2018 five-year ACS.

Household Income

Less than $45k

Source: 2018 five-year ACS.

$75k - $100k

Source: 2018 five-year ACS.

$125k - $150k

Source: 2018 five-year ACS.

$200k and above

Source: 2018 five-year ACS.

Methods

We analyzed the distribution of PredPol predictions for Birmingham, Ala. at the geographic level of a Census block group, which is a cluster of blocks with a population of between a few hundred to a few thousand people, generally. There are 279 block groups in Birmingham, Ala., the smallest block group had a population of approximately 141 and the largest had a population of approximately 8,344.

In Birmingham, Ala., we analyzed 327,476 predictions and used there locations to determine the block groups that were targeted the most, the median and the least. This data sheet presents the breakdown of the racial groups and household income ranges of the people who lived in those block groups. We also present the breakdowns for Birmingham, Ala. overall for comparison. The predictions we analyzed were between Sep 01, 2019 and Jan 30, 2021 , we received confirmation that Birmingham, Ala. department used the software between Sep 01, 2019 and Jan 30, 2021.

For the race/ethnicity and income analyses, we merged 2018 five-Year American Community Survey data and prediction data and observed the makeup of block groups that were targeted above and below the median, those targeted the most and those targeted the least. For the sake of consistency in our analysis we only used demographic groups for which we had reliable population estimates for all the jurisdictions in our data set. These are:

  • Racial Groups
    • Black
    • Asian
    • Latino
    • White
  • Household Income
    • Less than $45K
    • Between $75K-$100K
    • Between $125K-$150K
    • Greater than $200K

Definitions

We used the Census’ “designated place” boundaries as the boundaries for most jurisdictions. For Sheriff’s departments we confirmed the boundaries with the department.

We defined the most-targeted block groups as those in Birmingham, Ala. which encompassed the highest five percent of predictions. We defined the median-targeted block groups as the five percent around the median block group for predictions. And we defined the least-targeted block groups as those with the bottom five percent of predictions.

In some of the larger jurisdictions, more than five percent of block groups got zero predictions. In those cases, we chose the most populated block groups with no predictions for the five percent. Learn more about how we did this in our methodology.

We identified public housing through HUD’s online lookup tool available at https://resources.hud.gov

Data

The data used to generate this analysis can be found in our GitHub repository. It also contains the URLs for the rest of the data sheets from our analysis.