Conditions & Operations
Prison Classification Systems & Violence: Misclassification, Overclassification, and Safety Failures
This GPS research document comprehensively examines prison classification systems, documenting how both overclassification and underclassification drive violence, waste resources, and violate constitutional protections. The DOJ's October 2024 investigation of Georgia prisons found 142 homicides between 2018-2023, staffing at only 50% of full levels, close-security inmates housed in medium-security facilities, and classification decisions driven by bed availability rather than risk—conditions described as 'among the most severe violations' uncovered in any DOJ prison investigation. GPS's own analysis has identified systematic classification mismatch in Georgia's medium-security prisons, while national research and the New Mexico case study demonstrate that proper classification could simultaneously reduce both costs and violence.
Pre-written explainers based on this research
Key Findings
The most impactful data from this research collection.
142
142 homicides in Georgia prisons (2018–2023)
StatisticGeorgia prison homicides jumped from 7 to 35 in 5 years
Trend50%
Georgia correctional officers at only 50% staffing levels
StatisticAll Data Points
56 verified data points extracted from primary sources.
142 homicides in Georgia prisons (2018-2023) Statistic
Between 2018 and 2023, there were 142 homicides in Georgia state prisons according to the DOJ investigation findings.
142 homicides
Georgia prison homicides grew from 7 to 35 (2018-2023) Trend
Homicides in Georgia prisons grew from 7 in 2018 to 35 in 2023, a five-fold increase over five years.
5 homicides at 4 prisons in one month in 2023 Statistic
In 2023, there were 5 homicides at 4 different Georgia prisons in one month alone.
5 homicides
GDC homicide underreporting: 6 reported vs. 18 in incident reports Data gap
GDC reported 6 homicides for the first 5 months of 2024 in mortality data, but at least 18 deaths were categorized as homicides in incident reports, indicating significant underreporting.
Georgia correctional officer staffing at only 50% of full levels Statistic
State prison census has doubled since 1990 while correctional officer staffing is at only 50% of full levels.
50%
Some Georgia prisons have staffing vacancy rates exceeding 60% Statistic
Some Georgia prisons have staffing vacancy rates exceeding 60%.
60%
Single officer responsible for 400 beds at close-security prison Case detail
At one close-security Georgia prison, a single officer was responsible for tracking 400 beds.
Walker State Prison noted as exception with better staffing and safety Finding
Walker State Prison was a notable exception among Georgia prisons, with fewer incarcerated people reporting fear and a higher proportion of staff positions filled.
Georgia prison census doubled since 1990 Trend
Georgia's state prison census has doubled since 1990.
DOJ: GDC leadership has lost control of its facilities Quote
The DOJ found that 'the leadership of the Georgia Department of Corrections has lost control of its facilities.'
DOJ described Georgia findings as 'among the most severe violations' in any DOJ prison investigation Finding
The DOJ October 2024 findings report was described as revealing 'among the most severe violations' uncovered in any DOJ prison investigation.
DOJ: Near-constant life-threatening violence in Georgia prisons Quote
DOJ investigators described conditions in Georgia prisons as involving 'near-constant life-threatening violence.'
DOJ: GDC places too much blame on gangs, insufficient emphasis on understaffing Finding
DOJ investigators emphasized that GDC places too much blame on gangs and insufficient emphasis on understaffing as the primary driver of disorder.
DOJ: Close-security inmates housed in medium-security facilities in Georgia Finding
The DOJ found that close-security inmates ('escape risks, have assault histories, deemed dangerous') are housed in medium-security facilities not designed or staffed for that population in Georgia.
DOJ: Classification decisions driven by bed availability rather than risk assessment Finding
The DOJ found that classification decisions in Georgia prisons appear driven by bed availability rather than risk assessment.
DOJ: GDC does not adequately screen, classify, or track LGBTI individuals Finding
The DOJ found that GDC does not adequately screen, classify, or track LGBTI individuals to ensure their safety. Transgender women often housed with male inmates face heightened assault risk due to inadequate classification.
DOJ: Violent incidents consistently underreported and mischaracterized Finding
The DOJ found that violent incidents in Georgia prisons are consistently underreported due to lack of supervision and mischaracterized using inappropriate incident-type categories.
DOJ recommended GDC reevaluate housing and inmate classification process Policy
The DOJ explicitly recommended that GDC 'reevaluate the housing and inmate classification process' as one of its minimum remedial measures.
Kristen Clarke's remedial measures: staffing, classification, and reporting Quote
Assistant Attorney General Kristen Clarke's remedial measures included: 'adding supervision and staffing, fixing the classification and housing system, and correcting deficiencies when it comes to reporting and investigations.'
Staffing-classification-violence nexus mechanism in Georgia Finding
The DOJ identified a causal chain: (1) chronic understaffing at 50% or below, (2) inability to conduct basic daily counts or maintain supervision, (3) gangs fill the vacuum controlling housing units, (4) incarcerated people can unlock their own cell…
Georgia has 7 close-security prisons Statistic
Georgia has 7 close-security prisons for offenders who are escape risks, have assault histories, are considered dangerous, or have detainers for other serious crimes.
7 close-security prisons
Georgia has 14 medium-security prisons Statistic
Georgia has 14 medium-security prisons, the largest category. Offenders have no major adjustment problems, and most may work outside the prison fence under constant supervision.
14 medium-security prisons
Georgia Diagnostic and Classification Prison intake: 100-125 inmates per day Statistic
Approximately 100-125 diagnostic inmates arrive per day from county jails at the Georgia Diagnostic and Classification Prison (GDCP) in Jackson, Georgia.
100 inmates per day (100-125)
GDC diagnostic process takes 7-15 working days Policy
The diagnostic classification process at GDCP takes 7-15 working days, during which all inmates are considered close security.
GDC uses Next Generation Assessment (NGA) classification tool Policy
Georgia's classification tool is the Next Generation Assessment (NGA), which generates a security level using an automated algorithm that weighs factors including sentence length, nature of the crime, criminal history, history of violence, medical a…
GDC policy: Sex offenders can never be classified below medium security Policy
Under GDC policy, sex offenders can never be classified below medium security. Sexual predators have additional restrictions.
GDC wardens can override NGA-generated classification level Policy
The NGA generates a recommended security level which is reviewed by the Warden/Superintendent for approval. Wardens can submit override requests if they determine the system-generated level is inappropriate.
Nelson Mandela Rules call for lowest security category consistent with safety Legal fact
The Nelson Mandela Rules (UN Standard Minimum Rules for the Treatment of Prisoners) call for using the lowest security category consistent with safety and control requirements.
Overclassification bias: 'better to be safe than sorry' Finding
Research shows a systematic bias toward overclassification because correctional officials consider it 'better to be safe than sorry.' A false negative prediction (placing a high-risk inmate too low) may result in violence, death, or escape, and offi…
Women offenders tend to be overclassified by current classification tools Finding
Custody classification systems used today tend to overclassify women into higher risk categories than warranted by their behavior, increasing limitations on their freedoms and access to programming. Classification tools work better for male offender…
Federal BOP minimum-security per capita cost: $21,006/year Statistic
Federal Bureau of Prisons minimum-security per capita cost is $21,006 per inmate per year.
$21,006
Federal BOP high-security per capita cost: $33,930/year Statistic
Federal Bureau of Prisons high-security per capita cost is $33,930 per inmate per year.
$33,930
Federal community supervision (probation) cost: $3,433/year Statistic
Federal community supervision (probation) costs $3,433 per offender per year.
$3,433
New Mexico overclassification costs up to $28 million/year Statistic
Overclassification deviations from the scoring tool cost New Mexico up to $28 million per year.
$28M
New Mexico: 60% of new inmates scored minimum, only 29% housed there Statistic
In New Mexico, while 60% of new inmates from 2014-2016 scored at minimum security, only 29% were actually housed there — most were in medium security instead.
29% vs. percent scoring at minimum security
New Mexico minimum-security cost: $11,183/inmate/year Statistic
New Mexico minimum-security units cost $11,183 annual security cost per inmate in FY2019.
$11,183
New Mexico maximum-security cost: $37,585/inmate/year Statistic
New Mexico maximum-security units at all other facilities cost $37,585 per inmate per year in FY2019.
$37,585
1999 New Mexico Santa Rosa prison guard murder linked to classification failure Case detail
In 1999, a guard was murdered at the medium-security Santa Rosa prison — minutes after an inmate was stabbed in another part of the prison, one day after another serious assault, and nine days after an inmate was murdered in his cellblock. A review …
1980 PNM prison riot killed 33 inmates Case detail
The deadly 1980 Penitentiary of New Mexico (PNM) prison riot killed 33 inmates. The state was subsequently required under the Duran consent decree to implement a formal classification system, which it had lacked entirely prior to the riot.
Overcrowding linked to increased violence in California prisons Finding
Research on 14 California state prisons (2018 and 2022) found that populations exceeding original design capacity were associated with increased violence — both inmate-on-inmate and inmate-on-officer assaults and batteries. Overcrowding affects inma…
Perpetrators of staff assaults more likely to be younger, gang affiliated, higher security Finding
A study of New Zealand prisons (2016-2020) found that prisoners who assaulted staff were more likely to be younger, gang affiliated, and had higher security classifications compared to prisoners who did not assault staff. Perpetrators of serious vio…
Close-custody violence predictors: younger, prior prison violence, violent convictions Finding
Cunningham & Sorensen (2007) found that among close-custody prisoners, those who were younger, had previously perpetrated violence in prison, and were convicted of violent crimes were significantly more likely to commit assaults resulting in serious…
Few states have validated classification instruments for predicting prison violence Data gap
Few states have validated their classification instruments for predicting prison violence specifically; many rely on post-release recidivism as the only outcome variable (not institutional violence).
Static criminal history factors are weak predictors of institutional misconduct for women Finding
Static criminal history factors alone are weak predictors of institutional misconduct, especially for women. Gender-responsive needs assessment improves prediction of women's institutional misconduct.
Long-term and older inmates often require minimal internal security regardless of offense history Finding
Long-term and older inmates often require minimal internal security regardless of offense history.
Marbury v. Warden: Deliberate indifference includes pervasive staffing issues Legal fact
In Marbury v. Warden, 936 F.3d 1227, 1235 (11th Cir. 2019), the court held that deliberate indifference may include evidence of 'pervasive staffing and logistical issues rendering prison officials unable to address near-constant violence, tensions b…
Van Riper v. Wexford: Understaffing policies causing violence violate Eighth Amendment Legal fact
In Van Riper v. Wexford Health Sources, Inc., 67 F. App'x 501, 505 (10th Cir. 2003), the court held: 'When prison officials create policies that lead to dangerous levels of understaffing and, consequently, inmate-on-inmate violence, there is a viola…
PREA requires zero tolerance for sexual abuse and specific classification screening Legal fact
28 C.F.R. Part 115 (PREA) requires zero tolerance for sexual abuse and sexual harassment and mandates specific classification screening to protect vulnerable populations — requirements the DOJ found Georgia violating, particularly for LGBTI individu…
Beard v. Livesay: Classification regulations create liberty interest requiring due process Legal fact
In Beard v. Livesay, 798 F.2d 874 (6th Cir. 1986), the court held that in the Tennessee prison system, statutes and regulations that clearly defined security classification created a liberty interest requiring due process hearings.
GPS identified systematic classification mismatch in Georgia medium-security prisons Finding
Georgia Prisoners' Speak has identified through data analysis that medium-security prisons in Georgia house close-security inmates at rates far above what would be expected — a systematic classification mismatch that places lower-risk inmates in dan…
Proper classification could reduce both costs and violence simultaneously Finding
If Georgia's classification system systematically overclassifies low-risk inmates into higher-security settings while simultaneously housing high-risk inmates in insufficiently secure facilities, the state wastes money (higher per-inmate costs at hi…
GDCP opened in 1969 in Jackson, Georgia Case detail
The Georgia Diagnostic and Classification Prison (GDCP) opened in 1969 and is located in Jackson, Georgia. All adult male felons enter through GDCP.
Nine evidence-based reform pathways for Georgia classification Policy
Evidence-based reforms include: (1) independent validation of NGA using Georgia-specific data, (2) regular audits of classification decisions vs. actual housing, (3) elimination of bed-space-driven overrides, (4) LGBTI-specific screening and housing…
NGA validation gap: no evidence of Georgia-specific validation Data gap
The document identifies the need for independent validation of the NGA classification instrument using Georgia-specific data, implying no such validation has been conducted. Without validation of scoring tools, it is impossible to determine whether …
Overclassification effects per UNODC Finding
According to the UNODC Handbook on the Classification of Prisoners, overclassification: exposes individuals to harsher conditions than necessary, hinders social reintegration, results in significantly higher operating costs, can lead to unfair discr…
Underclassification driven by available bed space rather than security risk Finding
Classification decisions are frequently driven by available bed space rather than security risk. When prisons are overcrowded, classification integrity breaks down.
Sources
9 cited sources backing this research.
Primary
Official report
California Legislative Analyst's Office, Improving California's Prison Inmate Classification System
Primary
Academic
Cunningham & Sorensen (2007), characteristics associated with serious prison violence
Primary
Official report
GDC, State Prisons and Classification Fact Sheets
Primary
Official report
Georgia Department of Corrections, Security Classification Policy 220.02
Primary
Official report
New Mexico Legislative Finance Committee, Policy Spotlight on Inmate Classification
Tertiary
Academic
Oxford Bibliographies, Prison Classification — Criminology
Primary
Official report
U.S. Department of Justice, Civil Rights Division, Investigation of Georgia Prisons Findings Report
Primary
Official report
UNODC, Handbook on the Classification of Prisoners, Criminal Justice Handbook Series
Primary
Academic
Urban Institute, The Growth & Increasing Cost of the Federal Prison System
Key Entities
Organizations, people, facilities, and other named entities referenced in this research.
Beard v. Livesay
[case]
Duran consent decree
[case]
Federal Bureau of Prisons
[organization]
Georgia Department of Corrections
[organization]
Georgia Diagnostic and Classification Prison
[facility]
Georgia Prisoners' Speak
[organization]
Kristen Clarke
[person]
Marbury v. Warden
[case]
National Institute of Corrections
[organization]
Nelson Mandela Rules
[legislation]
New Mexico Corrections Department
[organization]
Next Generation Assessment
[program]
Prison Rape Elimination Act
[legislation]
Southern New Mexico Correctional Facility
[facility]
U.S. Department of Justice
[organization]
UNODC
[organization]
Van Riper v. Wexford Health Sources
[case]
Walker State Prison
[facility]