ARC LAB · Data Explorer
Advancing Research in Corrections · UNO School of Criminology & Criminal Justice
Experimental effects of a restrictive housing step-down program

Can people move to less restrictive housing without increasing misconduct?

A randomized evaluation in the Oregon Department of Corrections compared the Step-Up Program (SUP) with continued placement in the Intensive Management Unit (IMU), examining misconduct both during restrictive housing and after return to the general prison population.

Preregistered hypotheses · OSF
211
People randomized to SUP or continued IMU housing
52%
Assigned to the Step-Up Program (109 of 211)
47.7%
Of those assigned to SUP actually transferred to the unit
6 mo.
Official misconduct follow-up after return to general population
Research questions
Question 01

Does the Step-Up Program change misconduct while people are still in restrictive housing?

The study examines all, violent, and non-violent misconduct using both official records and self-reports.

Question 02

Does moving to Step-Up change misconduct after return to general population housing?

The study follows official misconduct for six months after participants leave restrictive housing.

FIGURE 1 · STUDY DESIGN

Everyone started in the same restrictive setting

People already housed in the Intensive Management Unit were randomly assigned to remain in the IMU or to be offered a transfer to the Step-Up Program. The study then measured misconduct during restrictive housing and, using official records, for six months after return to the general population.

Hypotheses preregistered in 2020 · Open Science Framework
From restrictive housing to outcomes
Starting point

IMU

Everyone in the randomized sample was already housed in long-term restrictive housing.

Random assignment

Two paths forward

Eligible participants were randomly assigned either to remain in the IMU or to be offered a transfer to SUP.

Continue IMU
n = 102
Assigned SUP
n = 109
Misconduct outcomes
01
During restrictive housing

While participants were still in IMU or SUP

Official records were observed between randomization and the end of the restrictive-housing stay.

Self-report: interviews also measured misconduct during restrictive housing.
02
After return to general population

Six-month follow-up

Official records captured misconduct during the six months following return to the general prison population.

Interview timing: about two weeks before the transition back to general population.
Why randomization matters: the treatment assignment itself was random. That gives the SUP and continued-IMU groups a common starting point, so differences in later misconduct can be interpreted as experimental estimates rather than simply reflecting pre-existing differences.
See the pre-randomization characteristics and equivalence checks
Full randomized sample

N = 211

102 assigned to IMU; 109 assigned to SUP.

Age: 32.0 years (SD 8.9)
Gang affiliation: 78.7%
Prior IMU: 40.3%
Examples of baseline characteristics

Groups were broadly similar

The published equivalence table compares age, race/ethnicity, gang affiliation, mental health designation, risk score, offense characteristics, prison history, prior misconduct, and IMU history.

Gang affiliation: IMU 77.5% · SUP 79.8% · p = .679
Pre-randomization misconduct: 1.2 vs. 1.1 violations/quarter · p = .538
IMU duration: 1.3 vs. 1.2 months · p = .682
What randomization did not eliminate

Small imbalances can occur

Of 24 comparisons in the full and interview samples, most were statistically equivalent; a small number of differences occurred at conventional thresholds and were adjusted for in the regression analyses.

Full sample: one difference at p < .05 and one at p < .10.
Interview sample: no differences at p < .05 and one at p < .10.
See the published study timeline and interview target · adapted from Figure 2
Before randomization

DSU → IMU

Participants spent an average of 142 days in the Disciplinary Segregation Unit before transfer to IMU.

After randomization

IMU → SUP or remain IMU

The full study cohort averaged 5.1 months from randomization to the end of restrictive housing. Those who transferred to SUP spent an average of 51 days in IMU before the move.

Interview timing

~2 weeks before return

Interviews were intended to occur shortly before participants returned to general population; 101 were conducted remotely and 11 in person.

See the participant flow from randomization to analysis · adapted from Figure 3 (CONSORT)
Allocation

211 randomized

IMU: n = 102
SUP: n = 109

Interview follow-up

112 consented

IMU: 50 · SUP: 62. Eleven people were missing the self-report misconduct section because of partial interviews and were excluded from that analysis.

Analysis

Administrative + survey

Official administrative data were available for all 211 randomized participants; the self-report analyses used the resulting 101-person interview sample.

THE INTERVENTION

SUP changed the conditions of restrictive housing

The Step-Up Program combined environmental changes, additional opportunities and programming, and a different approach to staff–prisoner interaction. It was intended to provide a less restrictive and more rehabilitative setting than the IMU.

5–6 hours
targeted daily out-of-cell time in SUP, compared with about 1 hour in IMU

More than a housing change

SUP combined changes to the physical environment, social interaction, programming, case management, recreation, and staff culture.

01 · OUT-OF-CELL TIME

More time outside the cell

The program was designed to provide 5–6 hours of out-of-cell time per day, compared with about 1 hour available in the IMU.

02 · SOCIAL CONTACT

More interaction

Participants had more opportunities for social interaction, including outdoor recreation with others and greater access to staff.

03 · PROGRAMMING

More rehabilitative services

SUP expanded access to treatment and recreational activities, including Free Your Mind in Segregation, DBT, GOGI, Living Well, HHAAP, and Charting a New Course.

04 · CASE MANAGEMENT

More individualized support

A correctional case counselor coordinated treatment plans and matched participants with programs based on their risks and needs.

05 · STAFF APPROACH

A more humanistic officer role

Correctional officers in SUP were encouraged to adopt a more humanistic approach, including participating in activities such as card or video games with prisoners. The authors note that this did not occur in the IMU.

The intervention was a package

SUP was not a single treatment component. The study evaluates a bundle of changes in living conditions, socialization, treatment access, recreation, case management, and staff culture.

FIGURE 2 · OFFICIAL MISCONDUCT

What happened to official misconduct?

The study estimates the difference between SUP and IMU housing. Click between the two outcome windows and estimators. Points are estimates; lines are 95% confidence intervals.

Outcome window
Estimator
View the published estimates (b values)
Published regression coefficients from Table 3. The coefficient is the estimated difference from the IMU reference group.
OutcomeITTpLATEp
FIGURE 3 · SELF-REPORT

Self-reports told a similar story inside restrictive housing

Because interviews were conducted before participants returned to general population, self-report outcomes were available only for misconduct during restrictive housing.

Estimator
View the published estimates (b values)
Published regression coefficients from Table 4. The coefficient is the estimated difference from the IMU reference group.
OutcomeITTpLATEp
All misconduct−0.003.974−0.005.973
Violent misconduct−0.052.350−0.078.335
Non-violent misconduct+0.001.995+0.001.995
HOW TO READ THE RESULT

The key contrast is inside restrictive housing versus after return

The experimental results are more nuanced than a single “worked” or “didn’t work” finding.

01 · During restrictive housing

A mixed pattern

The official-record estimates were mostly null. Violent misconduct was the exception in the LATE analysis: b = +0.097, p = .043. The authors interpret this partly through greater opportunity for interpersonal contact in SUP.

02 · After return to general population

No detected group differences

The official-record estimates did not show statistically significant differences between SUP and IMU across all three outcomes. The ITT coefficients were −0.021, +0.024, and −0.001.

WHAT TO KEEP IN MIND

The experimental test was strong; implementation was not complete

The authors identify several reasons to interpret the findings in context.

COVID-19

Program disruption

Enrollment was paused from May 2020 through March 2021, and the SUP operated in a modified format during part of the pandemic.

PARTIAL UPTAKE

Not everyone assigned to SUP transferred

Only 52 of the 109 people assigned to SUP agreed to move, which is why the study reports both ITT and LATE estimates.

STAFFING

The intended dosage was not always achieved

Staffing shortages meant the program did not consistently reach its intended 5–6 hours of out-of-cell time per day.

SAMPLE SIZE

Modest statistical power

The study included 211 randomized participants, and the interviewed sample was smaller, limiting the ability to detect subtle effects in self-reported outcomes.