United States v. Crisp

Case Information

E-FILED

Friday, 14 March, 2025 01:36:40 PM Clerk, U.S. District Court, ILCD IN THE UNITED STATES DISTRICT COURT CENTRAL DISTRICT OF ILLINOIS URBANA DIVISION UNITED STATES OF AMERICA, Plaintiff, v. Case No. 2:13-cr-20050-JEH-1 DAVID L. CRISP, JR.,

Defendant. Order and Opinion

Now before the Court is the Defendant, David L. Crisp’s, motion for early termination of supervised release. For the reasons stated, infra , the motion is granted. ^[1]

I A

David Crisp’s trouble with the law began at age seventeen, when he picked up a conviction for possession of cannabis and received a two-year sentence of probation. It didn’t take long for his probation to be revoked, resulting in a two- year stint in the Illinois Department of Corrections. While on parole for that offense, he committed new offenses of criminal trespass to land and another controlled substance offense. He again received a two-year sentence of probation, again had his probation revoked, and again received a two-year sentence in the Illinois Department of Corrections.

Between the ages of nineteen and twenty-one, he had two domestic battery convictions and one for resisting a police officer, all of which resulted in fines. Then, at age twenty-one, he received a six-year sentence for a third controlled substance offense. After being paroled, he was again imprisoned for another parole violation.

After again being released at age twenty-three, he committed his third domestic battery for which he received a conditional discharge sentence, and he then committed his fourth controlled substance offense at age twenty-four while on conditional discharge, for which he received a new four-year term in prison. While on parole again at age twenty-seven, he was sent back to prison for two years for an obstruction of justice offense. After release for that offense, he committed additional misdemeanor drug and other offenses until he was arrested on the federal charge in this case.

B

By the time a federal grand jury charged Crisp in this case on September 25, 2013, with possession with intent to distribute more than twenty-eight grams of crack cocaine, he was in his mid-thirties, had not gone past the ninth grade in school, had a checkered employment history when not in prison, had no money or job, had six prior felony convictions (four of which were narcotics-related offenses), had multiple misdemeanors (including four convictions for domestic battery), and multiple violations of various forms of supervision. Crisp eventually pleaded guilty to the federal charge, admitting to possessing with intent to distribute 41.4 grams of crack cocaine.

At sentencing for this offense, Judge Michael P. McCuskey, the presiding judge at the time, found Crisp to be a career offender pursuant to U.S.S.G. 4B1.1. Even without the career offender enhancement, Crisp had twenty criminal history points, which was seven more than necessary to be in the highest possible criminal

2 history category of six. Two of Crisp’s prior drug convictions increased his statutory range of penalties to a minimum of ten years and a maximum of a lifetime in prison. Regarding a supervised release term, Title 21 U.S.C. § 841(b)(1)(B) required a minimum supervised release term of at least eight years. Crisp had a guideline range of 262 to 327 months’ imprisonment. The court varied downward from the guideline range and imposed a sentence of 240 months’ imprisonment and the statutory minimum eight-year term of supervised release. In varying from the bottom of the guideline range of 262 months down to 240 months, Judge McCuskey noted that Crisp’s allocution demonstrated rehabilitative potential, as well as his exceptional acceptance of responsibility. (D. 33 at ECF pp. 39-40).

Crisp then appealed his sentence, challenging various conditions of supervised release. The Seventh Circuit vacated his sentence and remanded the case to this Court for a full re-sentencing hearing. At the time of the re-sentencing hearing on July 29, 2015, Crisp had spent a little over a year in prison on the current conviction. Judge James E. Shadid, the new presiding judge over the case at the time of re-sentencing, had this to say:

However, Judge McCuskey did make an observation about Mr. Crisp that Mr. Crisp has proven to be true. He pointed out that he thought that Mr. Crisp had great rehabilitative potential. And since your incarceration, Mr. Crisp, just a little over a year ago, if I do the math right, you have taken advantage of that and you have proven Judge McCuskey correct. You have completed 60 hours of adult education, 12-hour courses in algebra, geometry, psychology, commercial driver's license, African American history and creative writing, and then another 12 hours in parenting skills, drug education, all the while incurring no additional or no disciplinary violations.

So I believe that you should be credited for your post-offense rehabilitation as this will reduce the likelihood of re-offending. 3 (D. at ECF 62 pp. 30-31). Accordingly, Judge Shadid imposed an even lower sentence of 168 months’ imprisonment, but again imposed the mandatory minimum term of eight years of supervised release.

C

The rehabilitative efforts Crisp made during his first year in prison continued throughout his lengthy time in prison. After his re-sentencing, he earned his GED while in prison. He also completed a 120-hour Computer Concepts and Applications course, the Residential Drug Abuse Treatment Program, the Non-Residential Drug Treatment Program, and Adult Continuing Education courses entitled: “African American History” (in addition to the African American History class he took before his re-sentencing), “Money Smart,” and “Healthy Living.” He finished his entire term of incarceration without receiving a single disciplinary write-up.

In April of 2023, Crisp moved to a halfway house to finish his term of imprisonment. In early July of that year, the Bureau of Prisons released him to home confinement to finish his term of imprisonment, which he did on August 23, 2023. Also in July 2023, he found employment at Reliable Limo & Charter in Kankakee, Illinois, primarily driving travelers to and from Chicago’s O’Hare International Airport to various locations in the greater Chicago area— employment which he has maintained throughout the course of his supervised release term. He lives with his fiancée, who he will marry in July. In the eighteen months he has been on supervision, he has not violated any of the conditions. His current probation officer confirmed that Crisp has never provided a positive drug test and was taken off routine drug testing in November 2024. All he does now is electronically file monthly reports.

4

II

On January 27, 2025, a little over seventeen months into his supervision, Crisp filed a counseled motion to terminate his supervised release term early. He argues that given his perfect conduct while incarcerated and while on supervised release, he has demonstrated that he has been rehabilitated, does not pose a threat to society, and is in no need of services that might be available while on supervised release.

In support, he cites to the Judicial Conference of the United States’ Guide to Judiciary Policy , which, according to Crisp, encourages early termination of supervised release when a person has “substantially satisfied the requirements of the court order[] and . . . [] demonstrated a willingness and capability to remain lawful beyond the period of supervision.” Guide to Judiciary Policy¸ Volume 8E, Ch. 3, § 360.20. He also cites to a recent study performed by Thomas H. Cohen, the Branch Chief of the Data Unit for the Probation and Pretrial Services Office of the Administrative Office of the United States Courts (AO) entitled, Early Termination: Shortening Federal Supervision Terms Without Endangering Public Safety. See Appendix A to this Order and Opinion.

In this paper, Cohen examined data extracted from an administrative dataset maintained by the AO that included 296,023 persons whose supervision terms were successfully closed in the 94 federal district courts during the ten-year period encompassing fiscal years 2014 through 2023. Id. at 11. His examination reveals that, after matching individuals with similar offense and personal characteristics, supervisees with early termination are two percentage points less likely to recidivate (ten percent rearrested) than their regular termination counterparts (twelve percent rearrested). Id. at 19. Cohen therefore concludes that “supervisees granted early termination under current policies pose no greater risk to the community than those who serve a full term of supervision.” Id. at 23.

5 According to Crisp, this conclusion supports his argument that the early termination of his supervised release will not pose a greater risk to the community than if he was required to complete his entire eight-year term.

The Government opposes the early termination of Crisp’s supervised release. Although acknowledging his perfect conduct in federal prison and while on supervision, the Government argues that this Court should deny Crisp’s motion “based on his extensive criminal history, the limited percentage of supervised release term served, and the underlying conduct in this matter.” (D. 103 at ECF p. 2). ^[2] More specifically, the Government argues that the seriousness of Crisp’s offense as evidenced by the statutory mandatory maximum

sentence, his extensive criminal history, his past failures to comply with supervision, and the fact that he has served less than twenty percent of his originally imposed supervised release term all weigh against early termination. The Government also argues that terminating Crisp’s supervision early would create a disparity with other defendants similarly situated where the statutory mandatory minimum term of supervision is eight years. (D. 103 at ECF p. 6).

III A 1

Title 18 of the United States Code, Section 3583(e) governs a Court’s early termination of supervised release. Sub-section (1) allows a Court to “terminate a term of supervised release and discharge the defendant released at any time after the expiration of one year of supervised release.” 18 U.S.C. § 3583(e)(1) (emphasis added). Importantly, beyond the one-year requirement set forth in the text of the statute, attempts by courts to erect other arbitrary time barriers to considering motions for early termination have been rejected.

For example, in United States v. Lowe , 632 F.3d 996, 998–99 (7th Cir. 2011), the Seventh Circuit found that a district court abused its discretion by refusing to consider motions for early termination until the final twelve months of a defendant’s supervision term. In doing so, the court stated:

We find that this unexplained, clearly arbitrary policy certainly circumvents the intent and purposes of 18 U.S.C. § 3583(e)(1). Section 3583(e)(1) clearly provides an individual with the opportunity to submit a motion for early termination of supervised release “any time after the expiration of one year of supervised release.” Though § 3583(e)(1) gives the court discretion in granting a motion for early termination of supervised release, the district court’s failure to even consider such motions until twelve months before the probation’s end-date completely disregards the statute it must follow.

Lowe , 632 F. 3d 998-99. In the present case, the Government does not dispute that Crisp meets the one-year requirement set forth in the statute but argues that serving less than twenty percent of the originally imposed term is not enough time on supervision to warrant early termination. When asked at oral argument, the Government confirmed to the Court that the United States Attorney’s Office in the Central District of Illinois has a general policy of opposing any motions for supervised release where a defendant has not served at least two-thirds of the originally imposed term, although the policy is subject to exception with approval.

The Government, of course, is entitled to adopt any lawful policy it likes. However, the weight of the Government’s opposition to a motion is diminished when it is premised upon a counter-textual, arbitrary timeframe beyond the one year set by Congress. The same is true for any recommendation by the probation office based on a similar arbitrary policy. Congress has evidenced its intent

7 through the language of the statute that the only per se time bar to a Court’s ability to grant a motion for early termination is less than one year, and this Court must respect that intent. All of this is not to say that a court cannot consider the length of time a defendant has spent on supervision when deciding an early termination motion, but only in conjunction with an individualized determination of the other mandatory statutory factors which a court must consider, as discussed, infra .

2 Considering the approximately eighteen months Crisp has spent on supervision without a single violation or problem weighs in favor of an early termination. No doubt the reason Congress allowed for early termination motions to be considered after the first year of supervision is because that first year out of prison and back into society is likely to be the most difficult; if a defendant can get through that year successfully, he has provided persuasive evidence of rehabilitation and lack of danger to the public.

Crisp’s circumstances demonstrate this proposition. For approximately ten years, he was removed from society serving his prison term in this case. As the changes in society largely left him behind over those ten years, he one day in 2023 found himself out of prison, without a job, home, money, or income. Notwithstanding these circumstances, and with the assistance of the U.S. Probation Office and it services, he found employment, a stable home, and began to rebuild his life. Indeed, contrary to what happened after his release on other occasions, this time he fully complied with all his conditions and did not break the law again.

This stark difference in Crisp’s behavior between his release in this case and his previous releases from custody is persuasive evidence that something is different this time. Ironically, his abysmal criminal history and performance on prior forms of supervision makes his success on federal supervision the best

8 evidence of his reformation. If he were to make the decision to turn to a life of crime again, he would have likely done so when he was first released from custody, like he did on other occasions, when his circumstances were most difficult. Having not done so for eighteen months and having built a good life for himself with a stable job, income, home, and relationships, the likelihood that he will now or in the future throw that all away is slim given his behavior while on supervision.

B

Tuning to the § 3553(a) factors, if a defendant has spent at least one year on supervision as Crisp has, a court must next consider the factors set forth in Title 18, Section 3553, sub-sections “(a)(1), (a)(2)(B), (a)(2)(C), (a)(2)(D), (a)(4), (a)(5), (a)(6), and (a)(7). . . .” 18 U.S.C. § 3583(e). Conspicuously absent from this list of § 3553(a) factors is sub-section (a)(2)(A).

1 a

The above list of (a) factors specifically omits sub-section (a)(2)(A), which provides that a court, in determining the particular sentence to be imposed shall consider the need for the sentence imposed “to reflect the seriousness of the offense, to promote respect for the law, and to provide just punishment for the offense.” 18 U.S.C. § 3553(a)(2)(A). Congress included this factor among those a court must consider when imposing a sentence and when terminating a term of probation, but it specifically excluded this factor from a court’s consideration when terminating a term of supervised release. 18 U.S.C. §§ 3553(a), 3583(e), & 3564(c).

When it comes to revoking a term of supervision, as opposed to terminating one, there is a split among the circuits regarding the implication of this factor’s exclusion from § 3583(e). Five circuits, including the Seventh, have held that a court may consider this factor at least when deciding whether to revoke a term of

9 supervised release pursuant to § 3583(e)(3), notwithstanding Congress’s exclusion of sub-section (a)(2)(A) from § 3583(e). See United States v. Esteras , 88 F.4th 1163 (6th Cir. 2023), cert. granted, 145 S. Ct. 413 (2024); United States v. Vargas-Davila , 649 F.3d 129, 132 (1st Cir. 2011); United States v. Williams , 443 F.3d 35, 47 (2d Cir. 2006); United States v. Young , 634 F.3d 233, 239 (3d Cir. 2011); United States v. Clay , 752 F.3d 1106, 1108 (7th Cir. 2014). Four other circuits have held that a court may not consider the sub-section (a)(2)(A) factors when deciding whether to revoke a term of supervision. See United States v. Crudup , 461 F.3d 433, 439 (4th Cir. 2006); United States v. Miller , 634 F.3d 841, 844 (5th Cir. 2011); United States v. Miqbel , 444 F.3d 1173, 1182 (9th Cir. 2006); United States v. Booker , 63 F.4th 1254, 1261 (10th Cir. 2023). Although this split relates to revocations governed by § 3583(e)(3), the question of whether (a)(2)(A)’s exclusion from § 3583(e) applies equally to early terminations under sub-section (e)(1). See 18 U.S.C. § 3583(e)(1) & (3). The Supreme Court will hopefully resolve this question in a case currently pending before it, Esteras v. United States , in which the Court granted certiorari on the following question:

Even though Congress excluded section 3553(a)(2)(A) from section 3583(e)’s list of factors to consider when revoking supervised release, may a district court rely on the section 3553(a)(2)(A) factors when revoking supervised release?

145 S.Ct. at 413. At least for the time being and until the Supreme Court resolves this question, this Court is bound to follow the Seventh Circuit’s decision in Clay , which allows a court to consider in the revocation context the omitted (a)(2)(A) factor “so long as the district court relies primarily on the factors listed in § 3583(e), including the nature and circumstances of the violations, the history and characteristics of the defendant, the need to protect the public, and the need for adequate deterrence.” Clay , 752 F.3d at 1108. What Clay does not hold, however, is

10 that a Court must consider sub-section (a)(2)(A) factors—only that a court may , notwithstanding its exclusion from the list in § 3583(e).

b Recognizing that a court may consider the (a)(2)(A) factors under current circuit precedent, this Court exercises its discretion in this case not to do so. First, Congress created supervised release as “a unique method of post- confinement supervision.” Gozlon-Peretz v. United States , 498 U.S. 395, 407 (1991). Unlike parole (which supervised release replaced), a supervised release term is a separate part of the sentence, typically imposed at the judge’s discretion, to follow a determinate prison term. See 18 U.S.C. § 3583(c). And unlike parole, “supervised release wasn’t introduced to replace a portion of the defendant’s prison term, only to encourage rehabilitation after the completion of his prison term.” United States v. Haymond , 588 U.S. 634, 652 (2019) (emphasis in original); see also United States v. Johnson , 529 U.S. 53, 59 (2000) (“Congress intended supervised release to assist individuals in their transition to community life. Supervised release fulfills rehabilitative ends, distinct from those served by incarceration.”); S. REP. 98-225, 124 (“[T]he primary goal of such a term is to ease the defendant’s transition into the community after the service of a long prison term for a particularly serious offense, or to provide rehabilitation to a defendant who has spent a fairly short period in prison for punishment or other purposes but still needs supervision and training programs after release.”).

The Sentencing Reform Act of 1984 (SRA)’s text reflects Congress’s different goals for the different sentencing provisions. To punish the offender, Congress instructed courts to impose prison, probation, or a fine for every offense. See 18 U.S.C. § 3551. And when imposing any of these sanctions, Congress instructed courts to consider all the § 3553(a) factors, including the retribution factors in sub- section (a)(2)(A): the need for the sentence to reflect the seriousness of the offense,

11 to promote respect for the law, and to provide just punishment for the offense. See 18 U.S.C. § 3582(a) (prison); 18 U.S.C. § 3562(a) (probation); 18 U.S.C. § 3572(a) (fine).

Reflecting the rehabilitative goals of supervised release, though, the SRA omitted (a)(2)(A)’s retributive factors from the list of 3553(a) factors a court must consider when imposing a term of supervised release at sentencing. See Sentencing Reform Act of 1984, § 3583(c), 98 Stat. at 1999; S. REP. 98-225, 124 (“The Committee has concluded that the sentencing purposes of incapacitation and punishment would not be served by a term of supervised release.”). And Congress used the same list, reflecting the same goals, when instructing courts what to consider when addressing a request for early termination of supervised release. See Sentencing Reform Act of 1984, § 3583(e), 98 Stat. at 2000; see also S. REP. 98-225, 124 (noting that § 3583(e) directs the court to “consider[ ] the same factors considered in the original imposition of a term of supervised release”).

Comparing supervised release with probation is also instructive. Because judges may select probation as the complete sentence for an offense, see 18 U.S.C. § 3551, Congress instructed courts to consider all the applicable § 3553(a) factors when imposing it, see Sentencing Reform Act of 1984, § 3562(a), 98 Stat. at 1992. But because supervised release follows a prison sentence that the sentencing court has concluded satisfies the § 3553(a) factors, and because supervised release is a discretionary supplement that fulfills more limited goals, Congress did not require courts to consider the sub-section (a)(2)(A) when imposing a supervised release term. See Sentencing Reform Act of 1984, § 3583(c), 98 Stat. at 1999; see also S. REP. 98-225, 125 (“The term of supervised release is very similar to a term of probation, except that it follows a term of imprisonment and may not be imposed for purposes of punishment or incapacitation since those purposes will have been served to the extent necessary by the term of imprisonment.”).

12 This difference between probation and supervised release applies to requests for early termination as well. In its response to Crisp’s motion, the Government cited 18 U.S.C. § 3564(c) as the standard governing this Court’s decision, stating that “a Court must consider the factors set forth in 18 U.S.C. § 3553(a), to the extent applicable,” and setting forth the complete list of § 3553(a) factors. (D. 103 at ECF p. 8). As the Government conceded at oral argument, however, § 3564(c) applies to early termination of probation, not supervised release. The statute governing early termination of supervised release is § 3583(e)(2), which specially omits (a)(2)(A) from the list of factors a court must consider.

The difference between the probation termination and supervised release termination provisions makes sense in light of the different purposes served by them. Historically, a probation sentence was based on “suspend[ing] . . . a defendant’s prescribed prison term and afford[ing] him a conditional liberty as an ‘act of grace,’ subject to revocation.” Haymond , 588 U.S. at 643. The SRA retained from prior practice that probation “remains conditional and subject to revocation until its expiration or termination.” Sentencing Reform Act of 1984, § 3564(e), 98 Stat. at 1994. Thus, for a probation termination, the SRA instructed courts to consider all the applicable § 3553(a) factors. See Sentencing Reform Act of 1984, § 3565(a), 98 Stat. at 1995. For supervised release, in contrast, the SRA restricted what factors courts may consider—including precluding consideration of § 3553(a)(2)(A)’s retribution factors—and limited courts’ options for addressing a violation. See Sentencing Reform Act of 1984, § 3583(e), 98 Stat. at 2000. Those textual differences convey different meanings that reflect Congress’s different goals.

All of this is to say that, although allowed by current circuit precedent to consider the (a)(2)(A) retributive factors when deciding to terminate supervised release early, this Court exercises its discretion not to do so in light of the purposes

13 of supervised release. Punishment is not a purpose of supervised release and, accordingly, considering the (a)(2)(A) punitive factors when deciding whether to terminate a supervised release term early does not make much sense, even if allowed by the Seventh Circuit’s interpretation of § 3583(e).

2 Turning to factors Congress explicitly requires a court to consider when deciding whether to terminate supervision early, § 3553(a)(1) and § 3583(e) mandate a court consider the nature and circumstances of the offense and the history and characteristics of the defendant.

When considering the nature and circumstances of the offense, it is important to note that this consideration is not the same as the “seriousness” of the offense, listed at § 3553(a)(2)(A). As already stated, sub-section (a)(2)(A) is specifically excluded from the list of factors required to be considered by a court when imposing or terminating early supervised release. 18 U.S.C. §§ 3583(c) & (e). On the other hand, the “nature and circumstances of the offense” is listed as a factor to be considered when imposing a prison term and when imposing or terminating early supervision. 18 U.S.C. § 3553(a)(1); 18 U.S.C. §§ 3583(c) & (e). Congress clearly distinguished between the two by placing them in separate subsections in § 3553(a) and treating them differently for purposes of imposing a term of imprisonment and a term of supervised release. See Kungys v. United States , 485 U.S. 759, 778 (1988) (opinion of Scalia, J.) (noting the “cardinal rule of statutory interpretation that no provision should be construed to be entirely redundant”).

It is also significant that Congress placed both “seriousness of the offense” and “history and characteristics of the defendant” in the same section, i.e. , § 3553(a)(1). This placement of two factors in a single subsection is different than how Congress treated these same two factors for purposes of pre-trial release. Specifically, when listing the factors a court must consider when determining

14 whether a defendant should be granted pre-trial release, Congress set forth “the nature and circumstances of the offense” in § 3142(g)(1) but placed “history and characteristics of the defendant” in § 3142(g)(3). A court should presume Congress had a reason for this difference in treatment and not presume that the differences in treatment between § 3142(g) and § 3553(a) was “a simple mistake in draftmanship.” Pulsifer v. United States , 601 U.S. 124, 172 (2024) (quoting Russello v. United States , 464 U.S. 16, 23 (1983)).

Considering these textual differences and given that in the context of imposition or termination of supervision, as contrasted with imprisonment, the primary goal is rehabilitation (which also protects the public), the combined factors of “nature and circumstances of the offense” and the “history and characteristics of the defendant” should be considered together in such a way as to inform how these factors impact the goals of supervised release.

Performing this analysis here demonstrates the point. Crisp’s offense of conviction is for possessing with intent to distribute more than 28 grams of crack. Setting aside the seriousness of the offense, see supra, Section III.B.1.b., the nature and circumstances of this offense provide little insight into Crisp’s potential for rehabilitation and future dangerousness. However, when viewed in conjunction with his history and characteristics, different pictures emerge.

For example, at the time of Crisp’s original sentencing hearing, his multiple prior controlled substance convictions, other multiple convictions, and poor performance on various forms of parole and supervision signaled that the chances of him committing future controlled substance and other offenses were quite high, thereby necessitating a relatively lengthy term of supervised release to both promote rehabilitation and protect the public from future criminal activity. Even without the eight-year mandatory minimum supervised release term, in other words, a substantial term of supervised release was warranted.

15 At the time of Crisp’s re-sentencing hearing, however, the picture had somewhat changed. As Judge Shadid noted, in the year Crisp had spent in prison prior to the re-sentencing, Crisp had taken significant steps toward rehabilitation. Although Judge Shadid reflected this change in a reduced sentence of imprisonment, he could not do so in the length of the supervised release term because of the eight-year mandatory minimum term required by statute. See 21 U.S.C. § 841(b)(1)(B).

Now, upon consideration of the motion for early termination, however, Congress allows this Court to reduce or terminate supervision, taking into account the rehabilitative goals of supervision with a more complete picture. See 18 U.S.C. § 3583(e)(1). The picture Crisp presents today is quite different than presented to both Judge McCuskey at the original sentencing hearing and Judge Shadid at the re-sentencing hearing. For almost ten years, Crisp did not have a single disciplinary ticket while in prison. While incarcerated, he not only earned his GED, but availed him of numerous other courses and programs. Rather than returning to a life of crime upon his release from imprisonment, as he had done so many times before, Crisp found stable employment, a stable residence, built relationships, and has been in full compliance with all of his conditions. Accordingly, considering the nature and circumstances of Crisp’s offense and the complete picture of his history and characteristics, § 3553(a)(1) weighs heavily in favor of terminating his supervision; Crisp has been rehabilitated, and, accordingly, he does not present any greater risk to the public than others who are not under a form of court supervision.

3 Looking next to § 3553(a)(2)(B), (C), and (D) as required by § 3583(e), these factors weigh heavily in favor of early termination of Crisp’s supervision. Sub- section (a)(2)(B), relating to affording “adequate deterrence to criminal conduct,”

16 weighs in favor of early termination because, as a matter of specific deterrence, Crisp has already demonstrated his rehabilitation. The only effect supervised release has on Crisp at this point is the requirement he file an electronic monthly report. This requirement is not deterring him from returning to a life a crime; clearly, the benefits of living a healthy, stable, crime free life are doing the deterrence here.

Regarding general deterrence, terminating Crisp’s supervision early promotes it. Although being on supervision serves rehabilitative purposes, rather than punishment, there is no denying that being on federal supervision can be a heavy burden for supervisees. Dependent on the wide discretion of probation officers and judges, the sword of Damocles looms over a supervisee’s head, ready to strike in the form of a petition for revocation and potentially re-imprisonment, even for minor or technical violations. See e.g., United States v. Smith , 770 F.3d 653 (7th Cir. 2014) (defendant receiving fifteen-month sentence upon revocation for marijuana use and breaking halfway house rules).

Of course, the threat of re-imprisonment for violations, even minor or technical ones, can serve an important deterrent role for those on supervision, helping to incentivize rehabilitation. But the prospect of getting out from under the thumb of the criminal justice system early after demonstrating rehabilitation can also incentivize supervisees, as Crisp argues here. In other words, knowing that he could potentially reduce a multi-year supervision term by keeping his nose clean for at least a year can deter a supervisee from committing violations. By staying out of trouble in those initial, most difficult months after release, even if motivated only by both avoiding prison and reducing the supervised release term, a supervisee may begin to see the benefits of law-abiding life, promoting long- term, permanent rehabilitation. Therefore, terminating the supervision early of

17 someone like Mr. Crisp deters others from violating conditions of supervision by providing them with an additional incentive to turn over a new leaf.

Regarding sub-section (a)(2)(C) related to protecting the public from further crimes of the defendant, terminating a supervised release term, standing alone, does not present a greater risk to the public than requiring a defendant to complete the originally imposed term. The AO’s comparison of re-arrest rates for individuals whose supervision is terminated early compared with similar individuals who complete their terms demonstrates that the rearrest rates after supervision are lower for the early termination cohort. See Appendix A to this Order and Opinion. More specifically applying this factor to Crisp, his conduct in prison and while on supervision demonstrate his rehabilitation and that he is no more likely to commit crimes in the future than someone who would remain on supervision for several more years.

Sub-section (a)(2)(D) related to providing the defendant with needed education, training, medical care, and other correctional treatment, also weighs in favor of early termination. Crisp does not he need any services which being on supervision might offer him; he already receives no services related to his supervision and only files electronic reports monthly.

4 The final § 3553(a) factor to weigh is the “need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct. 18 U.S.C. § 3553(a)(6). ^[3] The Government argues that because Crisp’s offense required imposition of an eight-year term of supervised release at the time of sentencing, terminating his supervision after completing only eighteen months of that supervision would create an unwarranted disparity between him and other defendants receiving the eight-year mandatory minimum supervision term at sentencing. (D. 103 at ECF p. 6).

The Government is correct that if this Court terminates Crisp’s supervision after eighteen months, a disparity will exist between the supervision term he will have served in comparison to the term served by defendants whose original eight- year term of supervision was never terminated early. However, this disparity is not unwarranted , 18 U.S.C. § 3553(a)(6)(“unwarranted disparities”), but rather inherent in the fact the Congress specifically allows early termination of supervised release, even for those individuals who had a mandatory minimum term of supervision imposed originally. See, e.g. , United States v. Scott , 362 F. Supp. 2d 982 (N.D. Ill. 2005). A disparity created by terminating supervision early as allowed by Congress cannot be “unwarranted.”

Rather, what would create an unwarranted disparity would be to refuse to apply the law allowing for the early termination solely because an eight-year term was required at the time the sentence was imposed. Such a practice would set up an arbitrary bar to application of the law and would ignore the requirement that a court not only consider motions to terminate supervision after one year of supervision has elapsed but also consider all the factors set in § 3553(a) which are listed in § 3583(e). See also Lowe , 632 F.3d at 998–99 (prohibiting courts from erecting arbitrary bars to considering motions for early termination).

To be sure, unwarranted disparities do exist in the area of termination of supervised release . Arbitrary requirements set by judges, such as per s e requirements for time spent on supervision before an early termination motion will be considered and beyond the one-year threshold set by Congress create unwarranted disparities. See Lowe , 632 F.3d at 998–99. Some probation offices following and others not following the Judicial Conference’s polices for when to

19 consider and recommend early termination also create unwarranted disparities. See Guide to Judiciary Policy¸ Volume 8E, Ch. 3, § 360.20. ^[4]

Likewise, unwarranted disparities are created when some judges but not others erect other non-statutory, artificial barriers to the early termination of supervised release. For example, some Courts, including this one in the past, have required as a threshold requirement for early termination that a defendant have had absolutely no violations of supervised release of any kind during their entire time on supervision. United States v. Branscumb , No. 1:09-CR-10023-MMM-JEH, 2019 WL 6501208, at *3 (C.D. Ill. Nov. 12, 2019), report and recommendation adopted, No. 09-10023, 2019 WL 6499067 (C.D. Ill. Dec. 3, 2019). Indeed, some courts have gone so far as to state that having a perfect record on supervision is not enough to warrant early termination, stating that a defendant also should “demonstrate exceptionally good behavior or unforeseen circumstances warranting the early termination.” Id. (citing United States v. Kay , 283 F. App’x 944, 946–47 (3d Cir. 2008)) ; United States v. Lussier , 104 F.3d 32, 36 (2d Cir. 1997) ; Folks v. United States , 733 F. Supp. 2d 649, 651–52 (M.D.N.C. 2010) ; United States v. McKay , 352 F. Supp. 2d 359, 361 (E.D.N.Y. 2005) ; United States v. Caruso , 241 F. Supp. 2d 466, 468–69 (D.N.J. 2003) ; United States v. Medina , 17 F. Supp. 2d 245, 246–47 (S.D.N.Y. 1998) . Not only are these judge-made requirements beyond what § 3583(e) provides, but they are also of suspect origin.

The line of cases requiring this “perfect conduct plus” standard for early termination trace their origin to a 1997 case out of the Second Circuit, United States v. Lussier , 104 F.3d 32 (2d Cir. 1997). However, Lussier had nothing to do with early termination of supervised release, but instead, addressed “whether one component of a sentence—restitution—that was not challenged either on direct review or under 28 U.S.C. § 2255, may nonetheless be modified because restitution was made a condition of supervised release and supervised release may be modified pursuant to 18 U.S.C. § 3583(e)(2).” Id. at 33. This issue is a far cry from the proposition for which this case is cited in cases like Branscumb. Moreover, the statement in Lussier to the effect that § 3583(e) provides courts with authority to alter supervised release terms “in order to account for new or unforeseen circumstances” was created out of whole cloth, without citation to any authority.

21 Finally, the Second Circuit itself subsequently clarified this language in Lussier , stating that Lussier “does not require new or changed circumstances relating to the defendant in order to modify conditions of release, but simply recognizes that changed circumstances may in some instances justify a modification.” United States v. Parisi , 821 F.3d 343, 347 (2d Cir. 2016). By wrongly interpreting Lussier as requiring perfect conduct on supervision, courts applying this incorrect standard again create an unwarranted disparity in early terminations compared to courts who do not read that case to stand for something it does not. ^[5]

These types of non-statutory, judge-made artificial per se rules and requirements in the early termination context are what can create unwarranted disparities. Indeed, the identity of the district in which a defendant is sentenced is the single most significant fact in predicting whether an early termination motion will be granted, see Appendix A to this Order and Opinion at p. 17. The average percentage of early termination motions granted among the ninety-four federal districts range from less than five percent to over fifty-five percent. That disparity is unwarranted; any disparity created by terminating Crisp’s supervision early, where all the statutory factors weigh in favor of termination, is not.

IV

David Crisp is a success story, not only for himself but for the federal criminal justice system and society as a whole. At the time Judge McCuskey sentenced Crisp to 240 months’ imprisonment and eight years of supervised release, his personal history provided dim hope that he would be rehabilitated and no longer present a danger to the public. For reasons ultimately known only to Crisp, his time in prison on this occasion was different. He took advantage of every opportunity provided to him while in the Bureau of Prisons and, upon release, by the U.S. Probation Office. The purposes of sentencing as set forth in § 3553(a) were actually achieved in this case. Those § 3553(a) factors relevant to whether supervised release should be terminated early all weigh in favor of termination; the rehabilitative goals of supervised release have been achieved in Mr. Crisp’s case.

Accordingly, the motion to terminate supervised release early is granted. It is so ordered. Entered on March 14, 2025. s/Jonathan E. Hawley

U.S. DISTRICT JUDGE

23 Appendix A 2:13-cr-20050-JEH # 105 Filed: 03/14/25 Page 24 of 59 EARLY TERMINATION 1 Early Termination: Shortening Federal Supervision Terms Without Endangering Public Safety ^[1]

Thomas H. Cohen Branch Chief – Data Unit Probation and Pretrial Services Office Administrative Office of the U.S. Courts Abstract In the U.S. federal system, courts have the discretion to grant early termination of supervision for people on federal probation or supervised release under certain circumstances. This study sought to provide a profile of who was most likely to receive early terminations by examining 296,023 people whose supervision terms were successfully closed via early or regular termination during the 10 years encompassing fiscal years 2014 through 2023. Several research issues were explored, including the extent to which a supervisee’s risk profile correlated with the use of early termination and an examination of whether early terminations endangered community safety. Overall, research showed that this closure type had moderate correlations with a supervisee’s risk profile; however, other factors—including supervision time imposed and, most importantly, district of case closure—had the greatest bearing on whether a case would end by early termination. Moreover, using a process in which early and regular termed supervisees were matched on a variety of criteria associated with the risk of recidivism, the study found that early terminations did not threaten community safety. Early Termination: Shortening Federal Supervision Terms Without Endangering Public Safety IN THE FEDERAL supervision system, courts have the discretion to grant early termination of supervision for persons on federal probation or terms of supervised release (i.e., TSR) under certain circumstances. ^[2] Specifically, 18 U.S.C. §§ 3564(c) and 3583(e)(1) allow the court to terminate terms of probation in misdemeanor cases at any time and terms of supervised release or probation in felony cases after the expiration of one year of supervision if the court is satisfied that such action is warranted by the supervisee’s conduct and is in the interest of justice. In 2003 and 2005, the Judicial Conference promulgated policies that sought to increase the use of early termination to reduce officer workload burden and allow evidence to guide judicial decisions (Baber & Johnson, 2013). Specifically, in 2003, the Judicial Conference approved policies that encouraged officers to seek early termination whenever statutory eligibility was reached for those supervisees who have satisfied their conditions of supervision, have successfully reintegrated into the community, and did not pose a foreseeable risk to public safety generally or to any individual third party (Baber & Johnson, 2013).

In 2005, the Judicial Conference further revisited the early termination policy by allowing supervisees with outstanding balances on fines and restitutions to be considered for early termination as long as they were otherwise suitable and in compliance with their payment schedule (Baber & Johnson, 2013). The Judicial Conference in 2005 further revised the early termination policy by recommending provisions modeled after the U.S. Parole Commission regulations. Of particular importance, the Judicial Conference created a presumption in favor of early termination for non-career and non-violent supervisees who 1) have been under supervision for at least 18 months, present no identified risk to the public or victims, and are free from any moderate- or high-severity violations; or 2) have been under supervision for at least 42 months and are free from any moderate- or high-severity violations (Baber & Johnson, 2013).

These policies remain in effect today; as such, early termination is a practice that holds a great deal of promise as a positive incentive for persons under supervision and as a measure to contain costs in the judiciary without compromising the mission of public safety. While early terminations afford an opportunity both to contain costs and to move persons who present a low risk of re-offending off federal supervision, relatively little is known about how early terminations are implemented in the federal system and whether their application endangers community safety. The one study done on this topic is over ten years old and employed a relatively small number of matched cases ( n = 2,872) to examine the degree to which curtailing a person’s supervision term represented threats to public safety (Baber & Johnson, 2013).

The current study seeks to provide a fuller picture of the application of early terminations in the federal system by using a larger and newer dataset and more sophisticated statistical techniques. Specifically, the extant research will focus on examining the application of early termination in the federal system by exploring who receives this form of case closure and how key criteria, including a supervisee’s risk characteristics, most serious conviction offense, demographics, supervision time imposed, and district of case supervision might be associated with early termination. Then, the study will assess whether early termination endangers community safety by using matched samples of persons receiving early and full termination. Before delving into the study, a brief overview of early termination literature is provided. Literature on Early Termination The concept of early termination is based on the framework of goals-based supervision (Robina Institute of Criminal Law and Justice, 2020). Under the goals-based supervision framework, success is defined by a supervisee’s completion of specific requirements or programs rather than completion of an imposed supervision term that can last several years (Pew Charitable Trusts, 2020). There are a variety of terminologies that fall under the goals-based supervision rubric, including “early termination, earned discharge, earned compliance, and earned time credit.” (Petersilia, 2007; Robina Institute of Criminal Law and Justice, 2020: 1). The basic way these programs work is that the probation officer sets expectations that the supervisee will complete all program requirements and remain in compliance with all supervision conditions and that, should these expectations be met, the supervisee will be rewarded through the reduction of the probation/parole term (Jacobson et al., 2017; Pew Charitable Trusts, 2020; Robina Institute of Criminal Law and Justice, 2020; Smith et al., 2012). States that employ early termination programs do so through a variety of methods; in some states, a formula is used to shorten supervision terms, while other states allow reductions when a supervisee meets certain goals (e.g., successful program completion, payment of fines, fees, or restitution costs, etc.) (Pew Charitable Trusts, 2020; Robina Institute of Criminal Law and Justice, 2020). In most states where early termination is permitted, the discretion to grant an early discharge resides with judicial officials (Pew Charitable Trusts, 2020). State-Level Studies of Early Termination Programs Relatively few efforts to empirically evaluate early termination programs have occurred at the state level, and most have demonstrated that shortening supervision terms did not endanger community safety (Courtney et al., 2022; Jacobson et al., 2017; Pew Charitable Trusts, 2016). One notable study examined earned compliance in Missouri using an approach that compared the recidivism rates of supervisees who received early termination with a matched sample of supervisees discharged in that state prior to the policy’s implementation. Results showed no statistically significant differences in the recidivism rates between the early termed and comparison group; the policy reduced the supervision population by 18 percent (Pew Charitable Trusts, 2016). Another study conducted in New York City found that the aggressive implementation of early discharge policies generated an increase in the percentage of supervisees receiving this form of case closure from 3 percent to 17 percent; correspondingly, the early-term cohort manifested rearrest rates that were lower than a similarly situated group serving their full supervision terms (Jacobson et al., 2017: New York City Department of Probation, 2013). Another study, which examined the feasibility of reducing supervision terms in Oregon and South Carolina, found that more than 90 percent of probationers could have spent less time on supervision without compromising community safety; in addition, had these supervisees served the shortest supervision terms needed to minimize reoffending, the two states’ average daily supervision populations could have been reduced by 32 percent or more (Pew Charitable Trusts, 2020). Last, some studies focused on implementation challenges have found that early termination programs can sometimes be obstructed either by local criminal justice actors or by requirements that all financial conditions be satisfied before early termination is granted (Courtney et al., 2022; Griffin et al., 2013; Minnesota Department of Corrections, 2017; Smith et al., 2012). Federal-Level Studies of Early Termination Programs At the federal level, a study on early termination was conducted by Baber and Johnson in 2013. This study encompassed an examination of 2,872 supervisees whose cases were either early or fully termed in 2008, and these individuals were matched on a variety of criteria associated with the risk of recidivism, including risk prediction index score, criminal history category, gender, age, and district of case supervision. In general, the study found that persons serving their entire supervision term were nearly two times more likely to receive an arrest for new crimes after supervision compared to the early-termed cohort (19.2 percent vs. 10.2 percent, respectively). Moreover, the study showed that early-termed supervisees remained crime free for slightly longer periods than their full-termed counterparts (Baber & Johnson, 2013). Last, the study calculated that substantial savings could be realized through the application of early termination (Baber & Johnson, 2013).

While the Baber and Johnson (2013) study provided an important contribution to the understanding of early termination in the federal system, several key questions about this topic remain unanswered. Specifically, issues involving a profile of who receives early termination and how the application of early termination might vary by criteria associated with risk, conviction offenses, demographics, supervision time imposed, and the district of case supervision need addressing. Last, statistical techniques that allow for matching approaches encompassing a larger group of covariates than Baber and Johnson (2013) used are required to ascertain the extent to which early termination does or does not endanger community safety. Present Study The current study seeks to enhance the work begun by Baber and Johnson (2013) by exploring how early termination is being applied in the federal supervision system and examining whether the use of early termination could potentially endanger community safety. The initial part of this research is mostly descriptive and focuses on the ways that early termination is being applied for persons on federal community supervision (hereafter federal supervisees). The following issues form the main components of this research:

1. What percentage of federal supervisees with successful case closures (i.e., closed without a revocation or violation) receive an early termination, and has the early termination rate varied over time?

2. What is the profile of federal supervisees receiving early termination? To what extent does the granting of early termination vary by a supervisee’s risk characteristics as measured by the federal Post Conviction Risk Assessment (PCRA), most serious conviction offense, demographic characteristics (e.g., race/ethnicity, age, and gender), and supervision time imposed?

3. How much influence does the district of supervision have on the likelihood of early termination? Are supervisees in some districts more likely to receive an early termination than those in other districts?

4. How many months of supervision time are saved through the early termination process, and do the time savings vary by a supervisee’s risk profile, most serious conviction offenses, or supervision time imposed? The second part of this research delves into whether supervisees who receive early

termination are more, less, or equally likely to recidivate compared to their counterparts whose supervision terms ended through an expiration of their term. Using a matching approach (see methods) in which early- and full-termed supervisees are matched on a variety of criteria associated with the risk of recidivism, the following questions form this analysis:

1. Are persons who receive an early termination more or less likely to recidivate for any or violent offenses after completion of their supervision terms compared to a matched group of persons receiving full termination?

2. When rearrested, are early and full-term supervisees rearrested for similar offenses? Method Sample In order to address these questions, data were extracted from an administrative dataset maintained by the Administrative Office of the U.S. Courts (AO) that included 296,023 persons whose supervision terms were successfully closed in the 94 federal district courts during the ten- year period encompassing fiscal years 2014 through 2023. A successful closure means that the person either served the full term of supervision without any major violations resulting in a revocation (i.e., regular termination) or received an early termination of the supervision sentence (i.e., early termination). It should be noted that supervisees whose cases were closed unsuccessfully by revocation, violations, transfers, or other means were omitted from this study ( n lost = 132,511). ^[3] Supervisees with unsuccessful terminations, particularly revocations, were excluded from this study because these persons are by definition ineligible for early termination. Moreover, early terminations based on compassionate grounds, which comprised less than 1 percent of all successful closures, were also removed from this analysis ( n lost = 1,962). Last, persons placed on federal post-conviction supervision with successful case closures who were not U.S. citizens ( n lost = 18,230) or for whom rap sheet data were unavailable ( n lost = 9,051), were also removed from the study sample.

Table 1 provides a descriptive overview of the supervisees in the study sample. Among the population examined, 11 percent were rearrested for any offenses and 3 percent were rearrested for violent offenses within 24 months after their case closure date. About 37 percent of the study population comprised non-Hispanic White individuals, while Blacks accounted for similar portions of supervisees (36 percent). Hispanic individuals of any race comprised 23 percent of the study sample. Males accounted for 79 percent of the study population, and the average supervisee age was about 44 years (SD = 12.1). The offense conviction types that accounted for the largest portion of supervisees involved drug (45 percent), property/white collar (22 percent), and weapons/firearms offenses (13 percent); together these three offense types comprised 80 percent of the study sample. In regard to the PCRA risk distribution, which places supervisees into one of four risk categories (low, low/moderate, moderate, or high) (see Lowenkamp et al., 2015), similar percentages of supervisees were classified as either PCRA lows (43 percent) or low/moderates (43 percent) at their final assessments, while the remainder were PCRA moderates (11 percent) or PCRA highs (2 percent). Table 1 also provides details on the percentage of supervisees testing positive for illicit substances during their supervision terms and the average number of supervision months imposed on the study population.

_______________ Insert table 1 here ________________

Measures Early termination Early termination refers to instances where the court approves a request to shorten an individual’s supervision term. Although the request is typically initiated by the probation officer, it can also come from the federal defender. Though the general contours for early termination are provided by federal statute, the Judicial Conference in the early- to mid-2000s attempted to create a presumption in favor of early termination for supervisees meeting specified criteria (see Baber & Johnson, 2013). Among the 296,023 federal supervisees with successful case closures between fiscal years 2014 and 2023, 25 percent received an early termination (see Table 1).

Over time, the percentage of federal supervisees early termed has generally increased. Using administrative data that include all supervisees with successful case closures between fiscal years 2000 through 2023 (see Figure 1), the percentage of supervisees with early-term case closures increased from about 12 – 13 percent prior to the Judicial Conference changes (i.e., fiscal years 2000 through 2002) to 21 percent at the time when the Judicial Conference promulgated new policies about early termination (i.e., fiscal year 2005). During the 2005 to 2009 timespan, the percentage of supervisees placed on early termination declined from 21 percent to 18 percent; however, since then there has been a steady increase in the early termination rate, so that one year prior to the COVID-19 pandemic in 2019, 25 percent of supervisees had an early termination case closure. The early termination rate peaked at 28 percent during the COVID-19 pandemic year of 2020, declined slightly in 2021, and then increased back to 28 percent in 2023.

_______________ Insert Figure 1 here _________________

PCRA and Other Independent Variables The recidivism risk profile for supervisees receiving early and regular termination was measured through several PCRA risk domain scores. The PCRA is an actuarial risk assessment instrument used by federal officers to gauge a supervisee’s likelihood of being rearrested both during and after supervision. This instrument is a fourth-generation risk tool that can identify changeable dynamic risk factors and ascertain whether those factors have changed over multiple assessments (see Cohen et al., 2016; Johnson et al., 2011; Lowenkamp et al., 2013, 2015; Luallen et al., 2016). Implemented in 2009, the PCRA has nearly universal usage rates and has been shown to be a valid predictor of a supervisees’ risk of recidivism. Specifically, the PCRA risk tool manifests AUC-ROC scores in the .73-.74 range for any offenses and .76-.77 range for violent offenses (see Lowenkamp et al., 2015). These scores mean that the PCRA provides “good” to “excellent” predictive capacities (Desmarais & Singh, 2013) and hence, can be used as a risk classification measure for this research.

In this study, the supervisees’ risk profile was accounted for by controlling for several domain scores embedded within the PCRA instrument generated at the final assessment, including the criminal history (scores range 0-9), education/employment (scores range 0-3), substance abuse (scores range 0-2), social networks (scores range 0-3), and cognitions (scores range 0-1). In addition, the supervisees’ risk level at final assessment encompassing whether the supervisee was low, low-moderate, moderate, or high was incorporated as an additional risk control metric. Also included was the extent to which a supervisees’ risk categorization grouping changed between the first and final assessments that are measured to include either no changes or increases/decreases of plus or minus one or two or risk levels. Last, the score produced by the offender section of the PCRA that measures a supervisees’ general criminal thinking levels was included as another risk metric. The general criminal thinking score is based on the Psychological Inventory of Criminal Thinking Styles (PICTS), which has been shown to be a valid predictor of recidivism with AUC scores in the .65 range (Walters & Lowenkamp, 2016).

In addition to the PCRA, the supervisees’ age in years, drug use during supervision (both number of positive drug tests and positive tests for polysubstance use), fiscal year of case closure, most serious conviction offenses, number of conditions imposed, sex, supervision time imposed in months, and race/ethnicity were used as additional covariates for the current analysis. With the exception of age, number of positive drug tests, number of conditions imposed, and supervision time imposed, which are continuous variables, the remaining factors are captured using dummy (0/1) variable coding. Outcome Measures The primary outcome of interest involves whether a supervisee was rearrested within 24 months after their supervision case was closed. Rearrests for new criminal activity, which are also referred to as recidivism, were obtained from the National Crime Information Center (NCIC) and Access to Law Enforcement System (ATLAS). ATLAS is a software program used by the AO that provides an interface for performing criminal record checks through a systematic search of official state and federal rap sheets (Baber, 2010). Recidivistic events were defined to include arrests for either any (excluding arrests for technical violations) or violent offenses within 24 months of case closure. Violent arrest activity includes the following offenses: assault (including sexual), arson, kidnapping, murder, attempted murder, rape, robbery, or threats. Analytical Plan Descriptive statistics encompassing primarily crosstabs, means, and medians were used to examine the application of early termination for supervisees on federal post-conviction supervision. Statistical tests including chi-squares and ANOVAs were used to test for differences in the use of early termination across several covariates; moreover, effect sizes are reported when feasible to measure the magnitude of these relationships.

In addition to examining who receives early termination and how many months were saved through this method of closing cases, an assessment of the recidivism rates for the early and full-termed supervisees was conducted. When comparing the percentage of early- and regular-termed supervisees rearrested for new offenses after case closure, it is important to acknowledge that these two groups of supervisees differ in many ways on key domains associated with recidivism (see Figures 2 and 3, Table 2, and Appendix). Accounting for these differences is crucial when attempting to assess whether the use of early termination endangers community safety, and this issue was addressed through the application of propensity score matching (PSM) techniques to generate matched groups of supervisees who received early and regular termination. PSM has become a commonly employed technique to estimate treatment effects when randomized assignment is unavailable and it becomes necessary to account for covariates that could influence the outcome of interest (Rosenbaum & Rubin, 1983, 1985).

In this study, the “treated” supervisees constituted those who received early termination, while the “control” supervisees encompassed those with cases concluded through regular termination. Supervisees were matched on their final PCRA criminal history, education/employment, substance abuse, social networks, and cognition domain scores as well as their final risk levels (low, low/moderate, moderate, or high), and general criminal thinking scores. Moreover, supervisees were matched according to the extent to which their PCRA risk levels changed from their first to final assessments; this was one way to control for the presence of noncompliance between these two groups. Another way to account for noncompliance involved an effort to match early- and full-termed supervisees by the number of times they tested positive for any illegal substances and whether these supervisees tested positive for multiple substances (i.e., they were polysubstance users) during supervision. In addition to accounting for these recidivism risk criteria, supervisees were matched by most serious conviction offenses, felony vs. misdemeanor conviction, supervision time in months imposed, number of supervision conditions imposed, demographic characteristics (i.e., age, race/ethnicity, and sex), and fiscal year of case closure. It should be noted that matching by supervision district was explored, but ultimately not used because too many cases would have been lost by matching on this criterion and the results did not appreciably differ when supervision district was included in the matching algorithm. Last, a subset of the larger sample of early- and full-termed supervisees whose recidivism activity could be tracked for two years after case closure ( n = 244,941) was applied to the recidivism section of this research.

A two-step process was employed in the matching model. First, logistic regression was employed, in which the likelihood of a supervisee receiving an early termination was used to estimate the propensity scores. Then, the estimated likelihood scores were used to match the early termination group to a comparison group of supervisees with regular terminations, applying one-to-one nearest neighbor matching with a .0001 caliper setting (Guo & Fraser, 2014). This method resulted in matches being found for 77 percent of supervisees receiving an early termination. Results from the matching procedures (shown in the appendix) display the balance obtained between the early- and regular-termed supervisees. This appendix shows that the matching procedures generated early- and regular-termed groups of supervisees, showing strong balance on the key covariates of interest. Last, a more conservative alpha level of .001 was used to denote statistical significance because of the large sample sizes analyzed in this study. Findings Profile of Who Receives Early Termination A fuller profile of persons receiving early termination by their PCRA risk designation, along with statistical tests of significance and effect sizes, is provided in Figure 2. Regarding the relationship between risk and early termination outcomes, persons designated by the PCRA in the lowest risk category at their final assessment were about three times more likely to receive early termination (26 percent early termed) than people who were classified in the PCRA highest risk category (8 percent early termed). However, the relationship between risk and early termination was less substantial for people whom the PCRA placed in the low- and low/moderate-risk categories. For example, the percentage of supervisees with early termination in these risk categories was nearly the same, ranging from 26 percent for PCRA lows to 27 percent for PCRA low/moderates.

________________ Insert Figure 2 here ________________

A profile of the application of early termination across selected covariates, including statistical tests and effect size metrics, is provided in Table 2. Across the major offense categories, persons convicted of drug offenses were about two times more likely to receive this closure method (34 percent early termed) than persons convicted of other major offenses, including weapons/firearms (20 percent early termed), public order (19 percent early termed), or property offenses (16 percent early termed). Given the statute’s language favoring early terminations for persons convicted of misdemeanor offenses, it is interesting to note that supervisees with felony convictions were more likely to be early-termed (25 percent) than their misdemeanor counterparts (15 percent).

The use of early termination varies somewhat across the demographic categories of race/ethnicity, age, and gender. By race/ethnicity, there was a 4-percentage-point difference between the group most likely to receive early terminations (non-Hispanic Whites – 26 percent) and Hispanics, who were least likely to be early termed (22 percent); however, the data show little variation in the use of early terminations by non-Hispanic Whites (26 percent) and Blacks (25 percent). Moreover, there were essentially no differences in the use of early terminations of male (25 percent) and female (24 percent) supervisees. Last, about 25 percent of supervisees 50 and older received early terminations, compared to 19-20 percent of supervisees under the age of 30.

Unlike demographics, there exists a moderate relationship between early termination and supervision months imposed. For example, over a third of supervisees (36 percent) sentenced to a supervision term of four years or more and more than half (53 percent) of supervisees sentenced to a supervision term of more than five years were early termed from federal supervision. In comparison, about a fifth (22 percent) of persons placed on supervision for 2 – 3 years, constituting the largest federal supervision sentencing time frame, were early termed from federal supervision.

_______________ Insert Table 2 here ________________

Information on the application of early terminations across the federal judicial system is provided in Figure 3. Overall, Figure 3, which deidentifies the federal judicial districts, shows that there is substantial disparity regarding the use of early terminations at the district level, and these differences hold even when they are adjusted to account for factors driving the use of early termination, including PCRA risk scores, most serious conviction offenses, and supervision times imposed. The districts on the far right of the chart close over 50 percent of their cases by early termination; in comparison, the districts on the left side of the chart resolved less than 10 percent of their cases through early termination. The disparity in the application of early terminations across the federal judicial system is most likely the result of cultural differences and policy preferences about how this method of case closure should be applied at the local level.

________________ Insert Figure 3 here ________________

Supervision Months Saved Through Early Termination Early-term supervisees were sentenced to supervision terms that were on average about 12 months longer than those of supervisees who completed a full term of supervision (50 months vs. 38 months, respectively) (see Table 3). Although early-term supervisees had longer supervision sentences, on average they were on supervision 7 months fewer than persons with full terms (31 months to 38 months, respectively). Overall, half of supervisees with early terminations had their supervision sentences reduced by 16 months or more.

_______________ Insert Table 3 here _______________

Table 4 provides information on the average reduction in supervision sentences resulting from early termination and the relationship between these sentencing reductions and several covariates. ^[4] In general, the supervision time imposed and most serious conviction offense had the strongest relationship with early termination sentencing reductions. Specifically, persons convicted of sex and drug offenses who were early termed witnessed the largest reductions; their supervision sentences were reduced by 37 months and 22 months, respectively. Conversely, supervisees convicted of property or public order sentences saw their supervision terms shortened by less than 15 months. The average supervision sentences imposed at the time of sentencing also had a substantial bearing on early termination reductions. For example, supervisees sentenced to terms of 4 – 5 years witnessed drops in their supervision sentences of about two years (23 months). Moreover, persons sentenced to five years or more of supervision received reductions of about 4 years (47 months) when early termination was applied. Interestingly, there was a relatively weak correlation between risk as measured by the PCRA and early-term sentencing reductions, with the low-risk supervisees receiving slightly higher sentencing reductions (18 months) than their higher risk counterparts (15 months).

________________ Insert Table 4 here ________________

Early Termination and Community Safety The next part of this analysis explored whether the policy of early terminations represented a threat to community safety. Specifically, the focus here was on whether persons receiving early terminations recidivated at higher, similar, or lower rates compared to persons serving their full supervision terms. To reiterate, recidivism is defined by rearrest activity involving any or violent offenses that occurred within 24 months after the expiration of a person’s supervision term. Moreover, PSM matching techniques were employed to create “like” groups of persons receiving early and regular termination to examine their post-supervision recidivism rates (see Methods section and Appendix Table 1).

Differences in the post-supervision rearrest rates for matched persons with early and regular terminations are shown in Figure 4. Even though both groups have similar risk profiles through the matching process, supervisees with early termination were 2 percentage points less likely to recidivate (10 percent rearrested) than their regular termination counterparts (12 percent rearrested). These differential arrest rates widened from a 1-percentage-point difference for PCRA lows with early termination (4 percent) and regular termination (5 percent) to a 4- percentage-point difference for PCRA moderates with early termination (24 percent) and regular termination (28 percent). Among high-risk supervisees, the post-supervision arrest rates for regular and early terminations were essentially the same. While for the most part the early- and full-termed supervisees manifested rearrest rates that statistically differed at the .001 level, it is important to note that the effect size metrics (see Cramér’s V scores ) show the magnitude of these differences being relatively negligible.

In regard to recidivism for a violent offense, people whose supervision sentences were terminated early garnered violent arrest charges (2.9 percent) at rates similar to those receiving full supervision terms (3.2 percent) (see Figure 5). By risk level, there were no statistically significant differences in the violent rearrest rates for persons ending supervision with PCRA risk designations of low, low/moderate, or high. Only the persons in the moderate risk category manifested statistically different violent arrest rates, with the early-term group witnessing a violent arrest rate two percentage points lower (8 percent) than the regular-termed group (10 percent); however, the magnitude of these differences as shown by the Cramér’s V score indicates a relatively marginal effect size.

_______________ Insert Figure 4 here ________________ _______________ Insert Figure 5 here ________________

Information on the most common rearrest offenses for early- and regular-termed supervisees who recidivated was also examined (data not shown). The rearrest patterns for both groups are similar, with drugs accounting for the largest re-offense category, followed by violence, public-order, and property offenses. The majority of arrests for violent offenses involved assault-type crimes (85 percent) for both groups. Discussion The current analysis examined the application of early terminations in the federal supervision system. Specifically, it provided a profile of who was most likely to receive early terminations, explored district-level differences in the use of early terminations, assessed the time saving garnered by using early terminations, and analyzed whether early terminations endangered community safety. Overall, the research found that early terminations have been increasing, with about 1 out of 4 successful closures over the last 10 years occurring through early termination of the supervision sentences. Regarding who receives early terminations, the research showed that this closure type was moderately correlated with a supervisee’s risk profile and that other factors—particularly the number of supervision months imposed and (most important) the district of case closure—had the greatest bearing on whether a case would end by early termination. While persons who received early termination were sentenced to longer supervision terms than their regular-termed counterparts, on average they were on supervision for seven fewer months than persons who served their full supervision terms. In general, the greatest reduction in supervision sentences through the application of early termination was reserved for those convicted of drugs and sex offenses or for those sentenced to supervision sentences of four years or more. It is notable that sex offenders receiving early termination manifested the greatest sentencing reductions, because these persons tend to receive longer supervision terms compared to the average supervisee (USSC, 2012).

In findings mirroring research conducted by Baber and Johnson (2013) and work focusing on early termination at the state level (Courtney et al., 2022; Jacobson et al., 2017; Pew Charitable Trusts, 2016), this study found that early terminations did not endanger community safety. Specifically, when matched on a range of criteria associated with the risk of recidivism, supervisees with early terminations manifested post-supervision arrest rates that were two percentage points lower for any offenses than those of their regular-termed counterparts. Moreover, the post-supervision rearrest rates for violent offenses were relatively similar for the early- and regular-termed groups. While the current study produced findings showing a narrower band in the recidivism rates between the early- and regular-termed supervisees than that highlighted by Baber and Johnson (2013), these differences might be explained by the fact that the extant research uses more rigorous statistical approaches (e.g., propensity score matching) and matches on a greater range of variables than prior studies on federal early termination. Another recidivism finding was that the types of post-supervision arrest offenses between the two groups was essentially the same. Directions for Future Research Several directions for future research could be undertaken on the topic of early termination. Specifically, additional research could attempt to apply approaches that allow for more rigorous causal inferences to be drawn on the relationship between early termination and community safety. While the propensity-scoring models generated similar groups of regular- and early- termed individuals matched on several recidivism risk characteristics, it is possible that factors unaccounted for (such as drug addiction severity) could potentially explain the differences in the rearrest rates between the early- and regular-termed supervisees. If these unmeasured factors were included in the matching models, it is possible that the differences in rearrest outcomes between the early- and regular-termed individuals might change. Although methods that would allow for causal inferences, including random assignment of early termination, are probably infeasible, it might be possible to apply techniques such as interpreted time series designs to take advantage of those districts that decided to move from using early termination sparingly to making more extensive use of this closure method. Such efforts would allow for further exploration of whether early termination might be endangering the community.

Another potential avenue for future research could be delving into the differential use of early terminations at the district level. Methods that are more qualitatively oriented, including the use of focus groups or surveying judges, U.S. attorneys, or executive probation staff, on their attitudes or policies towards early termination would be required to better understand the disparities in the use of early termination across the judicial districts. The AO is seeking to collaborate with the Federal Judicial Center on conducting a more qualitative approach to better understand the use of early termination at the district level. It is anticipated that these efforts will illuminate district-level differences in early termination and perhaps suggest methods for bringing greater levels of uniformity to this technique of case closure. Conclusion Although the early termination research agenda could be further expanded, it is essential to highlight the compelling findings presented here. These results demonstrate that supervisees granted early termination under current policies pose no greater risk to the community than those who serve a full term of supervision (Baber & Johnson, 2013). These findings reaffirm the notion that early termination policies empower officers to make responsible decisions about which persons are suitable candidates for an early termination of their supervision term. References Baber, L. (2010). Results-based framework for post-conviction supervision recidivism

analysis. Federal Probation , 74 , 5. Baber, L. M., & Johnson, J. L. (2013). Early termination of supervision: No compromise to community safety. Federal Probation , 77(2) , 17-22. Cohen, T. H., Lowenkamp, C. T., & VanBenschoten, S. W. (2016). Does change in risk matter? Examining whether changes in offender risk characteristics influence recidivism outcomes. Criminology & Public Policy , 15 (2), 263-296.

Courtney, L., Pierce, B., Oglesby-Neal, A., & Nembhard, S. (2022). An assessment of probation sentencing reform in Louisiana and Georgia . The Urban Institute. Desmarais, S., & Singh, J. (2013). Risk assessment instruments validated and implemented in correctional settings in the United States . Council of State Governments Justice Center. Griffin, M. L., Hepburn, J. R., & Ginsburg, K. (2013). The effects of earned time credit on successful probation outcomes: An evaluation of Maricopa County, Arizona’s experience . Unpublished.

Guo, S., & Fraser, M. W. (2014). Propensity score analysis: Statistical methods and applications (Vol. 11). SAGE. Jacobson, M. P., Schiraldi, V., Daly, R., & Hotez, E. (2017). Less is more: How reducing probation populations can improve outcomes . Harvard Kennedy School, Program in Criminal Justice Policy and Management.

Johnson, J. L., Lowenkamp, C. T., VanBenschoten, S. W., & Robinson, C. R. (2011). The construction and validation of the federal Post Conviction Risk Assessment (PCRA). Federal Probation , 75(2) , 16.

Lowenkamp, C. T., Johnson, J. L., Holsinger, A. M., VanBenschoten, S. W., & Robinson, C. R. (2013). The federal Post Conviction Risk Assessment (PCRA): A construction and validation study. Psychological Services , 10 (1), 87.

Lowenkamp, C. T., Holsinger, A. M., & Cohen, T. H. (2015). PCRA revisited: Testing the validity of the Federal Post Conviction Risk Assessment (PCRA). Psychological Services , 12 (2), 149.

Luallen, J., Radakrishnan, S., & Rhodes, W. (2016). The predictive validity of the Post- Conviction Risk Assessment among federal offenders. Criminal Justice and Behavior , 43 (9), 1173-1189.

Minnesota Department of Corrections. (2017). Study of Earned Compliance Credit: Recommendations for Minnesota. Minnesota Department of Corrections. New York City Department of Probation. (2013). Do more good: A progress report from the NYC Department of Probation . New York City Department of Probation. Petersilia, J. (2007). Employ behavioral contracting for earned discharge parole. Criminology & Public Policy , 6 , 807-814. Pew Charitable Trusts. (2016). Missouri policy shortens probation and parole terms, protects public safety . The Pew Charitable Trusts. Pew Charitable Trusts. (2020). States can shorten probation and protect public safety . The Pew Charitable Trusts. Robina Institute of Criminal Law and Criminal Justice. (2020). Research in brief: Goal-based supervision . University of Minnesota. Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika , 70 (1), 41–55. Rosenbaum, P. R., & Rubin, D. B. (1985). Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. The American Statistician , 39 (1), 33–38.

Smith, S. M., Omori, M. K., Turner, S. F., & Jannetta, J. (2012). Assessing the earned discharge pilot project: The importance of context, capacity, and content. Criminology & Public Policy , 11 (2), 385-410.

U.S. Sentencing Commission (USSC) (2012). Report to Congress: Federal child pornography offenders . U.S. Sentencing Commission. Walters, G. D., & Lowenkamp, C. T. (2016). Predicting recidivism with the Psychological Inventory of Criminal Thinking Styles (PICTS) in community-supervised male and female federal offenders. Psychological Assessment , 28 (6), 652.

Statutes 18 U.S.C. § 3561 (1984) Sentence of probation 18 U.S.C. § 3564(c) (1984) Running of a term of probation – early termination 18 U.S.C. § 3583 (1984) Inclusion of a term of supervised release after imprisonment 18 U.S.C. § 3583(e)(1) (1984) Inclusion of a term of supervised release after imprisonment –

modifications of conditions Table 1: Descriptive statistics of federal supervisees in study sample Variables % (n) M (SD) Dependent variables Re-arrest any offense within 2 years of case closure 11.0 (32,641) Re-arrest violent offense within 2 years of case closure 3.1 (9,060) Independent variables - selected Early termination 24.7 (73,155) Race/ethnicity White, not Hispanic 36.7 (107,985) Black, not Hispanic 35.5 (104,575) Hispanic, any race 23.0 (67,798) Other race 4.8 (13,967) Male 78.8 (233,393) Age at supervision end date (in years) (296,023) 43.5 (12.1) Most serious conviction offense Drugs 44.6 (131,917) Property/White Collar 21.7 (64,050) Weapons/Firearms 13.3 (39,384) Public order 8.5 (25,137)

Immigration 4.5 (13,233) Violence 4.2 (12,354) Sex offense 2.8 (8,384) All other 0.4 (1,133) Misdemeanor or infraction conviction 5.9 (17,584) Final PCRA risk categories Low 43.4 (128,515) Low/Moderate 43.3 (128,199) Moderate 11.2 (33,155) High 2.1 (6,154) Drug use during supervision None 73.6 (217,948) Tested positive one substance 16.7 (49,474) Tested positive multiple substances 9.7 (28,601) Supervision time imposed (in months) (295,675) 40.8 (20.9) Note. Includes 296,023 federal supervises with cases closed successfully between fiscal years 2014 through 2023 with PCRA assessments. PCRA = Post Conviction Risk Assessment. Other race includes Asians, Pacific Islanders, Native Americans, Alaska Natives, and other non-identified races. Table 2: Profile of persons with successful closures who received an early termination by selected covariates Number of Percent early Cramer's χ ^[2] Selected covariates supervisees terminations V Most serious conviction offense Drugs 131,917 33.5 % 10000.0 * 0.18 Weapons/Firearms 39,384 20.2 Public order 25,137 19.2 Sex offense 8,384 17.8 Property/White Collar 64,050 16.4 Immigration 13,233 15.9 All other 1,133 15.4 Violence 12,354 15.3 Conviction severity Felony 278,405 25.3 % 910.0 * -0.06 Misdemeanor or infraction 17,584 15.2 Race/ethnicity White, non-Hispanic 107,985 26.3 % 538.7 * 0.04 Black, non-Hispanic 104,575 25.2 Other 13,967 23.8 Hispanic any race 67,798 21.5 Gender Male 233,393 24.9 % 26.3 * 0.01 Female 62,628 23.9 Age at supervision end date (in years) Under 20 143 18.9 % 601.3 * 0.05 20 - 29 34,694 20.0 30 - 39 88,153 24.2 40 - 49 85,866 26.6 50 and older 87,167 25.2 Supervision time imposed Less than 12 months 23,217 5.3 % 18000.0 * 0.25 12 months - 23 months 44,033 14.6 24 months - 35 months 124,079 21.7 36 months - 47 months 25,621 30.2 48 months - 59 months 65,028 35.7 60 months or more 13,697 52.9 Note. Includes 296,023 federal supervises with cases closed successfully between fiscal years 2014 through 2023 with PCRA assessments. Successful closures include those supervisees with regular or early terminations. PCRA = Post Conviction Risk Assessment. Other race includes Asians, Pacific Islanders, Native Americans, Alaska Natives, and other non-identified races. * p <.001 Table 3: Differences between imposed and actual supervision sentences for persons with early and regular terminations

Imposed term Served term (months) (months) Months saved Number Supervision term of closures Mean Median Mean Median Mean Median Early termination 72,845 49.8 47.9 30.8 27.8 19.0 15.6 Regular termination 222,830 37.8 35.9 37.8 35.9 0.0 0.0 Note. Numbers won't match those of table 1 because of missing data. Data on supervision time imposed and served available for over 99% of the total study sample. Table 4: Reduction in average number of supervision months for persons with early terminations by selected covariates Average supervision terms

Number with early Mean term Selected covariates terminations Imposed Served reduction PCRA 1.0 risk categories Low 32,921 47.0 29.5 17.5 Low/Moderate 34,653 52.8 32.2 20.6 Moderate 4,794 48.1 29.7 18.3 High 477 43.7 29.0 14.7 Significance F [3, 72841] = 210.6; p < .001; ω ^[2] = .009 Most serious conviction offense (not all offense types shown) Public order 4,813 33.4 21.1 12.3 Property/White Collar 10,514 39.2 25.3 13.9 Violence 1,870 42.6 27.9 14.8 Weapons/Firearms 7,918 41.8 26.8 15.0 Drugs 44,098 55.4 33.7 21.7 Sex offense 1,271 90.7 53.8 37.0 Significance F [7, 72755] = 828.8; p < .001; ω ^[2] = .074 Race/ethnicity Other race 3,306 43.4 27.5 15.9 Hispanic, any race 14,572 49.0 31.4 17.6 White, not Hispanic 28,162 47.0 29.0 18.0 Black, not Hispanic 26,355 54.1 32.8 21.4 Significance F [3, 72391] = 288.3; p < .001; ω ^[2] = .012 Gender Female 14,975 44.5 28.2 16.3 Male 57,870 51.2 31.4 19.7 Significance F [1, 72843] = 504.7; p < .001; ω ^[2] = .007 Supervision time imposed Less than 12 months 1,236 11.3 7.1 4.2 12 months - 23 months 6,446 23.0 15.5 7.5 24 months - 35 months 26,958 35.8 23.3 12.4 36 months - 47 months 7,747 47.5 30.5 17.0 48 months - 59 months 23,212 59.7 37.1 22.6 60 months or more 7,246 103.3 56.0 47.3 Significance F [5, 72839] = 10556.8; p < .001; ω ^[2] = .420 Note. Includes 72,845 federal supervises with cases closed through early terminations and with complete information on number of supervision months saved. Some categories may not sum to totals because of missing data. With exception of PCRA risk categories, data sorted from smallest to highest involving reductions in mean supervision terms. PCRA = Post Conviction Risk Assessment.

APPENDIX TABLE 1. Equivalent groups generated by propensity score matching - early terminations Panel A: Pre-matching group differences Panel B: Post-matching group differences Persons with Persons without Persons with Persons without early term (n = early term (n = early term (n early term (n = % Bias Matching covariates 59,026) 185,915) T-statistic p Value =45,251) 45,251) T-statistic p Value reduction Race Black 0.37 0.36 4.55 0.000 0.35 0.36 -0.98 0.327 71.3 White 0.39 0.36 12.50 0.000 0.39 0.39 0.14 0.892 98.5 Other 0.05 0.05 -2.26 0.024 0.05 0.05 -0.24 0.814 86.1 Hispanic, any race 0.20 0.23 -18.56 0.000 0.21 0.21 1.11 0.266 92.2 Sex (male = 1) 0.81 0.81 0.59 0.552 0.81 0.81 0.25 0.799 42.6 Average age (in years) 41.5 40.3 20.77 0.000 41.1 41.2 -0.73 0.463 95.3 Most serious conviction offense Drugs 0.63 0.41 88.22 0.000 0.57 0.58 -2.44 0.015 96.3 Weapons/Firearms 0.10 0.14 -21.86 0.000 0.12 0.12 1.49 0.137 91.4 Immigration 0.03 0.04 -19.06 0.000 0.03 0.03 0.45 0.652 97.3 Property/White Collar 0.14 0.23 -47.43 0.000 0.16 0.16 1.17 0.241 97.0 Sex offense 0.02 0.03 -18.00 0.000 0.01 0.01 0.17 0.868 99.1 Violence 0.02 0.05 -23.68 0.000 0.03 0.03 -0.18 0.856 99.2 Public order 0.06 0.09 -17.69 0.000 0.07 0.07 1.05 0.294 92.5 All other 0.00 0.00 -6.67 0.000 0.00 0.00 -0.77 0.444 86.9 Misdemeanor conviction 0.03 0.06 -25.91 0.000 0.04 0.04 1.06 0.291 95.4 Average final PCRA domain scores Total score 5.74 6.22 -31.31 0.000 5.69 5.70 -0.42 0.672 98.3 Criminal history 4.09 4.10 -1.43 0.153 4.00 4.02 -1.03 0.305 5.2 Employment/education 0.64 0.82 -41.22 0.000 0.66 0.66 0.43 0.667 98.7 Drugs/Alcohol 0.12 0.23 -45.03 0.000 0.13 0.13 0.49 0.627 98.8 Social networks 0.84 0.95 -31.42 0.000 0.85 0.85 0.88 0.380 96.6 Cognitions 0.04 0.12 -51.78 0.000 0.05 0.05 0.30 0.767 99.5

2:13-cr-20050-JEH # 105 Filed: 03/14/25 Page 54 of 59 EARLY TERMINATION 31 APPENDIX TABLE 1 (continued) PCRA risk levels Low 0.45 0.42 13.06 0.000 0.46 0.46 0.93 0.354 90.4 Low/Moderate 0.48 0.43 20.36 0.000 0.46 0.46 -1.10 0.271 92.7 Moderate 0.06 0.13 -40.79 0.000 0.07 0.07 0.26 0.793 99.3 High 0.01 0.02 -26.62 0.000 0.01 0.01 0.25 0.804 99.3 Changes PCRA risk levels first to final assessment Minus two or more 0.01 0.01 2.82 0.005 0.01 0.01 0.17 0.863 91.7 Minus one 0.21 0.17 17.20 0.000 0.20 0.20 -0.99 0.320 92.0 No change 0.75 0.74 4.62 0.000 0.76 0.75 1.35 0.178 61.8 Plus one 0.03 0.07 -34.93 0.000 0.03 0.03 -1.09 0.277 96.9 Plus two or more 0.00 0.00 -10.93 0.000 0.00 0.00 -0.30 0.765 97.9 General criminal thinking 90.0 91.7 -16.96 0.000 90.0 90.1 -0.99 0.320 92.5 Drug use during supervision None 0.82 0.71 54.93 0.000 0.81 0.81 0.15 0.879 99.7 Tested positive one substance 0.13 0.18 -30.33 0.000 0.14 0.13 0.31 0.756 98.7 Tested positive multiple

substances 0.05 0.11 -43.32 0.000 0.05 0.06 -0.73 0.464 98.3 Average number positive drug tests during supervision 0.74 1.40 -30.46 0.000 0.81 0.81 -0.04 0.970 99.9 Average supervision time imposed (in months) 49.6 37.8 120.22 0.000 44.5 44.3 1.48 0.140 98.6 Average number of conditions 6.7 7.5 -35.24 0.000 6.7 6.7 -0.16 0.876 99.5 Note: Nearest neighborhood matching with Caliper 0.0001 was used. A total of 13,775 persons with early terminations were lost through an inability to match with control group. In addition to matching on the above covariates, persons were matched on the fiscal year of case closure (not shown in table). Propensity score matching applied to a subset of supervisees (n = 244,941) whose rearrest activity could be tracked for a minimum of two years after case closure.

2:13-cr-20050-JEH # 105 Filed: 03/14/25 Page 55 of 59 EARLY TERMINATION 32 % early Figure 1. Percentage of successful closures ended by early termination, fiscal years 2000 - 2023 terminations 30% 28% 27% 28% 25% 25% 25% 24% 23% 23% 22% 22% 21% 21% 21% 21% 21% 20% 20% 19% 20% 19% 18% 17% 15% 13% 12% 12% 10% 5% 0%

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 Fiscal year of closure Note: Includes persons whose cases were closed through a successful termination (early or regular). Excludes cases closed through revocation, transfer, or other means. Early terminations based on compassionate grounds not included in figure; these cases accounted for less than 1% of all terminations.

2:13-cr-20050-JEH # 105 Filed: 03/14/25 Page 56 of 59 EARLY TERMINATION 33 Percent of supervisees granted early termination

Figure 2. Percentage of supervisees who received early termination by the final PCRA risk groups 30% 27.1% 25.8% 25% 20%

14.5% 15% 10%

7.8% 5% 0%

Low risk Low/Moderate risk Moderate risk High risk PCRA risk categories χ2 = 3300.0, p < .000; Cramer's V = 0.11 Note. PCRA = Post Conviction Risk Assessment. 2:13-cr-20050-JEH # 105 Filed: 03/14/25 Page 57 of 59 EARLY TERMINATION 34 Percent closed through early

Figure 3. Percent of successful closures ended through early termination by judicial district (de- termination identified) 60% 50% 40% 30% 20% 10% 0%

Federal judicial districts: de-identified Note. Percentages adjusted to take into account post-conviction risk assessment scores, most serious conviction offenses, and supervision time imposed. Federal judicial districts de-identified in figure.

Figure 4. Percent of early and regular term supervisees re-arrested for any offense within 24 months of their supervision end dates by final risk categorization

Percent Early termination Regular termination arrested 50% 38.8% 40% 37.7% 30% 27.9% 23.5% 20% 15.4% 13.2% 11.6% 9.9% 10% 5.1% 4.3% 0% All supervisees χ2 = 71.4 Low χ2 = 14.2 (p < .001) Low/Moderate χ2 = 44.1 Moderate χ2 = 15.9 (p < High χ2 = 0.1 (p= n.s.) V = (p < .001) V = -.03 V = -.02 (p < .001) V = -.03 .001) V = -.05 -.02 Final PCRA risk levels Note. Early and regular termed supervisees matched on a variety of criteria - see appendix table. PCRA = Post Conviction Risk Assessment

2:13-cr-20050-JEH # 105 Filed: 03/14/25 Page 59 of 59 EARLY TERMINATION 36 Figure 5. Percent of early and regular term supervisees re-arrested for violent offenses within 24 months Percent of their supervision end dates by final risk categorization violent arrests

Early termination Regular termination 20% 15%

13.1% 12.5% 9.9% 10% 7.5% 5% 4.3% 4.0% 3.2% 2.9% 1.0% 1.0% 0% All supervisees χ2 = 8.8 Low χ2 = 0.50 (p = n.s.) Low/Moderate χ2 = 2.3 Moderate χ2 = 10.7 (p < High χ2 = 0.05 (p = n.s.) (p = n.s.) V = -.01 V = .00 (p = n.s.) V = -.01 .001) V = -.04 V = -.01 Final PCRA risk levels Note. Early and regular termed supervisees matched on a variety of criteria - see appendix table. PCRA = Post Conviction Risk Assessment

NOTES

[1] This Court granted the motion at issue orally from the bench after a hearing on March 5, 2025, stating that a written Order and Opinion would follow. This is that Order and Opinion. 1

[2] Citations to the electronic docket are abbreviated as “D. ___ at ECF p. ___.” 6

[3] Sub-section 3553(a)(3) is excluded from the factors a court must consider as set forth in § 3583(e), and subsections (a)(4) and (a)(5) only have applicability to imposition and revocation of a supervised release term, not an early termination. Subsection (a)(7) relates to restitution, which is inapplicable in this case. 18

[4] Volume 8E, Chapter. 3, § 360.20 of the Guide contains the following policy for probation offices: (a) Under 18 U.S.C. §§ 3564(c) and 3583(e)(1), the court may terminate: (1) terms of probation in misdemeanor cases at any time; and (2) terms of supervised release or probation in felony cases after the expiration of one year of supervision, if the court is satisfied that such action is warranted by the conduct of the person under supervision and is in the interest of justice. (Note: Early termination of parole cases is governed by the U.S. Parole Commission Rules and Procedures Manual, Section 2.43.) (b) During the first 18 months of supervision, the appropriateness of early termination must be based on the person's overall progress in meeting supervision objectives, to include having: (1) substantially satisfied the requirements of the court order; and (2) demonstrated a willingness and capability to remain lawful beyond the period of supervision. Note: Officers should not recommend persons for early termination who have an identified higher risk to community safety. (c) At 18 months, there is a presumption in favor of recommending early termination for persons who meet the following criteria: (1) The person does not meet the criteria of a career drug offender or career criminal (as described in 28 U.S.C. § 994(h)) or has not committed a sex offense or engaged in terrorism; (2) The person presents no identified risk of harm to the public or victims; (3) The person is free from any court-reported violations over a 12-month period; (4) The person demonstrates the ability to lawfully self-manage beyond the period of supervision; (5) The person is in substantial compliance with all conditions of supervision; and (6) The person engages in appropriate prosocial activities and receives sufficient prosocial support to remain lawful well beyond the period of supervision. (d) After 18 months, higher risk persons under supervision who have demonstrated a reduction in risk (as demonstrated by a reduction in PCRA level/category) and who are in substantial compliance with the factors provided above must be considered for early termination. (e) The existence of an outstanding financial penalty does not adversely affect early termination eligibility, as long as the person under supervision is in compliance with the payment plan for the prior 12 months. (f) Officers should consider early termination for all persons who have been supervised for 12 months under low-risk supervision standards and who otherwise meet the eligibility criteria. At that time, the supervisor should approve, in a chronological record entry, any decision not to petition the court for early termination. The supervisor should then set the timeframe for the next early termination review. 20

[5] Judicial Conference policy creates a presumption in favor of probation officers recommending early termination of supervised release at eighteen months for certain persons who, among other things are in “substantial compliance with all conditions of supervision.” Guide , at Vol. 8E, Ch. 3, § 360.20. “Substantial compliance” is not “perfect” or “exceptional” compliance. 22

[1] The author would like to thank Aaron Wonneman, Christopher Lowenkamp, James Johnson, Nina Chipman, Scott VanBenschoten, and Vanessa Starr for their helpful suggestions and comments. Special thanks to Ellen Fielding for her careful editorial review. Direct correspondence to Thomas H. Cohen, Administrative Office of the U.S. Courts – Probation and Pretrial Services Office, One Columbus Circle, NE, Washington, DC 20544 (e-mail: thomas_cohen@ao.uscourts.gov).

[2] Supervised release refers to persons sentenced to a term of community supervision following a period of imprisonment within the Federal Bureau of Prisons (18 U.S.C. §3583). Probation refers to a period of supervision without any imposed incarceration sentence (18 U.S.C. §3561).

[3] About 85 percent of the unsuccessful closures involved a revocation from federal supervision.

[4] Table 4 also includes ANOVA statistical tests and effect size metrics.

Case Details

Case Name: United States v. Crisp

Court Name: District Court, C.D. Illinois

Date Published: Mar 14, 2025

Docket Number: 2:13-cr-20050

Court Abbreviation: C.D. Ill.