FORUM on Corrections Research

Classifying offenders serving life sentences

by Fred P. Luciani¹
Research Branch, Correctional Service of Canada

Initial security and reclassification of lifers present a number of unique challenges for correctional staff. While the gravity of the offence is common to all offenders serving life sentences, the security risk profile of lifers diverge in many important ways. The research evidence supports continuing the application of the Custody Rating Scale and Security Reclassification Scale to lifers, relying on the professional discretion of staff, considering additional guidelines and advising caution.

Objective measures of risk assessment in the management of offenders have yielded a number of tangible benefits for corrections. These include improvements in the consistency and accuracy of security placement, programming and discretionary release decisions, the formulation of publicly explicit and defensible norms for decision-making, and provision for regular tracking, verification and refinement of assessment measures.

Many actuarial assessment instruments rely on statistically weighted items and standardized decision rules that provide relative ratings of individual risk derived from the aggregate risk profile of the subject population. Objective instruments are now routinely applied by the Correctional Service of Canada (CSC) to support initial and subsequent reclassification of offenders, to assess offender risk and needs, reintegration potential, education aptitude, psychological functioning, program responsivity and in many other areas as determined by individual circumstances. Substantial evidence continues to accumulate to support the effectiveness of objective measures in both clinical² and correctional application³ and when combined with seasoned judgement a balanced decision-making modal can be expected. Despite the growing confidence in these measures, field testing⁴ and operational reviews⁵ of security classification scales, the ongoing analysis of scale overrides have raised concerns.

Research found that scales applied under conditions for which they are not designed and factors unrelated to risk can result in assessments contrary to ratings prescribed by the scale. Field responses also indicated that scale overrides are more likely to occur when risk factors with high face validity are not included in the instrument (often because of lack of statistical support) or item weightings are not in accord with the perceptions of the users. Finally, dealing with outliers or unique sub-groups within the subject population can present problems. Most clinical or correctional populations contain unique groups that may not fully share or contain characteristics distinct from the aggregate profile. Unless assessment instruments are specifically validated and recalibrated, caution is warranted and dedicated instructions for dealing with the groups are recommended. In the development and operational review of CSC’s classification scales many of these issues emerged in applying the scales to the “lifer” group and form the topic of the following discussion.

Custodial classification

For federal corrections, initial penitentiary placement of offenders begins with the Custody Rating Scale⁶ (CRS), a 12 item scale based on static, historical indices of institutional adjustment and public safety risk. The CRS assigns a security level rating and is applied at admission to all new offenders and those readmitted by way of conditional release revocation. Future security reviews for reclassification purposes take place at any time following the transfer of the offender to the initial placement site but no later then the first anniversary of the transfer and annually thereafter. Reclassification is supported by the Security Reclassification Scale⁷ (SRS), a 13 item scale that with the exception of History of Involvement Institutional Incident drawn from the CRS, all items are based on recent in-custody behaviour or progress. (The SRS was recently revised to include two additional items and adjustments to the item scores, cut-off values and override procedures.)

Life sentences

An indeterminate or life sentence can result from a number of dispositions including convictions for first- and second-degree murder, a dangerous offender designation including offenders designated under former dangerous sex offender legislation, and offenders convicted of a serious offence where life is imposed as a maximum sentence. The group is generally referred to as the “lifer” population.

A recent survey of the CSC offender population found that 2,732 or approximately 20% of the total population of offenders were currently serving an indeterminate sentence. The longest serving lifer was admitted in December 1955. Offenders serving time for first- or second-degree murder constitute the majority (85%) of lifers followed by offenders designated as dangerous offenders (9%) and offenders serving life as a maximum (4%). Lifers were far more likely to be admitted by warrant of committal (83%) and were most often incarcerated in medium-security (59%) facilities. Reintegration potential ratings were available on 2,649 lifers, most of whom received either low (45%) or medium (42%) reintegration potential ratings with the remainder rated high reintegration potential. Approximately 30% of the current lifer population were past parole eligibility and another 13% will reach eligibility within two years.

Classifying offenders serving life sentences

While lifers undergo similar assessment procedures as offenders serving determinate sentences, they often present unique classification challenges. The gravity of the offence is common to all offenders serving life sentences while the security risk profile of lifers diverge in many important ways. They range from your stereotypical, premeditated contract killer, hardened by a life of crime and incarceration, to the lifer whose sole offence and incarceration resulted from a domestic dispute, an impulsive act with no obvious antecedents and for which the lifer is often remorseful.

Static risk instruments assume that the best predictor of future criminal or institutional behaviour is past criminal or institutional behaviour. In the case of many lifers, their past bears little relationship to the offence that landed them in federal custody and often they have little or no incarceration history. Offence severity and sentence length are heavily weighted against the lifer by the CRS, however, these are offset by the very favourable scores on items that assess criminal or incarceration history, age and street stability and often result in a contradictory security risk profile. While incarcerated, lifers tend to be the most well-adjusted, co-operative offenders who maintain and even strengthen their community resources while making the most of programming opportunities to address criminogenic needs, improve their educational and employment skills. The Security Reclassification Scale (SRS) is dominated by in-custody, dynamic factors assessed during the current year of incarceration and thus provides ample opportunity for many lifers to quickly improve reclassification ratings irrespective of offence severity, sentence length and remote release potential.

Lifers form a unique group and many share characteristics distinct from those of the aggregate profile on which the scales are based. The following analyses are based on regular tracking of the classification scales and placements decisions for lifers and non-lifers. An examination of the operational data gathered serves to illustrate many of the issues encountered in applying the security classification scales to the lifer population.

Initial security classification

In fiscal year 1999/2000, 3,985 offenders were admitted to federal custody for which both a CRS rating and a final Offender Security Level (OSL) decision was available on the Offender Management System. (OSL represents the final classification decision rendered by the Parole Officer at time of admission). The sample consisted of 133 (3%) offenders identified as lifers.

Table 1 compares the concordance, override and distribution results of the CRS ratings and OSL decisions for lifers and non-lifers admitted for the year. The results for lifers are presented in the top left of each cell and for non-lifers in the bottom right. CRS/OSL concordance is presented on the shaded diagonal, override to higher security in the cells below the diagonal and to lower security in the cells above the diagonal. The distribution by CRS ratings across security levels is found on the row marginal and for OSL decisions on the column marginal. At admission, the CRS assigned more lifers maximum-security ratings (38%) than did the OSL decisions (20%), while there were no real differences in the distribution by CRS ratings (0%) or OSL decisions (1%) to minimum-security. (Note: The CRS is purposefully weighted to prevent the direct assignment of a minimum-security rating to a lifer at initial placement). OSL decisions leaned heavily toward medium decisions (79%), exceeding CRS scale ratings (62%).

Table 1

CRS/ OSL Concordance Lifers and Non- lifers: Admission Population
OSL Decision
LIFER NON-LIFER	MAXIMUM	MEDIUM	MINIMUM	CRS DISTRIBUTION
MAXIMUM	17% 4%	21% 4%	0% 0%	38% 8%
MEDIUM	4% 1%	58% 47%	1% 6%	62% 54%
MINIMUM	0% 0%	0% 9%	0% 30%	0% 39%
OSL DISTRIBUTION	20% 5%	79% 60%	1% 35%	(n=133) (n=3,852)
Note: Rows and columns may not tally accurately due to the effect of rounding.

The CRS/OSL concordance rate for lifers at admission was lower (74%) compared to the rate (80%) for non-lifers. As expected, far fewer maximum-security ratings (8%) and decisions (5%) were assigned to non-lifers than lifers and substantially more non-lifers (35%) received minimum-security decisions. Override of lifers were predominately to lower security (21%) consisting almost exclusively of offenders rated maximum security by the scale who received medium security OSL decisions. In contrast, overrides of non-lifers were evenly split to higher (10%) and lower (10%) security.

An operational review of the SRS⁸ was recently completed involving 6,993 applications of reclassification scale administered from the first of January to mid-November, 1999. SRS results and OSL decisions were available on 1,015 lifers, representing about 37% of the total lifer population. Table 2 compares the concordance, override and distribution results of the SRS ratings and OSL decisions for the resident lifer and non-lifer population. The results of the application of the SRS and security reclassification decisions of the lifer population provided some interesting contrasts to the CRS and initial decision results.

Table 2

SRS/ OSL Concordance for Lifers and Non- Lifers: Resident Population
OSL Decision
LIFER NON-LIFER	MAXIMUM	MEDIUM	MINIMUM	CRS DISTRIBUTION
MAXIMUM	5% 11%	0% 1%	0% 0%	5% 12%
MEDIUM	6% 6%	58% 59%	1% 3%	65% 68%
MINIMUM	0% 0%	12% 5%	17% 16%	29% 21%
OSL DISTRIBUTION	11% 17%	70% 65%	18% 19%	(n=1,015) (n=5,971)
Note:Rows and Columns may not tally accurately due to the effect of rounding.

The SRS rated fewer lifers maximum-security (5%) and more minimum-security (29%) than did OSL decisions (11% and 18% respectively). OSL decisions again favoured medium decisions (70%) over scale ratings (65%).

The SRS/OSL concordance rate (see diagonal of Table 2) for lifers at reclassification was lower (80%) than the rate (86%) for non-lifers. Non-lifers were more frequently rated maximum-security (12%) and fewer were rated minimum-security (21%) than the lifer sample. In contrast to admission overrides, lifer overrides at reclassification were predominately to higher security (19%) consisting largely of offenders rated minimum-security who received OSL decisions of medium-security (12%) and medium-rated who received maximum OSL decisions (6%). Less than 2% of lifer overrides was to lower security. Non-lifer overrides to higher security (11%) also exceeded those to lower security (4%).

In contrast to CRS ratings of lifers, the SRS was more liberal relative to SRS ratings of non-lifers and OSL decisions. Lifers were more frequently assigned to lower security levels by the SRS than non-lifers but more likely to be overridden by OSL decisions to higher security. The higher proportion of lifers assigned lower security ratings reflects the positive adjustment and ability of lifers to utilize incarceration opportunities to lower their risk profile. The large number of lifers assigned medium-security OSL decisions at admission in combination with lifers ability to earn minimum SRS ratings (often within a few years of arriving at a medium facility) becomes problematic.

The decision to assign a lifer to minimum-security must balance positive institutional adjustment against severity of the offence and length of sentence, and address the question of expanding resources designed to prepare offenders for release on those with no immediate release potential. In part, these issues explain the large number of minimum-rated lifers overridden to medium-security at the reclassification review.

Conclusions

While objective security classification measures offer many benefits, they are not intended to simply replace but rather to inform professional judgement, a maxim especially true in dealing with the security classification needs of offenders serving life sentences. For offenders serving life sentences overrides of both the CRS and SRS were more frequent and concordance rates lower and there was greater deviation from the design targets in the distribution across security levels. In addition, the classification override patterns of lifers suggest that the CRS overestimates security risk at admission and the SRS underestimates security risk at reclassification.

Security classification practice embraces the principle of “least restrictive method of confinement” and this is evident at admission where efforts are made to limit the placement of lifers to maximum-security. Consequently, many lifers begin their incarceration at medium-security where they adjust well, earn lower SRS scores and become candidates for minimum-security. The influence of sentence gravity, length and no immediate release potential, however, often overtake good adjustment resulting in frequent overrides to medium-security. Nonetheless, lifers earn high or medium reintegration potential ratings and many do migrate to lower security facilities where they continue their successful adjustment.

Considerable attention in this article and in earlier reports has been paid to CRS and SRS/OSL concordance rates and while they provide an indication of the concurrent validity of the instrument, the pursuit of higher concordance would do little to raise confidence in the scales. Correctional policy is to encourage staff to exercise their professional discretion and provides procedures to do so. It may be possible to recalibrate the CRS and SRS scales to improve concordance, however, before doing so it would be more helpful to gather additional information about how well the scales predict institutional adjustment, escape and public risk.

It may also be possible to design classification scales specifically for offenders serving life sentences, however, the net gains are not obvious. The evidence supports continuing the application of the scales to lifers and the prudent option would appear to be to continue current practice, rely on the discretion of staff, consider additional guidelines and advise caution.

---

1. Regional Headquarters, P.O. Box 1174, Kingston, Ontario K7L 4Y8.

2. Grove, W.M. and P.E. Meehl (1996). “Comparative Efficiency of Informal (Subjective, Impressionistic) and Formal (Mechanical, Algorithmic) Prediction Procedures: The Clinical-Statistical Controversy.”Psychology, Public Policy and Law, vol. 2, p. 293-323. See also Quinsey, V.L., G.T. Harris, Rice and C.A. Cormier (1998). Violent Offender: Appraising and Managing Risk. Washington DC: American Physiological Association.

3. Andrews, D.A., J. Bonta and R. Hoge (1990). “Classification for effective rehabilitation: Rediscovering psychology.” Criminal Justice and Behavior, 17, p. 19-52. See also Austin, J. (1983). “Assessing the new generation of prison classification models.” Crime and Delinquency, Vol. 29, no. 1, p. 561-577. And see Luciani, F.P., L.L. Motiuk and M. Nafekh (1996). An Operational Review of the Custody Rating Scale: Reliability, Validity and Practical Utility. Ottawa, ON: Correctional Service of Canada.

4. Luciani, F.P. and L.L. Motiuk (In press). The Security Classification of Offenders: Development of Security Reclassification Protocol. Ottawa, ON: Correctional Service of Canada.

5. Luciani, F.P. (In Press). The Security Classification of Offenders: Operational Review of the Security Reclassification Protocol.
Ottawa, ON: Correctional Service of Canada.

6. Solicitor General of Canada (1987). Development of a Security Classification Model for Canadian Federal Offenders: A Report to the Offender Management Division. Ottawa, ON: Correctional Service of Canada.

7. Luciani, F.P. (In Press). The Security Classification of Offenders: National Field Test Results of the Security Reclassification Protocol.Ottawa, ON: Correctional Service of Canada.

8. Luciani and Motiuk. (In press).