Correctional Service Canada
Symbol of the Government of Canada

Common menu bar links

Research Reports

Warning This Web page has been archived on the Web.

Women Offenders’ Institutional Behaviour: The Impact of Correctional Environment

2008 No R-190

Renée Gobeil,
Kelley Blanchette,
Meredith Robeson Barrett

Correctional Service Canada

September, 2007



Appreciation is extended to Kelly Taylor for her role in the SRSW field-test, from which a portion of the data used in this study were obtained.


Research has demonstrated that involvement in institutional misconducts is higher for women offenders at higher security classifications and placements than for their counterparts at lower security levels (Blanchette, Verbrugge, & Wichmann, 2002; Collie & Polaschek, 2003). Given the high rate of concordance between security classification and actual placement, it is not clear whether these differences are due to individual-level variables, the correctional environment, or some combination of the two. Some researchers have expressed concern that the correctional environment has a criminogenic influence on institutional behaviour (Harer & Langan, 2001). Though previous studies have found that this is not the case for male offenders (Baird, 1993; Camp & Gaes, 2005; Hanson, Moss, Hosford, & Johnson, 1983; Luciani, Motiuk, & Nafekh, 1996), no equivalent research has been conducted with women. This study sought to examine the possible effects of the correctional environment on women’s institutional behaviour.

The security reclassification procedure implemented for women offenders in 2005 includes an actuarial component (the Security Reclassification Scale for Women; SRSW). As the scale cannot encompass all possible case-specific factors, however, actual placement is not always consistent with the SRSW recommendation. As such, analyses for the current study involved contrasting the rates of involvement in institutional misconduct of women classified to the same security level by the SRSW but actually placed at different levels. We also compared the involvement of women actually placed at the same level, but classified to different levels by the scale. Data were gathered previously as part of the field test (Blanchette & Taylor, 2005) and revalidation (Gobeil, in press) of the SRSW, and involved a total of 580 and 384 reclassification reviews respectively.

Results demonstrated that actual placement was not associated with institutional misconduct after accounting for SRSW classification. Conversely, SRSW classification continued to be associated with institutional behaviour after accounting for actual placements. Together, these results demonstrate that it is individual-level variables which influence behaviour, and notthe correctional environment. These findings are supportive of current classification and placement practices.






To facilitate the management of offenders, Canada’s federal correctional system utilizes security level classifications. Offenders are placed at minimum, medium, or maximum security according to institutional adjustment, escape risk, and (should the offender escape), risk to the public for a new offence, as legislated by the Corrections and Conditional Release Act (CCRA, 1992). Research has demonstrated that, among women offenders, involvement in institutional misconducts increases with security classification and placement (Blanchette, Verbrugge, & Wichmann, 2002; Collie & Polaschek, 2003). Not surprisingly, women placed at minimum security engage in less misconduct than do those placed at medium security, who in turn engage in less misconduct than do those placed at maximum security.

Some have argued, however, that classification systems ultimately influence institutional behaviour through the streaming and placement of offenders (e.g., CAEFS, 1998; Harer & Langan, 2001). Specifically, there is concern regarding the potentially criminogenic influence of the prison environment. If prisons are criminogenic, then misconduct within an institution is a product of the environment, not of the offender’s individual characteristics; such a possibility runs counter to the rehabilitative goals of the correctional system. Within this context, this report presents the results of an investigation of the potential impact of the correctional environment on the institutional behaviour of women offenders. Moreover, the current study fulfills a commitment made in response to the Canadian Human Rights Commission’s 2003 report on the human rights of federally sentenced women (CSC, 2005).

Security Classification and Correctional Environments

To date, there is limited literature examining the criminogenic effect of correctional environments on institutional behaviour, especially in the field of women’s corrections. Given this gap, studies involving samples comprised exclusively or primarily of men are reviewed here, but these empirical results should be interpreted with caution as there is research suggesting significant differences in institutional adjustment patterns for women and their male counterparts (Warren, Hurt, Loper, & Chauhan, 2004).

An early study conducted by Hanson, Moss, Hosford, and Johnson (1983) sought to determine the ability of four classification methods to predict inmate penitentiary adjustment. The relevant classification methods for the purpose of the current literature review are the Security Level Designation and the Custody Level Classification. The Security Level Designation system assigns inmates to the appropriate institution based on variables such as severity of current offence, approximate length of incarceration, and history of escape or attempted escapes. The second classification system, the Custody Level Classification, assesses the required level of staff supervision within the assigned institution based on dynamic variables such as involvement with drugs or alcohol, type of most serious disciplinary report, and family and community ties. Hanson and colleagues collected data on these designations, as well as on institutional behaviour during a six-month follow-up, for 337 inmates in the U.S. Penitentiary at Lompoc. Using canonical correlation, the researchers found that Custody Level Classification best predicted inmate penitentiary adjustment (Rc = .90), defined in terms of disciplinary reports, time in segregation, and work performance rating, while Security Level Designation had little predictive power (Rc = -.08). The researchers speculated that the individual variables measured in the Custody Level Classification are most predictive of penitentiary adjustment as the variables measured reflect inter-individual differences in behaviour. In essence, these results demonstrate that it is the personal classification level, and not the environment of the prison, that influences prison behaviour.

A study of Tennessee’s inmate classification system in 1984 (Baird, 1993) allowed for an indirect examination of the criminogenic effects of prison environment. The study, conducted jointly by the Tennessee Department of Corrections and the National Council on Crime and Delinquency, found that despite the comprehensive inmate classification system, most offenders were still placed at the same level. Of 384 inmates placed at medium security, many were actually classified to other security levels. Specifically, 84 inmates were rated as ‘close’ custody (a classification level that is higher than medium but lower than maximum), 112 were rated as medium and 188 were rated as minimum. Despite receiving the same placement, inmates rated as close custody had significantly higher reported rates of prison misconduct than those rated as medium or lower.

In 1991, the Tennessee Department of Corrections reviewed its classification system again (Baird, 1993), this time focusing on minimum security inmates (classified as either minimum custody or the slightly higher minimum restricted custody) housed at medium security institutions. In this review, 18.9% of minimum security inmates and 22.1% of minimum restricted inmates committed infractions, compared to 58.0% of medium security inmates. The results also indicated that only 2.2% of minimum security inmates and 3.7% of minimum restricted inmates engaged in violent misconduct, compared to 15.8% of medium security inmates. Despite the higher security placement, minimum security inmates had significantly fewer infractions relative to medium security inmates. These studies were consistent in finding that inmates rated higher security engaged in more misconduct than those rated lower security, despite being placed in the same environment.

More recently, Luciani, Motiuk and Nafekh (1996) examined the reliability, validity and practical utility of the Correctional Service of Canada’s Custody Rating Scale, using inmate behaviour as the dependent variable. As part of this study, the researchers investigated whether incident rates were more reflective of individual-level variables or of the institutional environment. Specifically, they examined the incident rates of inmates according to both their classification on the Custody Rating Scale and to their actual security placement. If there was an effect of the institutional environment on behaviour, the researchers expected to see little to no differences in incident rates for inmates placed at the same security level, despite differing Custody Rating Scale ratings. However, among 6,745 inmates, those with medium security ratings had higher incident rates than did those with minimum security ratings, while those with maximum security ratings had higher incident rates than those with either minimum and medium security ratings, regardless of placement level. This positive linear relationship does not support the contention that the correctional environment is criminogenic; instead, it supports the notion that it is individual-level variables, as reflected by the security classifications, which affect the incident rates in prisons. The applicability of the study to the case of women offenders is uncertain however, due to the small number of women (n = 65) included in the study. As such, the incident rates and behaviour of women in different security placements is lost in the aggregate data.

Finally, Camp and Gaes (2005) examined a subset of data collected by Berk, Ladd, Graziano, and Baek (2003) to determine if there was a criminogenic effect of the prison environment on the behaviour of inmates. The researchers theorized that prison behaviour is influenced by a number of factors, including criminal behaviour (the individual proclivity of an inmate to commit criminal acts), inmate culture (the informal structure of the prison) and the formal organization of the prison. By holding security classification constant, the effect of inmate culture and prison regime can more accurately be determined. The sample was composed of 561 male inmates classified to Level III (one step down from the highest level of security). As part of an assessment of a new classification system, these inmates were split in half and assigned randomly to either a Level I prison (equivalent to minimum security), or a Level III prison. Once placed, inmates’ behaviour was recorded for two years. In analyzing the rates of misconduct as a function of placement, the researchers found no significant differences. Within the Level I prisons, 64% of the sample engaged in misconduct compared to 60% in the Level III prisons. The authors concluded that the environment of Level I prisons (hypothesized to be less violent and criminogenic) did not ultimately lower misconduct rates, nor did the environment of Level III prisons (thought to be more violent) encourage prison misconduct. Follow-up analyses focusing only on the more serious, violent behaviour again revealed no significant differences, with 33% of those placed at Level I and 36% of those placed at Level III engaging in serious misconduct. These results also suggest that it is the individual propensity to engage in misconduct that determines the behaviour of an inmate, and not the correctional environment.

Current Study

Given that the studies reviewed above all focused exclusively or primarily on male offenders, it is unclear whether this pattern of findings generalizes to women. As such, the present study comprises an investigation of the impact of correctional environment on women offenders’ institutional behaviour. The recent development and implementation of the Security Reclassification Scale for Women (SRSW; Blanchette & Taylor, 2005) provided an excellent opportunity to examine this issue.

According to Canadian legislation, security classifications are reviewed at a minimum of once a year for most offenders (CCRA, 1992). For women offenders, the SRSW acts as a key component in this reclassification process. It is an objective, gender-informed reclassification instrument developed specifically for federally sentenced women, implemented nationally in June 2005. In order to assess change since admission or the preceding security review, the scale focuses predominantly on dynamic factors, including escape risk, risk to the public if an escape were to occur, institutional adjustment and behaviour, and changes in the offender’s behaviour. Using actuarial methods, the scale provides a recommended security classification of minimum, medium, or maximum for each woman. Notably, a 10% discretionary range is included at the threshold between each security classification, allowing flexibility in assigning classifications to those women whose scores fall near the threshold values. These recommendations act as anchors for the reclassification decision, but caseworkers assigned to each woman also complete clinical appraisals of the woman’s risk, allowing consideration of case-specific factors or exceptional circumstances which are not contained within the scale’s component items. When the scale’s security classification and the results of the clinical appraisal differ, caseworkers need not assign the scale’s recommended security classification; however, caseworkers must document the rationale for their decision.

Differences between security classification recommendations produced by the SRSW and actual security placements allowed for the investigation of each’s association with institutional behaviour. Results indicating that involvement in misconducts was consistent with actual security placements rather than scale classifications would suggest that it is the correctional environment associated with each security level that influences institutional behaviour. Conversely, indications that involvement in misconducts was consistent with SRSW security classifications rather than placements would demonstrate that behaviour is related to factors associated with the women themselves, rather than the correctional environment. Finally, if both were found to be independently related to behaviour, a combined effect of correctional environment and individual-level characteristics would be supported.



Two samples of data were used in this study. The first corresponded to the sample from the SRSW field test (Blanchette & Taylor, 2005), and included a total of 580 security reclassification reviews completed between 2000 and 2003. Since some women underwent two or more security reviews during this time period, this cohort included a total of 323 individual women. The second data sample corresponded to the 443 reviews completed between 2005 and 2007 used to revalidate the scale (Gobeil, in press). Of these 443 reviews, actual classification data were not available for 59 reviews, leaving 384 reclassification events (268 individual women) for inclusion in this study.

The two samples were not combined for analysis due to differences in methodology of the two earlier studies. In the field test, the scale was completed after the clinical appraisal and actual placement decision were completed, while for the revalidation, the scale was completed before the actual placement decision was reached. Since it is unknown what impact these different methods may have had, the two samples were analysed separately.


Data used in this examination were originally gathered as part of the SRSW’s field test (Blanchette & Taylor, 2005) and revalidation (Gobeil, in press). Data were obtained from the Service’s automated offender information tools. In each case, the security classification recommendations produced by the SRSW, as well as the actual security placement (not always consistent with the SRSW recommendation), were recorded. In order to be consistent with how the scale is actually used in practice, the discretionary range was applied to all analyses.1 Data on the offender’s involvement in institutional misconducts after each review were also recorded. Misconducts were identified as being minor (e.g., disturbances, possession of unauthorized items), major (e.g., assaults, hostage-taking, possession of weapons or drugs), or any (either minor or major misconducts), and were coded as having occurred or not having occurred. The file verification corresponded to the time period for which each woman could be considered to be ‘at risk’ for institutional misbehaviour - from the date of review until the subsequent security review, release, or the end of data collection, whichever came first.


First, the demographic characteristics of each sample were examined. Second, rates of involvement in incidents were described in terms of both SRSW classification and actual placement. Next, chi-square tests of independence were used to assess differences in involvement in institutional misconduct for offenders matched on SRSW classification but differing in actual placements. Specifically, for those recommended to a medium classification by SRSW, comparisons were between those receiving actual placements of minimum, medium, or maximum security. The opposite procedure was also completed - for women actually placed in medium security, between-group differences in institutional behaviour were examined for offenders with SRSW recommendations of minimum, medium, or maximum security. Analyses focused on placements and classifications only to medium security, rather than across security levels, because the majority of cases fell into these categories. Analyses of other security levels had lower sample sizes, in many cases so much so that analyses became impossible (especially for the second sample, which included fewer cases).2

Since time at risk for institutional misbehaviour varied across security reviews, two types of analyses were considered: (1) analyses involving all misconducts occurring during the risk period, and (2) analyses limited to misconducts occurring in a fixed three-month follow-up. For the first option, results could potentially be biased if the ‘at risk’ period was significantly different according to pre-review security placement. For the second option, analyses would not include any such bias as all cases would be equated on time at risk. However, because some women in the original studies were reclassified or released before the end of their three-month follow-up periods, post-review data were not available for a portion of cases (nSAMPLE 1= 115; nSAMPLE 2= 31),3 thereby rendering some analyses impossible due to sample sizes.

Ultimately, pair-wise comparisons of the length of time at risk for institutional misconduct revealed no statistically significant differences according to pre-review security placement for either the field test or revalidation samples. As such, data on all misconducts occurring during the risk period were used in the analyses for the present report.4


Sample Characteristics

Table 1 presents the demographic characteristics of the women (not cases). Participants from each sample were similar in age, sentence length, proportion of women with life sentences, and marital status. A larger proportion of women in the second sample, however, were of Aboriginal ethnicity.5

Table 1. Demographic Characteristics

  Sample 1 Sample 2
Variable M (SD) % (n / 323) M (SD) % (n / 268)
Age at review 32.6 (8.3)   33.5 (9.3)  
Total Sentence Length (years) 3.7 (2.7)   4.0 (2.7)  
Life Sentenced   15 (49)   16 (43)
Caucasian   56 (182)   43 (114)
Aboriginal   35 (114)   49 (132)
Black   5 (16)   3 (9)
Other / Unknown   3 (11)   5 (13)
Marital Status        
Single / Widowed / Divorced   68 (220)   69 (186)
Married / Common-law   29 (94)   29 (79)
Unknown   3 (9)   1 (3)

Note. Proportions may not sum to 100 due to rounding.

Involvement in Institutional Misconducts

Overall, 184 cases from the first sample (31.7%) and 122 cases from the second sample (31.8%) were associated with subsequent involvement in institutional misconducts during the full at risk period. On average, this period was of 182.1 days in the first sample (range = 1 to 681 days) and 152.0 days in the second sample (range = 6 to 550 days). Figure 1 presents the distribution of involvement in institutional misconduct for each sample by type of misconduct and by SRSW recommendation. In both samples, a clear linear trend is apparent, with the proportion of women involved in each type of misconduct increasing together with SRSW security classifications.

Figure 1. Proportion of Cases Involved in Institutional Misconducts, by SRSW Security Classification

Note. All cases from each sample were included in calculating proportions.N SAMPLE 1 = 580; N SAMPLE 2 = 384.

distribution of involvement in institutional misconduct for each sample by type of misconduct and by SRSW recommendation

Figure 2 presents the proportion of cases involved in misconducts, broken down by type of misconduct and by actual security placement. Again, a linear trend is clearly present, with the proportion of women involved in each type of misconduct becoming larger as security placement increases. It is notable, however, that the distribution of proportions is less extreme when considering actual placement than when considering SRSW classifications. Specifically, the proportions of women at minimum security involved in misconducts (either minor, major, or any) are somewhat higher when considering actual placement. Similarly, the proportions of women at maximum security who were involved in institutional misconducts are slightly lower when considering actual placements. Put another way, SRSW classifications appear to be slightly more consistent with subsequent involvement in misconducts than actual placements.

Figure 2. Proportion of Cases Involved in Institutional Misconducts, by Actual Security Placement

Note. All cases from each sample were included in calculating proportions. N SAMPLE 1 = 580; N SAMPLE 2 = 384.

proportion of cases involved in misconducts, broken down by type of misconduct and by actual security placement

Actual security placements and SRSW classifications are generally quite consistent (72% and 66% concordant in the field test and revalidation respectively; Blanchette & Taylor, 2005; Gobeil, in press), but do not always coincide. It is those cases where scale recommendations and actual placements do not match that allow for examination of their relative associations with institutional behaviour.

Rates of Involvement, Controlling for SRSW Recommendation

First, the cases from each sample for which the SRSW recommended classification to medium security were examined. For each sample, the majority of women classified to medium security by the SRSW were also placed at medium security (69.4% and 68.6% for the field test and revalidation samples respectively), but the remainder were placed at an alternate level of security. Comparing the rates of involvement in misconducts for each level allows for the examination of the effect of actual placement, while methodologically holding constant the effect of security classification.

For these analyses, all women classified to medium security by the SRSW, regardless of actual placement, were included. As can be seen in Table 2, no statistical differences emerged for either sample. These findings demonstrate that the proportions of women with medium SRSW classifications engaging in institutional misconduct were similar for those actually placed at minimum, medium, and maximum security. Put another way, no association was found between involvement in institutional misconduct and actual security placement after controlling for SRSW classification.

Table 2. Involvement in Institutional Misconduct for Women with SRSW Classifications of Medium Security, by Actual Placement

  Actual Placement    
Misconduct Min. (%) Med. (%) Max. (%) Total (%) χ2
Sample 1          
Major % 4.2 9.0 19.0 13.4 3.34
(n / N ) (1 / 24) (24 / 186) (11 / 58) (36 / 268)  
Minor % 20.8 22.6 19.0 21.6 0.35
(n / N ) (5 / 24) (42 / 186) (11 / 58) (58 / 268)  
Any % 20.8 30.1 31.0 29.5 0.97
(n / N ) (5 / 24) (56 / 186) (18 / 58) (79 / 268)  
Sample 2          
Major % 0.0 24.6 23.3 22.7 0.18
(n / N ) (0 / 11) (29 / 118) (10 / 43) (39 / 172)  
Minor % 18.2 23.7 27.9 24.4 0.76
(n / N ) (2 / 11) (28 / 118) (12 / 43) (42 / 172)  
Any % 18.2 42.4 39.5 40.1 0.61
(n / N ) (2 / 11) (50 / 118) (17 / 43) (69 / 172)  

All results non-significant.

Rates of Involvement, Controlling for Actual Placement

A second series of analyses focused on the cases corresponding to actual placement at medium security, regardless of SRSW classification. Again, in most cases (69.9% and 60.5% for the first and second samples respectively), actual placement and SRSW classification were the same. Table 3 reports the proportions of women placed at medium security who were involved in each type of misconduct, according to SRSW classification. Significant differences are apparent in both samples. In the first sample, the proportion of women involved in incidents increased together with SRSW classification, with those classified to minimum security having the lowest rates of involvement, those classified to medium security having higher rates of involvement, and those classified to maximum security having the highest rates. This pattern, however, is less clear in the second sample.

Table 3. Involvement in Institutional Misconduct for Women Placed at Medium Security, by SRSW Security Classification

  SRSW Classification    
Misconduct Min. Med. Max. Total χ2
Sample 1          
Major  % 3.4 12.9 23.8 11.7 7.21*
(n / N (2 / 59) (24 / 186) (5 / 21) (31 / 266)  
Minor %  10.2 22.6 47.6 21.8 12.96**
(n / N (6 / 59) (42 / 186) (10 / 21) (58 / 266)  
Any % 13.6 30.1 52.4 28.2 12.65**
(n / N (8 / 59) (56 / 186) (11 / 21) (75 / 266)  
Sample 2          
Major % 3.1 24.6 15.4 16.9 13.60**
(n / N (2 / 64) (29 / 118) (2 / 13) (33 / 195)  
Minor %  4.7 23.7 15.4 16.7 10.72**
(n / N (3 / 64) (28 / 118) (2 / 13) (33 / 195)  
Any % 7.8 42.4 30.8 30.3 23.49****
(n / N (5 / 64) (50 / 118) (4 / 13) (59 / 195)  

* p  < .05.  ** p  < .01.  **** p  < .0001.

For the second sample, the proportion of women involved in misconducts was highest among those classified to medium security, followed by those classified to maximum security, and finally by those classified to minimum security. This pattern, however, may be attributable to the very small number of women (n = 13) classified to maximum security by the SRSW but actually placed at medium security. Indeed, follow-up analyses (using adjusted significance levels) supported this explanation. In comparing only women classified to minimum and medium security, the difference in the proportions of women involved in major, minor, and any incidents remained significant. Considering only those classified to medium and maximum security, however, no differences emerged. These analyses demonstrate that overall, for women placed at medium security, involvement in institutional misconducts was more common among those classified by the SRSW to higher security levels than among those classified to lower security levels.


The present investigation used Security Reclassification Scale for Women (SRSW) security level classifications and actual security placements to examine the impact of correctional environment on the institutional behaviour of federally sentenced women offenders. Across two samples, analyses consistently demonstrated that actual security placement was not associated with institutional conduct after accounting for SRSW security classification. Conversely, SRSW security classification continued to be associated with institutional behaviour when controlling methodologically for actual placement. Together, these results demonstrate that it is individual-level variables which influence institutional behaviour, and notcorrectional environment. Notably, these results are consistent with those found in samples composed primarily or solely of men (Baird, 1993; Camp & Gaes, 2005; Hanson et al., 1983; Luciani et al., 1996).

When considering only those cases where assessed risk was ‘medium’ (according to the SRSW), results of the present study suggested that institutional misconduct was independent of actual placement (minimum, medium, or maximum). While this may suggest that the correctional environment has no impact on behaviour, there is an alternative explanation. One American study revealed that being housed at higher levels of security had a suppressive effect on involvement in institutional misconduct,possibly due to greater supervision or higher levels of static security (Berecochea & Gibbs, 1991).

In the current context, it could be argued that the more secure environment at maximum security served to reduce potential misconducts in otherwise disruptive inmates.  The second series of analyses conducted in the present study, however, did not support this possibility. In these analyses, security level placement (the environment) was held constant, therefore any suppressive or amplifying effects associated with institutional security level would be consistent across women. Involvement in institutional misconduct according to SRSW security classifications continued to be apparent, however, demonstrating that a possible suppressive effect of incarceration at a higher level of security cannot adequately explain the present results.

Limitations and Future Directions

It is important to consider that not all incidents of institutional misbehaviour are likely to have been recorded in the data considered for these analyses. Indeed, empirical studies have determined that minor rule violations are inconsistently reported (e.g., Freeman, 2003; Hewitt, Poole, & Regoli, 1984). While the results of one study involving six male institutions in the U.S. demonstrated that rule enforcement differed by facility (Jenne & Kersting, 1998), these differences were not associated with institutional security level (D. Jenne, personal communication, August 24, 2007). As such, it is most likely that any differencesn reporting were consistent across actual placements, therefore not impacting the present findings.

It is also possible that correctional environment may have an impact on behaviour not assessed here - that is, that environment may contribute to post-release behaviour. It has been contended that prison can serve as a ‘training school’ for inexperienced offenders, in that these offenders may have the opportunity to learn criminal behaviours from their peers while incarcerated. Indeed, one group of researchers has suggested that ‘prisons breed crime’ (Petersilia, Turner, & Petersen, 1986, p. 37). It may be of interest to examine whether placement at different levels of security could moderate this effect. Investigation of this issue was impossible here, however, because Canadian legislation requires the periodic review of offenders’ security classifications (CCRA,1992), and most offenders are therefore housed at different security levels during their periods of incarceration. It would have been infeasible in the present study to methodologically disentangle the effects of different security placements throughout an offender’s sentence on her subsequent post-release behaviour.

Finally, this study differs from similar studies conducted with male inmates in that federal women offenders in Canada are housed in multi-level facilities, with those placed at minimum and medium security sharing programs, recreational activities, and social time, as well as being supervised by the same staff and being housed in very similar accommodations. As such, the correctional environment is likely not very different for women at these security levels, though their environment is quite different from that of women housed at maximum security. It is possible that in jurisdictions where there are greater differences in environment based on security level, results may have been different. Caution must therefore be used in extrapolating these findings tother correctional organizations.


The present study demonstrates that for women offenders, the association between security level and involvement in institutional misconduct is explained by the individual-level factors influencing security classification, and not by the correctional environment at each level of security. These findings should serve to alleviate any concerns that the environments at higher levels of security are criminogenic in nature. This is particularly encouraging given that security classification in Canada incorporates assessment of three areas of risk: institutional adjustment, risk of escape, and risk to the public in the event of an escape (CCRA, 1992). These results clearly indicate that when placement at a higher security level is necessary due to risk in one of the latter two areas, there is likely to be no associated negative effect on institutional adjustment. These findings, therefore, are supportive of Correctional Service Canada’s current classification and placement practices.


Baird, C. (1993). Objective classification in Tennessee: Management, effectiveness, and planning issues. In American Correctional Association (Ed.), Classification: A tool for managing today’s offenders. Arlington, VA: American Correctional Association.

Berecochea, J.E. & Gibbs, J.B. (1991). Inmate classification: A correctional program that works? Evaluation Review, 15, 333-363.

Berk, R.A., Ladd, H., Graziano H., & Baek, J.-H. (2003). Randomized experiment testing inmate classification systems. Criminology and Public Policy, 2, 215-242.

Blanchette, K. & Taylor, K. (2005). Development and field test of a gender-informed security reclassification scale for women offenders. Report R-167. Ottawa, ON: Correctional Service Canada.

Blanchette, K., Verbrugge, P., & Wichmann, C.G. (2002). The Custody Rating Scale, initial security level placement, and women offenders. Report R-127. Ottawa, ON: Correctional Service Canada.

Camp, S.D., & Gaes, G.G. (2005). Criminogenic effects of the prison environment on inmate behaviour: Some experimental evidence. Crime & Delinquency, 51, 425-442.

Canadian Association of Elizabeth Fry Societies (CAEFS; 1998). Position of the Canadian Association of ElizabethFry Societies (CAEFS) regarding the classification and carceral placement of women classified as maximum security prisoners. Retrieved August 29, 2007 from

Collie, R.M. & Polaschek, D.L.L. (2003). Sorting women’s risk: New Zealand women prisoner’s misconducts and internal security risk. New Zealand Journal of Psychology, 32, 101-109.

Correctional Service Canada (2005). CSC action plan in response to the report of the Canadian Human Rights Commission. Ottawa, ON: Author.

Corrections and Conditional Release Act. (1992).

Freeman, R.M. (2003). Social distance and discretionary rule enforcement in a women’s prison. The Prison Journal, 83, 191-205.

Gobeil, R. (in press). Revalidation of the Security Reclassification Scale for Women (SRSW). Ottawa, ON: Correctional Service Canada.

Hanson, R.W., Moss, C.S., Hosford, R.E., & Johnson, M.E. (1983). Predicting inmate penitentiary adjustment: An assessment of four classificatory methods. Criminal Justice and Behavior, 10, 293-309.

Harer, M.D., & Langan, N.P. (2001). Gender differences in predictors of prison violence: Assessing the predictive validity of a risk classification system. Crime & Delinquency, 47, 513-536.

Hewitt, J.D., Poole, E.D., & Regoli, R.M. (1984). Self-reported and observed rule breaking in prison: A look at disciplinary response. Justice Quarterly, 3, 437-448.

Jenne, D.L. & Kersting, R.C. (1998). Gender, power, and reciprocity in the correctional setting. The Prison Journal, 78, 166-185.

Luciani, F., Motiuk, L., & Nafekh, M. (1996). An operational review of the Custody Rating Scale: Validity, reliability, and practical utility. Report R-47. Ottawa, ON: Correctional Service of Canada.

Petersilia, J., Turner, S. & Petersen, J. (1986). Prison versus probation in California: Implications for crime and offender recidivism. Santa Monica, CA: RAND.

Warren, J., Hurt, S., Loper, A.B., & Chauhan, P. (2004). Exploring prison adjustment among female inmates: Issues of measurement and prediction. Criminal Justice and Behavior, 31, 624-645.

1 For example, if an offender’s score on the SRSW resulted in a recommended classification to medium security but fell within the discretionary range for maximum security, and the actual security placement was maximum security, the SRSW recommendation was recorded as maximum.

2 Chi-square analyses require minimum expected frequencies in each cell of the underlying contingency tables; these frequencies were not consistently achieved in analyses focused on cases placed or classified to levels of security other than medium.

3 These numbers refer only to those cases either placed at and / or classified to medium security, as those were the only ones included in the chi-square analyses. Considering only these cases, the sample sizes for the full at-risk period were of 348 and 249 cases respectively; for the fixed three-month follow-up, the sample sizes were of 233 and 218 cases respectively.

4 As an exercise, the analyses for which the three-month follow-up data included sufficient cases were also completed. Results of the two series of analyses were very similar.

5 Readers are referred to research reports summarizing the results of the SRSW field test (Blanchette & Taylor, 2005) and revalidation (Gobeil, in press) for more detailed reviews of the samples’ demographic characteristics.