In Spring 1999, the Anchorage Police Department (APD) and the Alaska Division of Juvenile Justice (DJJ) established a partnership to enhance the supervision and services provided to juvenile probationers in Anchorage. Modeled after a successful program in San Diego County, California, the Anchorage Coordinated Agency Network (CANS) program extended the supervisory arm of the youth probation office by having Anchorage police officers make random visits to juvenile probationers. This article discusses an evaluation conducted by the Justice Center of the pilot phase of the CANS program.
The Anchorage program targets juvenile probationers in the Anchorage Police Department service area with the hope of fulfilling two objectives. First, the program is designed to enhance overall levels of probationer supervision through intensive and systematic supervision. The police officers involved in the program are volunteers. Each participating officer is assigned at least one or, in some cases, two youth probationers. The police officer is expected to make at least two unannounced contacts with the assigned probationer each month to ensure that the juvenile is in compliance with the terms of probation. After each visit, the officer writes a brief description of the nature of the contact and forwards the information to the CANS program coordinator within the Anchorage Police Department.
The one-to-one police officer/probationer contact augments the supervision already provided by juvenile probation officers, each of whom is currently handling a caseload of 40 to 50 probationers. In fact, a juvenile in the CANS program can receive up to three times as many contacts as under regular probation. If a juvenile is not in compliance with the terms of his/her probation, the police department will notify the assigned probation officer, who is then able act upon any violations noted by the police officer.
The second objective of the CANS program is to provide positive role models for youth. Through their contact with juveniles, officers can engender mutual respect and foster positive interactions between youth and the police.
Program officials anticipate that, if the objectives described above are met, juvenile probationers participating in the CANS program will have lower levels of recidivism than those juveniles who do not participate. The pilot phase provided an opportunity to evaluate the outcomes for a select group of juvenile probationers before the program was implemented on a larger scale.
The Justice Center conducted a two-part examination of the CANS project. The first part of the study explored whether program participants had more favorable outcomes than a non-program control group. Second, an analysis was conducted to determine the relevant predictors of successful program outcomes.
Random assignment to the program was made prior to the design of the evaluation. The Division of Juvenile Justice Youth Probation Department produced a list of active youth probationers who were not institutionalized at the time. From this list, 95 juveniles were randomly assigned to be participants in the CANS pilot phase and 95 were assigned to a control group. The purpose of the control group was to provide a comparison. Unlike the CANS participants, juveniles in this group were supervised by DJJ youth probation officers only, not receiving additional supervision.
The design of the evaluation assumed that participants in the program were randomly assigned to the control or treatment groups. The Anchorage Police Department and DJJ provided data on each of the 190 juveniles participating in the study. APD produced data on the officer/probationer pairings and frequency of contacts. DJJ provided information on each probationer, including demographic characteristics such as age, sex, race, and education, as well as data on prior experiences, such as previous record and history of child abuse. In addition, DJJ supplied data regarding participants' probation violations and new offenses during the CANS pilot phase.
In order to be eligible for inclusion in the study, both control and experimental group members must have been on probation and not detained or institutionalized at the beginning of the CANS program, June 1, 1999. Thirty-five juveniles in the original group were removed from the analysis due to their ineligibility. The majority of ineligible cases were excluded because the juvenile was either off probation or institutionalized before the pilot period began. Three additional juveniles were excluded for being AWOL at the time of the program start, and another juvenile was excluded due to his move to Washington. The majority of ineligible juveniles were drawn from the control (non-CANS) group (88.6%); being off probation was the primary reason for ineligibility (51%).
The final sample used in the analysis consisted of 155 juveniles (91 experimental, 64 control). Table 1 presents descriptive statistics for juveniles in both groups. A chi-square test revealed that the disproportionate attrition of juveniles from the control group did not create differences between the two groups on any of the variables examined.
As noted above, CANS officers participated in the program voluntarily, with each participating police officer assigned one or two juveniles. Each officer was supposed to visit the juvenile two times per month.
CANS officers made a total of 186 contacts with the CANS participants between June 1, 1999 and December 31, 1999. Figure 1 shows that the majority of visits occurred in August and September. It is not surprising that the pattern of contacts takes on a bell-shaped curve. In June, the pilot phase was just beginning and officers were becoming acquainted with their role. As they learned about the expectations of the program, they gradually increased the number of contacts they were making. At the end of the pilot phase, the number of visits tapered off. This reduction might be attributed to several factors. Some juveniles were charged with new offenses or violated conditions of their probation and were institutionalized. Other juveniles simply completed their probation and were no longer a part of the program. In either case, there were fewer juveniles to supervise and contact at the latter stages of the pilot phase than at the beginning.
The number of contacts received by each CANS program juvenile varied. The majority of juveniles received between one and three visits during the pilot phase (56.1%). Twenty juveniles, or 22 per cent of CANS participants, did not receive any visits during the period. Most juveniles received one visit (30.8%) between June and December, 11 per cent received two visits, and 14.3 per cent received three visits. Twenty-two per cent of juveniles were contacted by their CANS officer four times or more. The mean number of contacts was 2.01 contacts over the seven-month period.
The results shown in Table 2 indicate that a slightly smaller proportion of CANS juveniles (19.8%) committed new offenses during the pilot phase when compared with control group participants (23.4%). The difference, however, was small and did not achieve statistical significance. The two groups did differ to a greater extent on new probation violations. While 17.2 per cent of control group members committed new technical violations during the period, 29.7 per cent of CANS juveniles committed new technical violations. (Note that the two groups may not necessarily differ in the actual number of new offenses or new probation violations; these are reported differences. Differences may be a product of the increased supervision and the increased likelihood of being caught.) Again, the differences were not significant at the commonly used .05 p value but the findings, consistent with findings reported in the literature, do lend some support to the idea that increased supervision leads to higher numbers of recorded probation violations.
The data also reveal that the 155 juveniles involved in the evaluation were involved in 43 incidents that resulted in new charges being filed. Of these incidents, 26 (60.4%) were committed by individuals in the CANS program while only 17 (39.5%) were committed by juveniles in the control group. The 26 experimental group incidents that resulted in new charges were committed by 18 different CANS program juveniles. Fifteen control group juveniles were responsible for 17 incidents that resulted in new charges. Table 3 presents the most serious charge for each of the total 43 events. It is worth noting that the differences between the control and the CANS group in the number of incidents is small when the size of each group is taken into account. That is, there were .286 incidents for each CANS program juvenile while there were .266 incidents for each control group member. This finding of very small differences in the rate of new incidents mirrors the findings above, suggesting no differences between the two groups in the likelihood of facing new charges.
A more sophisticated analysis was needed to determine possible predictors of program success. This analysis, which considered the influence of CANS participation while holding all other variables equal, was a stronger test of the relationship between CANS and new charges and new probation violations.
The results indicated that two variables might be significant predictors of new probation violations—participation in the CANS program and three or more changes in the juvenile's living situation. In addition, four other variables approached significance: work time, school time, prior history of abuse/neglect, and one or two changes in living situation.
To further explain this finding, the odds ratios need to be addressed. The results indicated that the odds of a juvenile in the CANS program having new probation violations, all else being equal, were 3.2 times greater than the odds of a juvenile not in the CANS program having new probation violations. In other words, CANS program juveniles were 220 per cent more likely to have new probation violations than control group participants. The findings here also suggested that juveniles with three or more changes in their living situation, regardless of whether they were CANS participants or not, were over 10 times more likely to have new probation violations than juveniles with no changes.
The variable of one or two changes in living situation—which approaches statistical significance—may play some role in predicting new probation violations, not surprising given the finding that three or more changes also predicted probation outcomes. Two other variables, work time and school time, also approached statistical significance. Therefore, there is some evidence to suggest that juveniles who were not working or not working regularly and juveniles not attending school or not attending regularly were more likely to have new probation violations. Finally, a child with a prior history of abuse was somewhat more likely to have new probation violations, although, the results were not statistically significant. (It is worth restating that the higher significance levels for each of these four variables mean that the differences in outcomes may be due to chance alone rather than any true differences in living situation, work and school situation, and child abuse history.)
When the impact of CANS was examined using new charges as the dependent variable, the only significant relationship found was between age and three or more changes in living situation and new charges. The relationship is negative, indicating that older juveniles were less likely to have new charges filed against them than younger juveniles. In other words, for each unit increase in age (e.g., one-year increase in age), the odds of new charges being filed were reduced by 37 per cent. In addition, having three or more changes in the living situation increased the likelihood of new charges being filed by over 195 per cent, compared with a juvenile with no changes. The other variables provided no significant contribution to predicting outcomes and were excluded from the statistical equation.
One final analysis produced interesting results. In this analysis, only participants in the CANS program were included. A model was constructed that included only those variables that are relevant in predicting the outcome: educational level, work time, three or more changes in living situation, and total number of visits. Of these, only total visits and three or more changes in living situation were statistically significant. As in the other analyses, the living situation variable increased the odds of new offenses. Among CANS participants only, the increase in odds is over 400 per cent. That is, a CANS juvenile with three or more changes in living situation was more than five times as likely to have new charges during the pilot phase than a juvenile with no changes.
More important, perhaps, for the CANS program was the impact of visitation. The results indicated that, among CANS participants, for every one visit increase, the odds of new charges being filed decreased by 37 per cent. In other words, an increase in the number of contacts reduced the likelihood of a CANS juvenile having new charges filed. If one considers the earlier finding that 63.8 per cent of juveniles received fewer than three contacts during the study period, it is not surprising that CANS participation exerted no significant impact on new violations in the earlier analysis. However, if the results of the analysis of CANS individuals only are any indication, the impact of the CANS program may be more pronounced if the number of contacts is increased. Table 4 further illustrates this point. Thirty-five per cent of CANS juveniles who received zero contacts had new charges filed against them. In contrast, only five per cent of juveniles who were contacted four or more times had new charges filed against them. In sum, there is some evidence to suggest that increasing the number of visits may have a preventive impact on new offenses. Frequency of contacts did not, however, significantly predict the likelihood of new probation violations.
This evaluation of CANS involved directly comparing a group of 91 CANS participants with 64 juvenile probationers not in the CANS program. The first outcome examined was new probation violations. The data reveal that almost 30 per cent of CANS juveniles had new technical violations compared to slightly more than 17 per cent of control group members. Furthermore, a more sophisticated logistic regression analysis identified CANS participation as a significant predictor of new probation violations. That is, CANS participants were over three times as likely to have new technical violations when compared to the control group. This fact does not imply that CANS or ISP participants commit more technical violations but, rather, there are increased opportunities of being caught due to enhanced surveillance.
Further analysis revealed no significant differences between control and CANS group members in the likelihood of committing offenses that resulted in new charges being filed. At first glance, this finding suggests the program does not meet its objectives. However, this finding should not be interpreted alone; instead, it is also necessary to consider the intensity of contacts received by each juvenile. Only 22 per cent of juveniles in the CANS program were contacted more than three times during the study period. Of the remaining juveniles, more than 28 per cent did not receive a single visit. It is true that each CANS juvenile received a letter indicating that an assigned police officer would visit. However, the letter was the only real distinction between the 22 per cent of CANS juveniles who received no visits and the control group. With this in mind, an analysis was undertaken to examine the impact of contact intensity on CANS participants only. The findings were revealing. With all other variables controlled, each contact reduced the odds of new charges being filed against a juvenile by 37 per cent. It is entirely possible that the large number of CANS juveniles with few contacts masked any differences that did exist between CANS juveniles and the control group. If this possibility is correct, one could expect the preventive impact of CANS to be greater as the number of contacts increases. However, an alternative explanation for this finding reverses the causal relationship. Instead of the number of visits predicting probation success, it is possible that probation success determined the number of visits. For example, a juvenile who commited a new offense and was institutionalized was, for obvious reasons, unable to receive visits from CANS officers. Therefore, while a relationship between the number of visits and probation success exists, the issue of causality is an empirical question which cannot be answered here.
Two additional findings are worth noting. Instability in a juvenile's living situation was consistently found to be related to the likelihood of having new probation violations or new charges. More specifically, three or more changes increased the likelihood of having new probation violations ten times more than for a juvenile with no changes. Similarly, juveniles with three or more changes were nearly four times as likely to have new charges filed against them. The influence of residential instability was independent of CANS participation. In other words, all else being equal, a juvenile with three or more changes in living situation was more likely to have probation violations or new charges. A second interesting finding is that increasing age reduced the probability of new charges. Again, the impact of age existed regardless of CANS participation.
It is important to note that other factors may play a role in predicting outcomes. The results above consider only those variables that were included in the analysis. Additional factors related to officer contacts with juveniles may be pertinent. For example, the variable "number of contacts" included in the above analyses assumes that all visits are equal. The variable is not able to discern what the officer does on each visit, who the contact is (e.g., juvenile, guardian), how long the contact lasts, or where the contact was made. It is possible that certain types of visits are more beneficial and have a greater effect on program outcomes than other types.
Despite the caveat noted above, evidence does exist that the objectives of the CANS program are being achieved. Although juveniles in CANS are more likely to have technical violations, the increased supervision also enhances the level of accountability for a juvenile's actions. In addition, there is some evidence suggesting that increasing the number of contacts may reduce the levels of new charges.
Matthew Giblin is a researcher with the Justice Center.