Evaluating the Effectiveness of Electronic Monitoring with Probationers

ABSTRACT

Despite growth in the use of electronic monitoring (EM), and Global Positioning Satellite (GPS) tracking in particular over the last 20 years, limited research has examined the effectiveness of these approaches on community supervision outcomes (e.g. likelihood of failure on community supervision). This study makes use of data from over 59,000 probation cases collected over a 40-year time span in a midwestern state. Using Propensity Score Matching (PSM), the study compares the rates of unsuccessful sentence completion among those who were supervised with and without any EM (and also GPS specifically). Results reveal that after matching cases on a variety of relevant controls (e.g. demographics, risk level, current offense type), those receiving EM while on probation were significantly more likely to fail to complete their term of supervision, with similar results for those who received GPS, specifically. Limitations of the study and implications for the use of EM/GPS with probationers as a potential alternative to prison are discussed.

KEYWORDS:

• Electronic monitoring
• probation
• alternative sanctions

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. Other more recently developed forms of EM include Secure Continuous Remote Alcohol Monitoring (SCRAM) first implemented in 2003. The use of SCRAM technology typically requires the individual to wear a monitoring device attached to their ankle which can periodically analyzes their perspiration to assess the person’s use of alcohol. These monitors may also include GPS tracking capabilities in addition to assessing alcohol consumption (SCRAM Systems, 2024). We do not examine the impact of SCRAM separately in this study because there were relatively few such individuals available in our data.

2. We examined whether the outliers in time of probation supervision influenced the findings presented in Tables 3 and 4. We “top coded” time on probation to the 95^th percentile and re-estimated the propensity scores. The findings are substantively identical when the effect of the outliers is constrained.

3. We performed several checks for the robustness of the findings. These findings are presented in the Appendix. We first examined whether the large sample size influenced the matching of individuals in the treatment group to the comparison group. One concern is that a large sample size may bias estimates of a treatment effect because the likelihood of finding a nearest-neighbor match based upon the propensity score increases as the sample size increases (see King & Nielsen, 2019 for a review of criticisms of PSM). We used a set seed in Stata 14.2 and randomly selected a sample size that is approximately 3 times as large as the treatment group to examine whether the findings are influenced by sample size. Appendix A and Appendix B reveal that the findings are substantively identical as those presented in Tables 3 and 4. Next, we examined whether receiving EM influences the odds of being unsuccessful by method of a multivariable logistic regression. Appendix C and Appendix D reveal that individuals who received EM have significantly higher odds of being unsuccessful on community supervision. These findings align with those of the PSM analyses performed in Tables 3 and 4. Finally, as an added check for the robustness of the findings we examined inverse probability of treatment weights (IPTW). Unlike traditional logistic regression, IPTW adjusts the analysis by using a weight of the treatment effect (e.g., receiving EM). Appendix E and F reveal that individuals who received EM have significantly higher odds of being unsuccessful on community supervision. Across each of our checks for the robustness of the findings presented in Tables 3 and 4, the takeaway is that placement on EM increases the odds of being unsuccessful on community supervision, which supports our final conclusion of the study.