«From Inception to Implementation: How SACPA has affected the Case Processing and Sentencing of Drug Offenders in One California County DISSERTATION ...»
This is not ideal, however, I have at least three other post-implementation years of data to work with and STATA time series software can navigate the gap without problems. CJSC data, besides being under-reported state-wide, is relatively good and free from serious issues for the offenses and time period I am looking at for Orange County. Furthermore, this under-reporting is consistent between years, so year-to-year comparisons can be made with confidence. In addition to these data from the California Criminal Justice Statistics Center (CJSC), data from the Orange District Attorney’s Office are also used to assess the impact of Proposition 36 on court processing and sentencing trends.
the Orange County District Attorney’s Office. This dataset includes information on all adults charged with a SACPA-eligible drug offense between July 1, 1998 and June 30,
2004. For each felony case, I have the offender’s charged offense/s, offender’s race, gender, and age at time of arrest and disposition. For each misdemeanor case, I have all of the above except final disposition. The District Attorney’s Office does not keep records of an individual’s arrest offense so the information is on charging offense only. These data are used to discern case processing and sentencing trends, such as the number of drug cases filed by the District Attorney, and to illuminate changes in characteristics of offenders prior to and after the law.
Corrections Data The Orange County Probation Department (OCPD) was the primary data source at the corrections stage. OCPD provided three datasets for research purposes28.
As previously stated, I was unable to secure prison admission data for drug offenders
the number of new admissions of drug offenders placed on probation for SACPAeligible offenses on a monthly basis from July 1, 1995 – May 30, 2006. This indicates whether the probation department is supervising a larger number of drug offenders after SACPA went into effect than before. I use this data in conjunction with CJSC I also requested a 4th dataset, the number of drug possession offenders who received treatment as a condition of probation prior to SACPA. Unfortunately this information was not easily available.
disposition data to get an overall picture of how offenders in Orange County were sentenced before and after SACPA went into effect.
contains a count of new SACPA-referred probationers each month for the period, July 1, 2001 to May 30, 2006. It does not include offense or offender information. It indicates how Orange County Probation Department caseloads changed as a result of SACPA. It does not necessarily tell me how many more (or fewer) probationers are being supervised as a result of SACPA because we do not know how many would have been sentenced to probation before SACPA. However, this information does tell me how many drug offending probationers are on SACPA.
and needs information for the population of offenders on probation for SACPAeligible offenses between July 1, 1995 and May 30, 2006. One of the most important questions to answer in this dissertation is whether observed changes are due to SACPA, offender characteristics, or other factors. This dataset contains more than 25 variables, including: offender characteristics, initial risk score, the number of probation violations, criminal history, financial issues, family issues, and other information pertinent to an offender’s risk profile for all probationers for the past 15 years. However, the scores after June 2002 are not reliable due to different assessment strategies by probation officers (Shirley Hunt, personal communication, October 22, 2005). Therefore the study period for this question is limited to July 1, 1998 to June 30, 2002.
I utilize this information to answer the question whether any changes in the number of offenders sentenced to prison, jail or probation are due to SACPA, differences in offender characteristics, or other factors. For instance, if SACPA is actually diverting offenders from prison, I expect to find higher risk scores for offenders sentenced to probation after SACPA implementation and as a result of SACPA than before. The dataset illustrates how the drug offender sentenced to probation has changed over the years. This could lend credibility to the assertion that the drug offenders sentenced under SACPA are indeed the offenders with higher risks
The current research project uses interrupted time series analysis, analysis of variance and chi-squares to interpret the quantitative data. Interrupted time-series analysis was used to ascertain whether any observed sentencing changes could be attributed to Proposition 36 implementation. ANOVAs were used to evaluate the observed differences in sentencing trends before and after Proposition 36. Chi-squares for cross-tabulation tables were used to determine whether observed offender characteristics were different after Proposition 36 than before Proposition 36 for data from the district attorney’s office and probation department.29 Chi-squares are suitable to answer the questions of interest in the current project. However, it is possible that more-advanced statistical techniques will be performed in the future to exploit the information about individuals contained in these datasets.
Interrupted Time Series Analysis Time-series analysis involves collecting many time points of aggregate level data before and after an intervention (in my case, SACPA implementation on 7/1/01) to determine if that intervention (SACPA) had any effect on the issue being studied (case processing and sentencing trends). Time series analysis requires an analysis of each trend prior to an intervention (SACPA implementation) to create a projection of what that trend would have looked like after the intervention (implementation date), had the intervention (SACPA) not occurred. It then compares this predicted trend to the actual trend after the intervention (SACPA implementation) took place to determine if they are statistically different from one another.
Time-series analysis is particularly useful for evaluating the effects of fullcoverage programs (interventions which apply uniformly to an entire population, such as policy interventions) like SACPA (Rossi, Freeman, and Lipsey, 1999). Time series designs “are the strongest way of examining full-coverage programs” (Rossi, Freeman, and Lipsey,1999: 268). A minimum of 30 time points are recommended before an intervention in order to obtain a correct projection of the trend line. I have a minimum of three years of data before and after SACPA implementation for this reason, as monthly observations for three years will yield 36 time points before intervention and 36 time points after intervention. This should be enough to ensure proper modeling and fit of each time-series. In most cases, I have many more time points30.
To maximize internal validity I collect data from January 1995 to December 2006 (providing 144 time points of data), when possible.
Because SACPA was passed at a discrete point in time, interrupted time-series analysis is a powerful statistical procedure which identifies and controls for both the seasonal variation and the long-term trending of the data (Shadish et al., 2002). This makes it a particularly strong and useful research strategy that is considered an “alternative to randomized designs when [randomized designs] are not feasible and a time-series can be found” (Shadish et al., 2002: p174)31.
ARIMA (Auto Regressive Integrated Moving Averages) models were built
McDowall, McCleary, Meidinger, and Hay (1980) and McCleary and Hay (Chapters 2-5, 1980) which provide more accessible developments of ARIMA model-building were also used. Data series were transformed into their natural logarithm when it was necessary (see McGarrell et al., 2006). STATA 10 was used to build ARIMA models.
It was evident, however that the series were extremely complex and SCA 8 software was used at the end to identify the best fit ARIMA models for some of the series.
Analysis of Variance Analysis of variance (ANOVA) tests whether the independent variable has an impact on the dependent variable by comparing the pre and post experiment levels of the dependent variable. A dummy variable in used to differentiate the pre-intervention period from the post-intervention period. ANOVA than compares the mean of the pre-intervention period to the mean of the post-intervention period to determine The main threat to validity is history. The best way to control for this threat is to collect information on similar “no treatment” jurisdictions – jurisdictions which are similar to Orange County, California but which did not experience any change in policy affecting drug offenders. In this study, history is not controlled for directly.
conjunction with time series analysis to assess the impact of Prop36 on sentencing variables.
Chi-Squares Chi-square (Χ2) is a statistic used to test whether there is any association between two (or more) variables using observed and expected values. It is particularly useful for investigating whether there are any differences in drug offenders before and after the law. Chi-square is based on the null hypothesis, which is the assumption that there is no relationship between the two variables of interest (such as offender gender and Proposition 36) and computes expected values based on the observed values. The value of chi-square tells us the likelihood of the observed value and the expected value being different by chance. Data from the probation department and district attorney’s office contain nominal and ordinal level individual data that are easily interpreted using chi-square analyses.
data to determine if there was any impact of Proposition 36 on the type of crime (felony or misdemeanor) offenders were charged with, the offense charged with, or the disposition. In addition DA data were analyzed to determine if there were any legislative impacts on the characteristics (race, gender, age) of offenders charged with SACPA-eligible drug crimes.
the risk/needs data from the probation department to identify changes in average drug The limitations of this test for time series data are acknowledged.
probationer risk score over time. If SACPA is diverting offenders from prison, I expect to find offenders sentenced to probation as a result of SACPA have higher risk scores than offenders sentenced to probation prior to the law. Such a discovery would indicate that drug offenders sentenced under SACPA are indeed more serious offenders than past offenders sentenced to probation and may signify that SACPA offenders are precisely those who would have been sentenced to prison prior to 2001.
In order to establish comparable pre- and post- intervention time periods, analyses were limited to one year before the law and one year after the law took effect.
The pre-intervention period was January 1, 2000 to December 31, 2000 and the postintervention time period was July 1, 2001 to June 30, 2002. January 1, 2001- June 30, 2001 was not used because Orange County conducted a pilot study in early 2001 prior to the start of Proposition 36 and it was possible that the pilot study population could confound the results of the analyses in unknown ways.
Chapter 3 Proposition 36 and the Orange County Experience Proposition 36 completely changed how California deals with minor drug offenders – from a crime control model to an addiction treatment model. It was written by drug reformers and opposed by many in the criminal justice system. The California District Attorney’s Office, the California Association of Drug Court Professionals, the California Peace Officer’s Association, judges, Attorney General Bill Lockyer, and U.S. Drug Czar Barry McCaffrey all came out against the law when it was on the ballot (Booth and Sanchez, 2000; Sauer, 2000; Wallace, 2000). These same groups that opposed the law were then required to implement it – not just adjust to it – but proactively create the infrastructure and shape the philosophy that would guide and govern how “Proposition 36” worked in Orange County. This was a monumental task, not only because of the new procedures that had to be put into place, but also because the scope was so large (36,000 expected offenders state-wide). So just how does a county go about implementing a new protocol for thousands of offenders each year? This chapter describes Orange County’s experience implementing Proposition 36.
The law was passed on November 7, 2000 with a mandatory implementation date of July 1, 2001. State, County and local agencies had slightly less than eight months to react to, plan and prepare for the law change. The statute mandated sentencing changes, required probation departments to work with treatment providers, and prescribed how probation violations (both drug and non-drug related) would be handled. However, it did not dictate how counties had to organize the various pieces of the process. Structural issues were left almost completely up to individual counties.
Thus, counties varied widely in their implementation strategies.
passed. An Orange County judge, who incidentally gave anti-Proposition 36 speeches up to Election Day, drafted Orange County Superior Court’s position statement within weeks of the election (Confidential Informant ECV, personal communication). It is clear from interviews with multiple criminal justice practitioners involved in the implementation of Proposition 36 that Judge Day33 took P the lead and spearheaded the effort to organize Orange County’s implementation effort. Judge Day organized a meeting of drug court managers in December 2000 to discuss the law and strategize how Orange County should proceed.
postpone their hearings until Proposition 36 became effective on July 1, 2001. If many offenders postponed their hearings, it would put pressure on both the jail and the court because many of these defendants would be in jail awaiting their postponed hearings and court staff would be unable to support the expected bulge of offenders.
He felt that if they “got moving quickly,” Orange County could avoid this anticipated bottleneck.