The following pages are intended as a rough draft and discussion piece as to how decision support software might appear. Because of the various perspectives of potential users, the default "home" for a judge or attorney preparing for or participating in a sentencing hearing might be a different "home" screen than for a pretrial release or post conviction supervision officer, but the package ought to be flexible enough to allow easy navigation among the various screens.

Central to my focus for increasing the accuracy with which we aim sanctions at offenders in a way most likely to reduce their criminal behavior is my "home" screen - the screen which first presents itself when I access the software. I would first see "Display Sanctions by Performance" with the available sanctions for a given sentencing decision arrayed in terms of rate of recidivism for the default category of offender most appropriate for the offender before me. Details for the case and the offender would come from those screens, and the decision reached with the assistance of the "Sanctions by Performance" screen would be passed on to the "Disposition Finalization and Recording Screen" to document and record the result.

Starting with the Display Sanctions by Performance Screen, the offender category would be gleaned by the software from what we know about the offender and the case; likewise the available sanctions. In other words, in an ideal world, the software would "know" which sanctions to present based on what it "knows" about the nature of the offense, the eligibility criteria of the sanctions, and the offender's prior history and performance. The recidivism (outcome) measure would by default the one I have ended up using most for this type of case - ideally, the software would learn my approach by category of offense and offender.

Even if the software is not smart enough to construct these parts for me, and in any event after presenting an initial array, I would be able to refine any of the three variables: the offender category, the sanctions displayed, and the outcome measure. Each screen would present buttons for regularly used defaults, and each would allow user-defined categories by reference to the appropriate supporting screen or subscreens.

I'm not exactly sure of the relationship between the "offender profile selector" and the "risk assessment instrument screens," but my notion is that the available tools (the jail's matrix, the corrections department's risk assessment tool, and the Restitution Center's screening tool) will help us construct useful offender profiles. I expect that our experience with this system will greatly improve our ability to formulate offender profiles. It may be that all we need is a numerical rating (as is the output for these tools, as I understand them), but I expect that we will want more. We will probably at least start with categories of offenders based on level of prior involvement (criminal history), current charge, and substance abuse involvement, with basic distinctions among such factors as generalized violence, situational violence, property crimes only, substance abuse, substance abuse crimes only. A link to available literature or synopses of literature would ideally enable users to research the nuance of categorization in question.

The Sanction Screen is probably the same thing as the contemplated "Sanction Tracking Project." It would allow the user to view information about all available sanctions, including such information as their target population, their eligibility criteria, cost, waiting list, and counter-indications (e.g., "no sex offenders need apply.") Flexibility and power are key - I should be able to click on a category such as "anger management" and drill down to details on each available program inn and out of custody. Regularly useful categories should be available at a mouse-click, such as inpatient vs. outpatient substance abuse, SB 710 vs. non "listed" alcohol treatment, and the like. Once I selected which programs or groups of programs to consider, they would be among those displayed in the "Display Sanctions by Performance" screen. Links to literature might assist the user in appreciating the implications for different treatment approaches, for example.

The "Outcome Measure Selector" would enable me to choose from among different flavors of recidivism by which to compare outcomes. Defaults would probably range from "any misbehavior" through "new felony conviction," and would probably include mild violence, major violence, minor property, major property, and substance abuse crimes only. We might well find, for example, that as between programs or dispositions x, y, and z, the only significant difference in outcomes for offenders profiled like the one before us appears when "serious violence" is the only measure of recidivism. Please remember that outcomes to be compared often include some term of incarceration, local or state, and that we have the figures for these as well in our system. We should also be able to craft specific measures of "recidivism" for individual cases. Links to literature might help us refine or select outcome measures.

It is essential, however, that outcome measures focus on some measure of apparent impact on future criminal conduct. One choice should be the period after graduation from the sanction which is used to assess outcome. One, three, and five years are reasonable default choices. We must not let this piece degenerate to "success" during treatment or counseling (except, perhaps, in the rare situations of treatment which is expected to continue indefinitely into the future, such as some categories of mental health and sex offender treatment). On the other hand, we need to be able to see what percentage of offenders sent to a sanction graduate from it (as opposed to fail to show up or to complete the sanction), as a program whose graduates commit no crimes is qualified in its usefulness if almost none of its clients graduate.

Case Screens and  Offender Screens	Display Sanctions by Performance   Array available dispositions or choices so as to display relative success at producing reduced criminal behavior (and, in case of pretrial release decisions, reduced FTAs) for a given offender profile and a given outcome measure.  Probably a bar chart, each bar designated by a sanction title, with flags to indicate chosen offender profile and outcome measure, and links to the screens below to modify each or both.  Use or revise dispositions to be compared from the Sanction Screen, Offender Profiles from the Profile Selector Screen, and Outcome Measures from the Outcome Measure Selector Screen.  Click to see Multnomah County's screens	Disposition Finalization and Recording Screen
Offender Profile Selector   Select from among default categories of offenders, with the suggested category, drawn from this offender's data, highlighted or flagged, or construct new or modify default categories, with access to the various risk assessment tools.	Sanction Screen  Displays available dispositions (custodial and otherwise), ideally based upon legal and factual criteria to include the theoretically available choices.  Choose from this screen to determine which sanctions will be compared in the "Sanctions by Performance" screen; refer to detail screens (below) to determine whether to add or remove possible dispositions from the comparison.	Outcome Measure Selector   Select from among default categories of outcome measures, ranging from negative police contacts through arrests to convictions, with the capacity to screen any or all of these measures by property vs violent crime, with capacity to create user-defined outcome measures.
To Risk Assessment Instrument Screens and to Literature	To Sanction Detail Screens and to Literature	To Literature